| @@ -1,7510 +1,7510 | |||
|
|
1 | 1 | # v3.0-devel |
|
|
2 | 2 | import numpy |
|
|
3 | 3 | import math |
|
|
4 | 4 | from scipy import optimize, interpolate, signal, stats, ndimage |
|
|
5 | 5 | from scipy.fftpack import fft |
|
|
6 | 6 | import scipy |
|
|
7 | 7 | from scipy.optimize import least_squares |
|
|
8 | 8 | import re |
|
|
9 | 9 | import datetime |
|
|
10 | 10 | import copy |
|
|
11 | 11 | import sys |
|
|
12 | 12 | import importlib |
|
|
13 | 13 | import itertools |
|
|
14 | 14 | from multiprocessing import Pool, TimeoutError |
|
|
15 | 15 | from multiprocessing.pool import ThreadPool |
|
|
16 | 16 | import time |
|
|
17 | 17 | |
|
|
18 | 18 | import matplotlib.pyplot as plt |
|
|
19 | 19 | from scipy.optimize import fmin_l_bfgs_b #optimize with bounds on state papameters |
|
|
20 | 20 | from .jroproc_base import ProcessingUnit, Operation, MPDecorator |
|
|
21 | 21 | from schainpy.model.data.jrodata import Parameters, hildebrand_sekhon |
|
|
22 | 22 | from scipy import asarray as ar,exp |
|
|
23 | 23 | from scipy.optimize import fmin, curve_fit |
|
|
24 | 24 | from schainpy.utils import log |
|
|
25 | 25 | import warnings |
|
|
26 | 26 | from numpy import NaN |
|
|
27 | 27 | from scipy.optimize.optimize import OptimizeWarning |
|
|
28 | 28 | warnings.filterwarnings('ignore') |
|
|
29 | 29 | |
|
|
30 | 30 | |
|
|
31 | 31 | SPEED_OF_LIGHT = 299792458 |
|
|
32 | 32 | |
|
|
33 | 33 | '''solving pickling issue''' |
|
|
34 | 34 | |
|
|
35 | 35 | def _pickle_method(method): |
|
|
36 | 36 | func_name = method.__func__.__name__ |
|
|
37 | 37 | obj = method.__self__ |
|
|
38 | 38 | cls = method.__self__.__class__ |
|
|
39 | 39 | return _unpickle_method, (func_name, obj, cls) |
|
|
40 | 40 | |
|
|
41 | 41 | def _unpickle_method(func_name, obj, cls): |
|
|
42 | 42 | for cls in cls.mro(): |
|
|
43 | 43 | try: |
|
|
44 | 44 | func = cls.__dict__[func_name] |
|
|
45 | 45 | except KeyError: |
|
|
46 | 46 | pass |
|
|
47 | 47 | else: |
|
|
48 | 48 | break |
|
|
49 | 49 | return func.__get__(obj, cls) |
|
|
50 | 50 | |
|
|
51 | 51 | # @MPDecorator |
|
|
52 | 52 | class ParametersProc(ProcessingUnit): |
|
|
53 | 53 | |
|
|
54 | 54 | METHODS = {} |
|
|
55 | 55 | nSeconds = None |
|
|
56 | 56 | |
|
|
57 | 57 | def __init__(self): |
|
|
58 | 58 | ProcessingUnit.__init__(self) |
|
|
59 | 59 | |
|
|
60 | 60 | self.buffer = None |
|
|
61 | 61 | self.firstdatatime = None |
|
|
62 | 62 | self.profIndex = 0 |
|
|
63 | 63 | self.dataOut = Parameters() |
|
|
64 | 64 | self.setupReq = False #Agregar a todas las unidades de proc |
|
|
65 | 65 | |
|
|
66 | 66 | def __updateObjFromInput(self): |
|
|
67 | 67 | |
|
|
68 | 68 | self.dataOut.inputUnit = self.dataIn.type |
|
|
69 | 69 | |
|
|
70 | 70 | self.dataOut.timeZone = self.dataIn.timeZone |
|
|
71 | 71 | self.dataOut.dstFlag = self.dataIn.dstFlag |
|
|
72 | 72 | self.dataOut.errorCount = self.dataIn.errorCount |
|
|
73 | 73 | self.dataOut.useLocalTime = self.dataIn.useLocalTime |
|
|
74 | 74 | |
|
|
75 | 75 | self.dataOut.radarControllerHeaderObj = self.dataIn.radarControllerHeaderObj.copy() |
|
|
76 | 76 | self.dataOut.systemHeaderObj = self.dataIn.systemHeaderObj.copy() |
|
|
77 | 77 | self.dataOut.channelList = self.dataIn.channelList |
|
|
78 | 78 | self.dataOut.heightList = self.dataIn.heightList |
|
|
79 | 79 | self.dataOut.dtype = numpy.dtype([('real','<f4'),('imag','<f4')]) |
|
|
80 | 80 | # self.dataOut.nHeights = self.dataIn.nHeights |
|
|
81 | 81 | # self.dataOut.nChannels = self.dataIn.nChannels |
|
|
82 | 82 | # self.dataOut.nBaud = self.dataIn.nBaud |
|
|
83 | 83 | # self.dataOut.nCode = self.dataIn.nCode |
|
|
84 | 84 | # self.dataOut.code = self.dataIn.code |
|
|
85 | 85 | # self.dataOut.nProfiles = self.dataOut.nFFTPoints |
|
|
86 | 86 | self.dataOut.flagDiscontinuousBlock = self.dataIn.flagDiscontinuousBlock |
|
|
87 | 87 | # self.dataOut.utctime = self.firstdatatime |
|
|
88 | 88 | self.dataOut.utctime = self.dataIn.utctime |
|
|
89 | 89 | self.dataOut.flagDecodeData = self.dataIn.flagDecodeData #asumo q la data esta decodificada |
|
|
90 | 90 | self.dataOut.flagDeflipData = self.dataIn.flagDeflipData #asumo q la data esta sin flip |
|
|
91 | 91 | self.dataOut.nCohInt = self.dataIn.nCohInt |
|
|
92 | 92 | # self.dataOut.nIncohInt = 1 |
|
|
93 | 93 | # self.dataOut.ippSeconds = self.dataIn.ippSeconds |
|
|
94 | 94 | # self.dataOut.windowOfFilter = self.dataIn.windowOfFilter |
|
|
95 | 95 | self.dataOut.timeInterval1 = self.dataIn.timeInterval |
|
|
96 | 96 | self.dataOut.heightList = self.dataIn.heightList |
|
|
97 | 97 | self.dataOut.frequency = self.dataIn.frequency |
|
|
98 | 98 | #self.dataOut.noise = self.dataIn.noise |
|
|
99 | 99 | |
|
|
100 | 100 | def run(self): |
|
|
101 | 101 | |
|
|
102 | 102 | #---------------------- Voltage Data --------------------------- |
|
|
103 | 103 | |
|
|
104 | 104 | if self.dataIn.type == "Voltage": |
|
|
105 | 105 | |
|
|
106 | 106 | self.__updateObjFromInput() |
|
|
107 | 107 | self.dataOut.data_pre = self.dataIn.data.copy() |
|
|
108 | 108 | self.dataOut.flagNoData = False |
|
|
109 | 109 | self.dataOut.utctimeInit = self.dataIn.utctime |
|
|
110 | 110 | self.dataOut.paramInterval = self.dataIn.nProfiles*self.dataIn.nCohInt*self.dataIn.ippSeconds |
|
|
111 | 111 | if hasattr(self.dataIn, 'dataPP_POW'): |
|
|
112 | 112 | self.dataOut.dataPP_POW = self.dataIn.dataPP_POW |
|
|
113 | 113 | |
|
|
114 | 114 | if hasattr(self.dataIn, 'dataPP_POWER'): |
|
|
115 | 115 | self.dataOut.dataPP_POWER = self.dataIn.dataPP_POWER |
|
|
116 | 116 | |
|
|
117 | 117 | if hasattr(self.dataIn, 'dataPP_DOP'): |
|
|
118 | 118 | self.dataOut.dataPP_DOP = self.dataIn.dataPP_DOP |
|
|
119 | 119 | |
|
|
120 | 120 | if hasattr(self.dataIn, 'dataPP_SNR'): |
|
|
121 | 121 | self.dataOut.dataPP_SNR = self.dataIn.dataPP_SNR |
|
|
122 | 122 | |
|
|
123 | 123 | if hasattr(self.dataIn, 'dataPP_WIDTH'): |
|
|
124 | 124 | self.dataOut.dataPP_WIDTH = self.dataIn.dataPP_WIDTH |
|
|
125 | 125 | return |
|
|
126 | 126 | |
|
|
127 | 127 | #---------------------- Spectra Data --------------------------- |
|
|
128 | 128 | |
|
|
129 | 129 | if self.dataIn.type == "Spectra": |
|
|
130 | 130 | |
|
|
131 | 131 | self.dataOut.data_pre = [self.dataIn.data_spc, self.dataIn.data_cspc] |
|
|
132 | 132 | self.dataOut.data_spc = self.dataIn.data_spc |
|
|
133 | 133 | self.dataOut.data_cspc = self.dataIn.data_cspc |
|
|
134 | 134 | # for JULIA processing |
|
|
135 | 135 | self.dataOut.data_diffcspc = self.dataIn.data_diffcspc |
|
|
136 | 136 | self.dataOut.nDiffIncohInt = self.dataIn.nDiffIncohInt |
|
|
137 | 137 | # for JULIA processing |
|
|
138 | 138 | self.dataOut.nProfiles = self.dataIn.nProfiles |
|
|
139 | 139 | self.dataOut.nIncohInt = self.dataIn.nIncohInt |
|
|
140 | 140 | self.dataOut.nFFTPoints = self.dataIn.nFFTPoints |
|
|
141 | 141 | self.dataOut.ippFactor = self.dataIn.ippFactor |
|
|
142 | 142 | self.dataOut.abscissaList = self.dataIn.getVelRange(1) |
|
|
143 | 143 | self.dataOut.spc_noise = self.dataIn.getNoise() |
|
|
144 | 144 | self.dataOut.spc_range = (self.dataIn.getFreqRange(1) , self.dataIn.getAcfRange(1) , self.dataIn.getVelRange(1)) |
|
|
145 | 145 | # self.dataOut.normFactor = self.dataIn.normFactor |
|
|
146 | 146 | self.dataOut.pairsList = self.dataIn.pairsList |
|
|
147 | 147 | self.dataOut.groupList = self.dataIn.pairsList |
|
|
148 | 148 | self.dataOut.flagNoData = False |
|
|
149 | 149 | |
|
|
150 | 150 | if hasattr(self.dataIn, 'ChanDist'): #Distances of receiver channels |
|
|
151 | 151 | self.dataOut.ChanDist = self.dataIn.ChanDist |
|
|
152 | 152 | else: self.dataOut.ChanDist = None |
|
|
153 | 153 | |
|
|
154 | 154 | #if hasattr(self.dataIn, 'VelRange'): #Velocities range |
|
|
155 | 155 | # self.dataOut.VelRange = self.dataIn.VelRange |
|
|
156 | 156 | #else: self.dataOut.VelRange = None |
|
|
157 | 157 | |
|
|
158 | 158 | if hasattr(self.dataIn, 'RadarConst'): #Radar Constant |
|
|
159 | 159 | self.dataOut.RadarConst = self.dataIn.RadarConst |
|
|
160 | 160 | |
|
|
161 | 161 | if hasattr(self.dataIn, 'NPW'): #NPW |
|
|
162 | 162 | self.dataOut.NPW = self.dataIn.NPW |
|
|
163 | 163 | |
|
|
164 | 164 | if hasattr(self.dataIn, 'COFA'): #COFA |
|
|
165 | 165 | self.dataOut.COFA = self.dataIn.COFA |
|
|
166 | 166 | |
|
|
167 | 167 | |
|
|
168 | 168 | |
|
|
169 | 169 | #---------------------- Correlation Data --------------------------- |
|
|
170 | 170 | |
|
|
171 | 171 | if self.dataIn.type == "Correlation": |
|
|
172 | 172 | acf_ind, ccf_ind, acf_pairs, ccf_pairs, data_acf, data_ccf = self.dataIn.splitFunctions() |
|
|
173 | 173 | |
|
|
174 | 174 | self.dataOut.data_pre = (self.dataIn.data_cf[acf_ind,:], self.dataIn.data_cf[ccf_ind,:,:]) |
|
|
175 | 175 | self.dataOut.normFactor = (self.dataIn.normFactor[acf_ind,:], self.dataIn.normFactor[ccf_ind,:]) |
|
|
176 | 176 | self.dataOut.groupList = (acf_pairs, ccf_pairs) |
|
|
177 | 177 | |
|
|
178 | 178 | self.dataOut.abscissaList = self.dataIn.lagRange |
|
|
179 | 179 | self.dataOut.noise = self.dataIn.noise |
|
|
180 | 180 | self.dataOut.data_snr = self.dataIn.SNR |
|
|
181 | 181 | self.dataOut.flagNoData = False |
|
|
182 | 182 | self.dataOut.nAvg = self.dataIn.nAvg |
|
|
183 | 183 | |
|
|
184 | 184 | #---------------------- Parameters Data --------------------------- |
|
|
185 | 185 | |
|
|
186 | 186 | if self.dataIn.type == "Parameters": |
|
|
187 | 187 | self.dataOut.copy(self.dataIn) |
|
|
188 | 188 | self.dataOut.flagNoData = False |
|
|
189 | 189 | |
|
|
190 | 190 | return True |
|
|
191 | 191 | |
|
|
192 | 192 | self.__updateObjFromInput() |
|
|
193 | 193 | self.dataOut.utctimeInit = self.dataIn.utctime |
|
|
194 | 194 | self.dataOut.paramInterval = self.dataIn.timeInterval |
|
|
195 | 195 | |
|
|
196 | 196 | return |
|
|
197 | 197 | |
|
|
198 | 198 | |
|
|
199 | 199 | def target(tups): |
|
|
200 | 200 | |
|
|
201 | 201 | obj, args = tups |
|
|
202 | 202 | |
|
|
203 | 203 | return obj.FitGau(args) |
|
|
204 | 204 | |
|
|
205 | 205 | class RemoveWideGC(Operation): |
|
|
206 | 206 | ''' This class remove the wide clutter and replace it with a simple interpolation points |
|
|
207 | 207 | This mainly applies to CLAIRE radar |
|
|
208 | 208 | |
|
|
209 | 209 | ClutterWidth : Width to look for the clutter peak |
|
|
210 | 210 | |
|
|
211 | 211 | Input: |
|
|
212 | 212 | |
|
|
213 | 213 | self.dataOut.data_pre : SPC and CSPC |
|
|
214 | 214 | self.dataOut.spc_range : To select wind and rainfall velocities |
|
|
215 | 215 | |
|
|
216 | 216 | Affected: |
|
|
217 | 217 | |
|
|
218 | 218 | self.dataOut.data_pre : It is used for the new SPC and CSPC ranges of wind |
|
|
219 | 219 | |
|
|
220 | 220 | Written by D. ScipiΓ³n 25.02.2021 |
|
|
221 | 221 | ''' |
|
|
222 | 222 | def __init__(self): |
|
|
223 | 223 | Operation.__init__(self) |
|
|
224 | 224 | self.i = 0 |
|
|
225 | 225 | self.ich = 0 |
|
|
226 | 226 | self.ir = 0 |
|
|
227 | 227 | |
|
|
228 | 228 | def run(self, dataOut, ClutterWidth=2.5): |
|
|
229 | 229 | |
|
|
230 | 230 | self.spc = dataOut.data_pre[0].copy() |
|
|
231 | 231 | self.spc_out = dataOut.data_pre[0].copy() |
|
|
232 | 232 | self.Num_Chn = self.spc.shape[0] |
|
|
233 | 233 | self.Num_Hei = self.spc.shape[2] |
|
|
234 | 234 | VelRange = dataOut.spc_range[2][:-1] |
|
|
235 | 235 | dv = VelRange[1]-VelRange[0] |
|
|
236 | 236 | |
|
|
237 | 237 | # Find the velocities that corresponds to zero |
|
|
238 | 238 | gc_values = numpy.squeeze(numpy.where(numpy.abs(VelRange) <= ClutterWidth)) |
|
|
239 | 239 | |
|
|
240 | 240 | # Removing novalid data from the spectra |
|
|
241 | 241 | for ich in range(self.Num_Chn) : |
|
|
242 | 242 | for ir in range(self.Num_Hei) : |
|
|
243 | 243 | # Estimate the noise at each range |
|
|
244 | 244 | HSn = hildebrand_sekhon(self.spc[ich,:,ir],dataOut.nIncohInt) |
|
|
245 | 245 | |
|
|
246 | 246 | # Removing the noise floor at each range |
|
|
247 | 247 | novalid = numpy.where(self.spc[ich,:,ir] < HSn) |
|
|
248 | 248 | self.spc[ich,novalid,ir] = HSn |
|
|
249 | 249 | |
|
|
250 | 250 | junk = numpy.append(numpy.insert(numpy.squeeze(self.spc[ich,gc_values,ir]),0,HSn),HSn) |
|
|
251 | 251 | j1index = numpy.squeeze(numpy.where(numpy.diff(junk)>0)) |
|
|
252 | 252 | j2index = numpy.squeeze(numpy.where(numpy.diff(junk)<0)) |
|
|
253 | 253 | if ((numpy.size(j1index)<=1) | (numpy.size(j2index)<=1)) : |
|
|
254 | 254 | continue |
|
|
255 | 255 | junk3 = numpy.squeeze(numpy.diff(j1index)) |
|
|
256 | 256 | junk4 = numpy.squeeze(numpy.diff(j2index)) |
|
|
257 | 257 | |
|
|
258 | 258 | valleyindex = j2index[numpy.where(junk4>1)] |
|
|
259 | 259 | peakindex = j1index[numpy.where(junk3>1)] |
|
|
260 | 260 | |
|
|
261 | 261 | isvalid = numpy.squeeze(numpy.where(numpy.abs(VelRange[gc_values[peakindex]]) <= 2.5*dv)) |
|
|
262 | 262 | if numpy.size(isvalid) == 0 : |
|
|
263 | 263 | continue |
|
|
264 | 264 | if numpy.size(isvalid) >1 : |
|
|
265 | 265 | vindex = numpy.argmax(self.spc[ich,gc_values[peakindex[isvalid]],ir]) |
|
|
266 | 266 | isvalid = isvalid[vindex] |
|
|
267 | 267 | |
|
|
268 | 268 | # clutter peak |
|
|
269 | 269 | gcpeak = peakindex[isvalid] |
|
|
270 | 270 | vl = numpy.where(valleyindex < gcpeak) |
|
|
271 | 271 | if numpy.size(vl) == 0: |
|
|
272 | 272 | continue |
|
|
273 | 273 | gcvl = valleyindex[vl[0][-1]] |
|
|
274 | 274 | vr = numpy.where(valleyindex > gcpeak) |
|
|
275 | 275 | if numpy.size(vr) == 0: |
|
|
276 | 276 | continue |
|
|
277 | 277 | gcvr = valleyindex[vr[0][0]] |
|
|
278 | 278 | |
|
|
279 | 279 | # Removing the clutter |
|
|
280 | 280 | interpindex = numpy.array([gc_values[gcvl], gc_values[gcvr]]) |
|
|
281 | 281 | gcindex = gc_values[gcvl+1:gcvr-1] |
|
|
282 | 282 | self.spc_out[ich,gcindex,ir] = numpy.interp(VelRange[gcindex],VelRange[interpindex],self.spc[ich,interpindex,ir]) |
|
|
283 | 283 | |
|
|
284 | 284 | dataOut.data_pre[0] = self.spc_out |
|
|
285 | 285 | |
|
|
286 | 286 | return dataOut |
|
|
287 | 287 | |
|
|
288 | 288 | class SpectralFilters(Operation): |
|
|
289 | 289 | ''' This class allows to replace the novalid values with noise for each channel |
|
|
290 | 290 | This applies to CLAIRE RADAR |
|
|
291 | 291 | |
|
|
292 | 292 | PositiveLimit : RightLimit of novalid data |
|
|
293 | 293 | NegativeLimit : LeftLimit of novalid data |
|
|
294 | 294 | |
|
|
295 | 295 | Input: |
|
|
296 | 296 | |
|
|
297 | 297 | self.dataOut.data_pre : SPC and CSPC |
|
|
298 | 298 | self.dataOut.spc_range : To select wind and rainfall velocities |
|
|
299 | 299 | |
|
|
300 | 300 | Affected: |
|
|
301 | 301 | |
|
|
302 | 302 | self.dataOut.data_pre : It is used for the new SPC and CSPC ranges of wind |
|
|
303 | 303 | |
|
|
304 | 304 | Written by D. ScipiΓ³n 29.01.2021 |
|
|
305 | 305 | ''' |
|
|
306 | 306 | def __init__(self): |
|
|
307 | 307 | Operation.__init__(self) |
|
|
308 | 308 | self.i = 0 |
|
|
309 | 309 | |
|
|
310 | 310 | def run(self, dataOut, ): |
|
|
311 | 311 | |
|
|
312 | 312 | self.spc = dataOut.data_pre[0].copy() |
|
|
313 | 313 | self.Num_Chn = self.spc.shape[0] |
|
|
314 | 314 | VelRange = dataOut.spc_range[2] |
|
|
315 | 315 | |
|
|
316 | 316 | # novalid corresponds to data within the Negative and PositiveLimit |
|
|
317 | 317 | |
|
|
318 | 318 | |
|
|
319 | 319 | # Removing novalid data from the spectra |
|
|
320 | 320 | for i in range(self.Num_Chn): |
|
|
321 | 321 | self.spc[i,novalid,:] = dataOut.noise[i] |
|
|
322 | 322 | dataOut.data_pre[0] = self.spc |
|
|
323 | 323 | return dataOut |
|
|
324 | 324 | |
|
|
325 | 325 | |
|
|
326 | 326 | |
|
|
327 | 327 | class GaussianFit(Operation): |
|
|
328 | 328 | |
|
|
329 | 329 | ''' |
|
|
330 | 330 | Function that fit of one and two generalized gaussians (gg) based |
|
|
331 | 331 | on the PSD shape across an "power band" identified from a cumsum of |
|
|
332 | 332 | the measured spectrum - noise. |
|
|
333 | 333 | |
|
|
334 | 334 | Input: |
|
|
335 | 335 | self.dataOut.data_pre : SelfSpectra |
|
|
336 | 336 | |
|
|
337 | 337 | Output: |
|
|
338 | 338 | self.dataOut.SPCparam : SPC_ch1, SPC_ch2 |
|
|
339 | 339 | |
|
|
340 | 340 | ''' |
|
|
341 | 341 | def __init__(self): |
|
|
342 | 342 | Operation.__init__(self) |
|
|
343 | 343 | self.i=0 |
|
|
344 | 344 | |
|
|
345 | 345 | |
|
|
346 | 346 | # def run(self, dataOut, num_intg=7, pnoise=1., SNRlimit=-9): #num_intg: Incoherent integrations, pnoise: Noise, vel_arr: range of velocities, similar to the ftt points |
|
|
347 | 347 | def run(self, dataOut, SNRdBlimit=-9, method='generalized'): |
|
|
348 | 348 | """This routine will find a couple of generalized Gaussians to a power spectrum |
|
|
349 | 349 | methods: generalized, squared |
|
|
350 | 350 | input: spc |
|
|
351 | 351 | output: |
|
|
352 | 352 | noise, amplitude0,shift0,width0,p0,Amplitude1,shift1,width1,p1 |
|
|
353 | 353 | """ |
|
|
354 | 354 | print ('Entering ',method,' double Gaussian fit') |
|
|
355 | 355 | self.spc = dataOut.data_pre[0].copy() |
|
|
356 | 356 | self.Num_Hei = self.spc.shape[2] |
|
|
357 | 357 | self.Num_Bin = self.spc.shape[1] |
|
|
358 | 358 | self.Num_Chn = self.spc.shape[0] |
|
|
359 | 359 | |
|
|
360 | 360 | start_time = time.time() |
|
|
361 | 361 | |
|
|
362 | 362 | pool = Pool(processes=self.Num_Chn) |
|
|
363 | 363 | args = [(dataOut.spc_range[2], ich, dataOut.spc_noise[ich], dataOut.nIncohInt, SNRdBlimit) for ich in range(self.Num_Chn)] |
|
|
364 | 364 | objs = [self for __ in range(self.Num_Chn)] |
|
|
365 | 365 | attrs = list(zip(objs, args)) |
|
|
366 | 366 | DGauFitParam = pool.map(target, attrs) |
|
|
367 | 367 | # Parameters: |
|
|
368 | 368 | # 0. Noise, 1. Amplitude, 2. Shift, 3. Width 4. Power |
|
|
369 | 369 | dataOut.DGauFitParams = numpy.asarray(DGauFitParam) |
|
|
370 | 370 | |
|
|
371 | 371 | # Double Gaussian Curves |
|
|
372 | 372 | gau0 = numpy.zeros([self.Num_Chn,self.Num_Bin,self.Num_Hei]) |
|
|
373 | 373 | gau0[:] = numpy.NaN |
|
|
374 | 374 | gau1 = numpy.zeros([self.Num_Chn,self.Num_Bin,self.Num_Hei]) |
|
|
375 | 375 | gau1[:] = numpy.NaN |
|
|
376 | 376 | x_mtr = numpy.transpose(numpy.tile(dataOut.getVelRange(1)[:-1], (self.Num_Hei,1))) |
|
|
377 | 377 | for iCh in range(self.Num_Chn): |
|
|
378 | 378 | N0 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][0,:,0]] * self.Num_Bin)) |
|
|
379 | 379 | N1 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][0,:,1]] * self.Num_Bin)) |
|
|
380 | 380 | A0 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][1,:,0]] * self.Num_Bin)) |
|
|
381 | 381 | A1 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][1,:,1]] * self.Num_Bin)) |
|
|
382 | 382 | v0 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][2,:,0]] * self.Num_Bin)) |
|
|
383 | 383 | v1 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][2,:,1]] * self.Num_Bin)) |
|
|
384 | 384 | s0 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][3,:,0]] * self.Num_Bin)) |
|
|
385 | 385 | s1 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][3,:,1]] * self.Num_Bin)) |
|
|
386 | 386 | if method == 'generalized': |
|
|
387 | 387 | p0 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][4,:,0]] * self.Num_Bin)) |
|
|
388 | 388 | p1 = numpy.transpose(numpy.transpose([dataOut.DGauFitParams[iCh][4,:,1]] * self.Num_Bin)) |
|
|
389 | 389 | elif method == 'squared': |
|
|
390 | 390 | p0 = 2. |
|
|
391 | 391 | p1 = 2. |
|
|
392 | 392 | gau0[iCh] = A0*numpy.exp(-0.5*numpy.abs((x_mtr-v0)/s0)**p0)+N0 |
|
|
393 | 393 | gau1[iCh] = A1*numpy.exp(-0.5*numpy.abs((x_mtr-v1)/s1)**p1)+N1 |
|
|
394 | 394 | dataOut.GaussFit0 = gau0 |
|
|
395 | 395 | dataOut.GaussFit1 = gau1 |
|
|
396 | 396 | |
|
|
397 | 397 | print('Leaving ',method ,' double Gaussian fit') |
|
|
398 | 398 | return dataOut |
|
|
399 | 399 | |
|
|
400 | 400 | def FitGau(self, X): |
|
|
401 | 401 | # print('Entering FitGau') |
|
|
402 | 402 | # Assigning the variables |
|
|
403 | 403 | Vrange, ch, wnoise, num_intg, SNRlimit = X |
|
|
404 | 404 | # Noise Limits |
|
|
405 | 405 | noisebl = wnoise * 0.9 |
|
|
406 | 406 | noisebh = wnoise * 1.1 |
|
|
407 | 407 | # Radar Velocity |
|
|
408 | 408 | Va = max(Vrange) |
|
|
409 | 409 | deltav = Vrange[1] - Vrange[0] |
|
|
410 | 410 | x = numpy.arange(self.Num_Bin) |
|
|
411 | 411 | |
|
|
412 | 412 | # print ('stop 0') |
|
|
413 | 413 | |
|
|
414 | 414 | # 5 parameters, 2 Gaussians |
|
|
415 | 415 | DGauFitParam = numpy.zeros([5, self.Num_Hei,2]) |
|
|
416 | 416 | DGauFitParam[:] = numpy.NaN |
|
|
417 | 417 | |
|
|
418 | 418 | # SPCparam = [] |
|
|
419 | 419 | # SPC_ch1 = numpy.zeros([self.Num_Bin,self.Num_Hei]) |
|
|
420 | 420 | # SPC_ch2 = numpy.zeros([self.Num_Bin,self.Num_Hei]) |
|
|
421 | 421 | # SPC_ch1[:] = 0 #numpy.NaN |
|
|
422 | 422 | # SPC_ch2[:] = 0 #numpy.NaN |
|
|
423 | 423 | # print ('stop 1') |
|
|
424 | 424 | for ht in range(self.Num_Hei): |
|
|
425 | 425 | # print (ht) |
|
|
426 | 426 | # print ('stop 2') |
|
|
427 | 427 | # Spectra at each range |
|
|
428 | 428 | spc = numpy.asarray(self.spc)[ch,:,ht] |
|
|
429 | 429 | snr = ( spc.mean() - wnoise ) / wnoise |
|
|
430 | 430 | snrdB = 10.*numpy.log10(snr) |
|
|
431 | 431 | |
|
|
432 | 432 | #print ('stop 3') |
|
|
433 | 433 | if snrdB < SNRlimit : |
|
|
434 | 434 | # snr = numpy.NaN |
|
|
435 | 435 | # SPC_ch1[:,ht] = 0#numpy.NaN |
|
|
436 | 436 | # SPC_ch1[:,ht] = 0#numpy.NaN |
|
|
437 | 437 | # SPCparam = (SPC_ch1,SPC_ch2) |
|
|
438 | 438 | # print ('SNR less than SNRth') |
|
|
439 | 439 | continue |
|
|
440 | 440 | # wnoise = hildebrand_sekhon(spc,num_intg) |
|
|
441 | 441 | # print ('stop 2.01') |
|
|
442 | 442 | ############################################# |
|
|
443 | 443 | # normalizing spc and noise |
|
|
444 | 444 | # This part differs from gg1 |
|
|
445 | 445 | # spc_norm_max = max(spc) #commented by D. ScipiΓ³n 19.03.2021 |
|
|
446 | 446 | #spc = spc / spc_norm_max |
|
|
447 | 447 | # pnoise = pnoise #/ spc_norm_max #commented by D. ScipiΓ³n 19.03.2021 |
|
|
448 | 448 | ############################################# |
|
|
449 | 449 | |
|
|
450 | 450 | # print ('stop 2.1') |
|
|
451 | 451 | fatspectra=1.0 |
|
|
452 | 452 | # noise per channel.... we might want to use the noise at each range |
|
|
453 | 453 | |
|
|
454 | 454 | # wnoise = noise_ #/ spc_norm_max #commented by D. ScipiΓ³n 19.03.2021 |
|
|
455 | 455 | #wnoise,stdv,i_max,index =enoise(spc,num_intg) #noise estimate using Hildebrand Sekhon, only wnoise is used |
|
|
456 | 456 | #if wnoise>1.1*pnoise: # to be tested later |
|
|
457 | 457 | # wnoise=pnoise |
|
|
458 | 458 | # noisebl = wnoise*0.9 |
|
|
459 | 459 | # noisebh = wnoise*1.1 |
|
|
460 | 460 | spc = spc - wnoise # signal |
|
|
461 | 461 | |
|
|
462 | 462 | # print ('stop 2.2') |
|
|
463 | 463 | minx = numpy.argmin(spc) |
|
|
464 | 464 | #spcs=spc.copy() |
|
|
465 | 465 | spcs = numpy.roll(spc,-minx) |
|
|
466 | 466 | cum = numpy.cumsum(spcs) |
|
|
467 | 467 | # tot_noise = wnoise * self.Num_Bin #64; |
|
|
468 | 468 | |
|
|
469 | 469 | # print ('stop 2.3') |
|
|
470 | 470 | # snr = sum(spcs) / tot_noise |
|
|
471 | 471 | # snrdB = 10.*numpy.log10(snr) |
|
|
472 | 472 | #print ('stop 3') |
|
|
473 | 473 | # if snrdB < SNRlimit : |
|
|
474 | 474 | # snr = numpy.NaN |
|
|
475 | 475 | # SPC_ch1[:,ht] = 0#numpy.NaN |
|
|
476 | 476 | # SPC_ch1[:,ht] = 0#numpy.NaN |
|
|
477 | 477 | # SPCparam = (SPC_ch1,SPC_ch2) |
|
|
478 | 478 | # print ('SNR less than SNRth') |
|
|
479 | 479 | # continue |
|
|
480 | 480 | |
|
|
481 | 481 | |
|
|
482 | 482 | #if snrdB<-18 or numpy.isnan(snrdB) or num_intg<4: |
|
|
483 | 483 | # return [None,]*4,[None,]*4,None,snrdB,None,None,[None,]*5,[None,]*9,None |
|
|
484 | 484 | # print ('stop 4') |
|
|
485 | 485 | cummax = max(cum) |
|
|
486 | 486 | epsi = 0.08 * fatspectra # cumsum to narrow down the energy region |
|
|
487 | 487 | cumlo = cummax * epsi |
|
|
488 | 488 | cumhi = cummax * (1-epsi) |
|
|
489 | 489 | powerindex = numpy.array(numpy.where(numpy.logical_and(cum>cumlo, cum<cumhi))[0]) |
|
|
490 | 490 | |
|
|
491 | 491 | # print ('stop 5') |
|
|
492 | 492 | if len(powerindex) < 1:# case for powerindex 0 |
|
|
493 | 493 | # print ('powerindex < 1') |
|
|
494 | 494 | continue |
|
|
495 | 495 | powerlo = powerindex[0] |
|
|
496 | 496 | powerhi = powerindex[-1] |
|
|
497 | 497 | powerwidth = powerhi-powerlo |
|
|
498 | 498 | if powerwidth <= 1: |
|
|
499 | 499 | # print('powerwidth <= 1') |
|
|
500 | 500 | continue |
|
|
501 | 501 | |
|
|
502 | 502 | # print ('stop 6') |
|
|
503 | 503 | firstpeak = powerlo + powerwidth/10.# first gaussian energy location |
|
|
504 | 504 | secondpeak = powerhi - powerwidth/10. #second gaussian energy location |
|
|
505 | 505 | midpeak = (firstpeak + secondpeak)/2. |
|
|
506 | 506 | firstamp = spcs[int(firstpeak)] |
|
|
507 | 507 | secondamp = spcs[int(secondpeak)] |
|
|
508 | 508 | midamp = spcs[int(midpeak)] |
|
|
509 | 509 | |
|
|
510 | 510 | y_data = spc + wnoise |
|
|
511 | 511 | |
|
|
512 | 512 | ''' single Gaussian ''' |
|
|
513 | 513 | shift0 = numpy.mod(midpeak+minx, self.Num_Bin ) |
|
|
514 | 514 | width0 = powerwidth/4.#Initialization entire power of spectrum divided by 4 |
|
|
515 | 515 | power0 = 2. |
|
|
516 | 516 | amplitude0 = midamp |
|
|
517 | 517 | state0 = [shift0,width0,amplitude0,power0,wnoise] |
|
|
518 | 518 | bnds = ((0,self.Num_Bin-1),(1,powerwidth),(0,None),(0.5,3.),(noisebl,noisebh)) |
|
|
519 | 519 | lsq1 = fmin_l_bfgs_b(self.misfit1, state0, args=(y_data,x,num_intg), bounds=bnds, approx_grad=True) |
|
|
520 | 520 | # print ('stop 7.1') |
|
|
521 | 521 | # print (bnds) |
|
|
522 | 522 | |
|
|
523 | 523 | chiSq1=lsq1[1] |
|
|
524 | 524 | |
|
|
525 | 525 | # print ('stop 8') |
|
|
526 | 526 | if fatspectra<1.0 and powerwidth<4: |
|
|
527 | 527 | choice=0 |
|
|
528 | 528 | Amplitude0=lsq1[0][2] |
|
|
529 | 529 | shift0=lsq1[0][0] |
|
|
530 | 530 | width0=lsq1[0][1] |
|
|
531 | 531 | p0=lsq1[0][3] |
|
|
532 | 532 | Amplitude1=0. |
|
|
533 | 533 | shift1=0. |
|
|
534 | 534 | width1=0. |
|
|
535 | 535 | p1=0. |
|
|
536 | 536 | noise=lsq1[0][4] |
|
|
537 | 537 | #return (numpy.array([shift0,width0,Amplitude0,p0]), |
|
|
538 | 538 | # numpy.array([shift1,width1,Amplitude1,p1]),noise,snrdB,chiSq1,6.,sigmas1,[None,]*9,choice) |
|
|
539 | 539 | # print ('stop 9') |
|
|
540 | 540 | ''' two Gaussians ''' |
|
|
541 | 541 | #shift0=numpy.mod(firstpeak+minx,64); shift1=numpy.mod(secondpeak+minx,64) |
|
|
542 | 542 | shift0 = numpy.mod(firstpeak+minx, self.Num_Bin ) |
|
|
543 | 543 | shift1 = numpy.mod(secondpeak+minx, self.Num_Bin ) |
|
|
544 | 544 | width0 = powerwidth/6. |
|
|
545 | 545 | width1 = width0 |
|
|
546 | 546 | power0 = 2. |
|
|
547 | 547 | power1 = power0 |
|
|
548 | 548 | amplitude0 = firstamp |
|
|
549 | 549 | amplitude1 = secondamp |
|
|
550 | 550 | state0 = [shift0,width0,amplitude0,power0,shift1,width1,amplitude1,power1,wnoise] |
|
|
551 | 551 | #bnds=((0,63),(1,powerwidth/2.),(0,None),(0.5,3.),(0,63),(1,powerwidth/2.),(0,None),(0.5,3.),(noisebl,noisebh)) |
|
|
552 | 552 | bnds=((0,self.Num_Bin-1),(1,powerwidth/2.),(0,None),(0.5,3.),(0,self.Num_Bin-1),(1,powerwidth/2.),(0,None),(0.5,3.),(noisebl,noisebh)) |
|
|
553 | 553 | #bnds=(( 0,(self.Num_Bin-1) ),(1,powerwidth/2.),(0,None),(0.5,3.),( 0,(self.Num_Bin-1)),(1,powerwidth/2.),(0,None),(0.5,3.),(0.1,0.5)) |
|
|
554 | 554 | |
|
|
555 | 555 | # print ('stop 10') |
|
|
556 | 556 | lsq2 = fmin_l_bfgs_b( self.misfit2 , state0 , args=(y_data,x,num_intg) , bounds=bnds , approx_grad=True ) |
|
|
557 | 557 | |
|
|
558 | 558 | # print ('stop 11') |
|
|
559 | 559 | chiSq2 = lsq2[1] |
|
|
560 | 560 | |
|
|
561 | 561 | # print ('stop 12') |
|
|
562 | 562 | |
|
|
563 | 563 | oneG = (chiSq1<5 and chiSq1/chiSq2<2.0) and (abs(lsq2[0][0]-lsq2[0][4])<(lsq2[0][1]+lsq2[0][5])/3. or abs(lsq2[0][0]-lsq2[0][4])<10) |
|
|
564 | 564 | |
|
|
565 | 565 | # print ('stop 13') |
|
|
566 | 566 | if snrdB>-12: # when SNR is strong pick the peak with least shift (LOS velocity) error |
|
|
567 | 567 | if oneG: |
|
|
568 | 568 | choice = 0 |
|
|
569 | 569 | else: |
|
|
570 | 570 | w1 = lsq2[0][1]; w2 = lsq2[0][5] |
|
|
571 | 571 | a1 = lsq2[0][2]; a2 = lsq2[0][6] |
|
|
572 | 572 | p1 = lsq2[0][3]; p2 = lsq2[0][7] |
|
|
573 | 573 | s1 = (2**(1+1./p1))*scipy.special.gamma(1./p1)/p1 |
|
|
574 | 574 | s2 = (2**(1+1./p2))*scipy.special.gamma(1./p2)/p2 |
|
|
575 | 575 | gp1 = a1*w1*s1; gp2 = a2*w2*s2 # power content of each ggaussian with proper p scaling |
|
|
576 | 576 | |
|
|
577 | 577 | if gp1>gp2: |
|
|
578 | 578 | if a1>0.7*a2: |
|
|
579 | 579 | choice = 1 |
|
|
580 | 580 | else: |
|
|
581 | 581 | choice = 2 |
|
|
582 | 582 | elif gp2>gp1: |
|
|
583 | 583 | if a2>0.7*a1: |
|
|
584 | 584 | choice = 2 |
|
|
585 | 585 | else: |
|
|
586 | 586 | choice = 1 |
|
|
587 | 587 | else: |
|
|
588 | 588 | choice = numpy.argmax([a1,a2])+1 |
|
|
589 | 589 | #else: |
|
|
590 | 590 | #choice=argmin([std2a,std2b])+1 |
|
|
591 | 591 | |
|
|
592 | 592 | else: # with low SNR go to the most energetic peak |
|
|
593 | 593 | choice = numpy.argmax([lsq1[0][2]*lsq1[0][1],lsq2[0][2]*lsq2[0][1],lsq2[0][6]*lsq2[0][5]]) |
|
|
594 | 594 | |
|
|
595 | 595 | # print ('stop 14') |
|
|
596 | 596 | shift0 = lsq2[0][0] |
|
|
597 | 597 | vel0 = Vrange[0] + shift0 * deltav |
|
|
598 | 598 | shift1 = lsq2[0][4] |
|
|
599 | 599 | # vel1=Vrange[0] + shift1 * deltav |
|
|
600 | 600 | |
|
|
601 | 601 | # max_vel = 1.0 |
|
|
602 | 602 | # Va = max(Vrange) |
|
|
603 | 603 | # deltav = Vrange[1]-Vrange[0] |
|
|
604 | 604 | # print ('stop 15') |
|
|
605 | 605 | #first peak will be 0, second peak will be 1 |
|
|
606 | 606 | # if vel0 > -1.0 and vel0 < max_vel : #first peak is in the correct range # Commented by D.ScipiΓ³n 19.03.2021 |
|
|
607 | 607 | if vel0 > -Va and vel0 < Va : #first peak is in the correct range |
|
|
608 | 608 | shift0 = lsq2[0][0] |
|
|
609 | 609 | width0 = lsq2[0][1] |
|
|
610 | 610 | Amplitude0 = lsq2[0][2] |
|
|
611 | 611 | p0 = lsq2[0][3] |
|
|
612 | 612 | |
|
|
613 | 613 | shift1 = lsq2[0][4] |
|
|
614 | 614 | width1 = lsq2[0][5] |
|
|
615 | 615 | Amplitude1 = lsq2[0][6] |
|
|
616 | 616 | p1 = lsq2[0][7] |
|
|
617 | 617 | noise = lsq2[0][8] |
|
|
618 | 618 | else: |
|
|
619 | 619 | shift1 = lsq2[0][0] |
|
|
620 | 620 | width1 = lsq2[0][1] |
|
|
621 | 621 | Amplitude1 = lsq2[0][2] |
|
|
622 | 622 | p1 = lsq2[0][3] |
|
|
623 | 623 | |
|
|
624 | 624 | shift0 = lsq2[0][4] |
|
|
625 | 625 | width0 = lsq2[0][5] |
|
|
626 | 626 | Amplitude0 = lsq2[0][6] |
|
|
627 | 627 | p0 = lsq2[0][7] |
|
|
628 | 628 | noise = lsq2[0][8] |
|
|
629 | 629 | |
|
|
630 | 630 | if Amplitude0<0.05: # in case the peak is noise |
|
|
631 | 631 | shift0,width0,Amplitude0,p0 = 4*[numpy.NaN] |
|
|
632 | 632 | if Amplitude1<0.05: |
|
|
633 | 633 | shift1,width1,Amplitude1,p1 = 4*[numpy.NaN] |
|
|
634 | 634 | |
|
|
635 | 635 | # print ('stop 16 ') |
|
|
636 | 636 | # SPC_ch1[:,ht] = noise + Amplitude0*numpy.exp(-0.5*(abs(x-shift0)/width0)**p0) |
|
|
637 | 637 | # SPC_ch2[:,ht] = noise + Amplitude1*numpy.exp(-0.5*(abs(x-shift1)/width1)**p1) |
|
|
638 | 638 | # SPCparam = (SPC_ch1,SPC_ch2) |
|
|
639 | 639 | |
|
|
640 | 640 | DGauFitParam[0,ht,0] = noise |
|
|
641 | 641 | DGauFitParam[0,ht,1] = noise |
|
|
642 | 642 | DGauFitParam[1,ht,0] = Amplitude0 |
|
|
643 | 643 | DGauFitParam[1,ht,1] = Amplitude1 |
|
|
644 | 644 | DGauFitParam[2,ht,0] = Vrange[0] + shift0 * deltav |
|
|
645 | 645 | DGauFitParam[2,ht,1] = Vrange[0] + shift1 * deltav |
|
|
646 | 646 | DGauFitParam[3,ht,0] = width0 * deltav |
|
|
647 | 647 | DGauFitParam[3,ht,1] = width1 * deltav |
|
|
648 | 648 | DGauFitParam[4,ht,0] = p0 |
|
|
649 | 649 | DGauFitParam[4,ht,1] = p1 |
|
|
650 | 650 | |
|
|
651 | 651 | return DGauFitParam |
|
|
652 | 652 | |
|
|
653 | 653 | def y_model1(self,x,state): |
|
|
654 | 654 | shift0, width0, amplitude0, power0, noise = state |
|
|
655 | 655 | model0 = amplitude0*numpy.exp(-0.5*abs((x - shift0)/width0)**power0) |
|
|
656 | 656 | model0u = amplitude0*numpy.exp(-0.5*abs((x - shift0 - self.Num_Bin)/width0)**power0) |
|
|
657 | 657 | model0d = amplitude0*numpy.exp(-0.5*abs((x - shift0 + self.Num_Bin)/width0)**power0) |
|
|
658 | 658 | return model0 + model0u + model0d + noise |
|
|
659 | 659 | |
|
|
660 | 660 | def y_model2(self,x,state): #Equation for two generalized Gaussians with Nyquist |
|
|
661 | 661 | shift0, width0, amplitude0, power0, shift1, width1, amplitude1, power1, noise = state |
|
|
662 | 662 | model0 = amplitude0*numpy.exp(-0.5*abs((x-shift0)/width0)**power0) |
|
|
663 | 663 | model0u = amplitude0*numpy.exp(-0.5*abs((x - shift0 - self.Num_Bin)/width0)**power0) |
|
|
664 | 664 | model0d = amplitude0*numpy.exp(-0.5*abs((x - shift0 + self.Num_Bin)/width0)**power0) |
|
|
665 | 665 | |
|
|
666 | 666 | model1 = amplitude1*numpy.exp(-0.5*abs((x - shift1)/width1)**power1) |
|
|
667 | 667 | model1u = amplitude1*numpy.exp(-0.5*abs((x - shift1 - self.Num_Bin)/width1)**power1) |
|
|
668 | 668 | model1d = amplitude1*numpy.exp(-0.5*abs((x - shift1 + self.Num_Bin)/width1)**power1) |
|
|
669 | 669 | return model0 + model0u + model0d + model1 + model1u + model1d + noise |
|
|
670 | 670 | |
|
|
671 | 671 | def misfit1(self,state,y_data,x,num_intg): # This function compares how close real data is with the model data, the close it is, the better it is. |
|
|
672 | 672 | |
|
|
673 | 673 | return num_intg*sum((numpy.log(y_data)-numpy.log(self.y_model1(x,state)))**2)#/(64-5.) # /(64-5.) can be commented |
|
|
674 | 674 | |
|
|
675 | 675 | def misfit2(self,state,y_data,x,num_intg): |
|
|
676 | 676 | return num_intg*sum((numpy.log(y_data)-numpy.log(self.y_model2(x,state)))**2)#/(64-9.) |
|
|
677 | 677 | |
|
|
678 | 678 | class Oblique_Gauss_Fit(Operation): |
|
|
679 | 679 | ''' |
|
|
680 | 680 | Written by R. Flores |
|
|
681 | 681 | ''' |
|
|
682 | 682 | def __init__(self): |
|
|
683 | 683 | Operation.__init__(self) |
|
|
684 | 684 | |
|
|
685 | 685 | def Gauss_fit(self,spc,x,nGauss): |
|
|
686 | 686 | |
|
|
687 | 687 | |
|
|
688 | 688 | def gaussian(x, a, b, c, d): |
|
|
689 | 689 | val = a * numpy.exp(-(x - b)**2 / (2*c**2)) + d |
|
|
690 | 690 | return val |
|
|
691 | 691 | |
|
|
692 | 692 | if nGauss == 'first': |
|
|
693 | 693 | spc_1_aux = numpy.copy(spc[:numpy.argmax(spc)+1]) |
|
|
694 | 694 | spc_2_aux = numpy.flip(spc_1_aux) |
|
|
695 | 695 | spc_3_aux = numpy.concatenate((spc_1_aux,spc_2_aux[1:])) |
|
|
696 | 696 | |
|
|
697 | 697 | len_dif = len(x)-len(spc_3_aux) |
|
|
698 | 698 | |
|
|
699 | 699 | spc_zeros = numpy.ones(len_dif)*spc_1_aux[0] |
|
|
700 | 700 | |
|
|
701 | 701 | spc_new = numpy.concatenate((spc_3_aux,spc_zeros)) |
|
|
702 | 702 | |
|
|
703 | 703 | y = spc_new |
|
|
704 | 704 | |
|
|
705 | 705 | elif nGauss == 'second': |
|
|
706 | 706 | y = spc |
|
|
707 | 707 | |
|
|
708 | 708 | |
|
|
709 | 709 | # estimate starting values from the data |
|
|
710 | 710 | a = y.max() |
|
|
711 | 711 | b = x[numpy.argmax(y)] |
|
|
712 | 712 | if nGauss == 'first': |
|
|
713 | 713 | c = 1.#b#b#numpy.std(spc) |
|
|
714 | 714 | elif nGauss == 'second': |
|
|
715 | 715 | c = b |
|
|
716 | 716 | else: |
|
|
717 | 717 | print("ERROR") |
|
|
718 | 718 | |
|
|
719 | 719 | d = numpy.mean(y[-100:]) |
|
|
720 | 720 | |
|
|
721 | 721 | # define a least squares function to optimize |
|
|
722 | 722 | def minfunc(params): |
|
|
723 | 723 | return sum((y-gaussian(x,params[0],params[1],params[2],params[3]))**2) |
|
|
724 | 724 | |
|
|
725 | 725 | # fit |
|
|
726 | 726 | popt = fmin(minfunc,[a,b,c,d],disp=False) |
|
|
727 | 727 | #popt,fopt,niter,funcalls = fmin(minfunc,[a,b,c,d]) |
|
|
728 | 728 | |
|
|
729 | 729 | |
|
|
730 | 730 | return gaussian(x, popt[0], popt[1], popt[2], popt[3]), popt[0], popt[1], popt[2], popt[3] |
|
|
731 | 731 | |
|
|
732 | 732 | def Gauss_fit_2(self,spc,x,nGauss): |
|
|
733 | 733 | |
|
|
734 | 734 | |
|
|
735 | 735 | def gaussian(x, a, b, c, d): |
|
|
736 | 736 | val = a * numpy.exp(-(x - b)**2 / (2*c**2)) + d |
|
|
737 | 737 | return val |
|
|
738 | 738 | |
|
|
739 | 739 | if nGauss == 'first': |
|
|
740 | 740 | spc_1_aux = numpy.copy(spc[:numpy.argmax(spc)+1]) |
|
|
741 | 741 | spc_2_aux = numpy.flip(spc_1_aux) |
|
|
742 | 742 | spc_3_aux = numpy.concatenate((spc_1_aux,spc_2_aux[1:])) |
|
|
743 | 743 | |
|
|
744 | 744 | len_dif = len(x)-len(spc_3_aux) |
|
|
745 | 745 | |
|
|
746 | 746 | spc_zeros = numpy.ones(len_dif)*spc_1_aux[0] |
|
|
747 | 747 | |
|
|
748 | 748 | spc_new = numpy.concatenate((spc_3_aux,spc_zeros)) |
|
|
749 | 749 | |
|
|
750 | 750 | y = spc_new |
|
|
751 | 751 | |
|
|
752 | 752 | elif nGauss == 'second': |
|
|
753 | 753 | y = spc |
|
|
754 | 754 | |
|
|
755 | 755 | |
|
|
756 | 756 | # estimate starting values from the data |
|
|
757 | 757 | a = y.max() |
|
|
758 | 758 | b = x[numpy.argmax(y)] |
|
|
759 | 759 | if nGauss == 'first': |
|
|
760 | 760 | c = 1.#b#b#numpy.std(spc) |
|
|
761 | 761 | elif nGauss == 'second': |
|
|
762 | 762 | c = b |
|
|
763 | 763 | else: |
|
|
764 | 764 | print("ERROR") |
|
|
765 | 765 | |
|
|
766 | 766 | d = numpy.mean(y[-100:]) |
|
|
767 | 767 | |
|
|
768 | 768 | # define a least squares function to optimize |
|
|
769 | 769 | popt,pcov = curve_fit(gaussian,x,y,p0=[a,b,c,d]) |
|
|
770 | 770 | #popt,fopt,niter,funcalls = fmin(minfunc,[a,b,c,d]) |
|
|
771 | 771 | |
|
|
772 | 772 | |
|
|
773 | 773 | #return gaussian(x, popt[0], popt[1], popt[2], popt[3]), popt[0], popt[1], popt[2], popt[3] |
|
|
774 | 774 | return gaussian(x, popt[0], popt[1], popt[2], popt[3]),popt[0], popt[1], popt[2], popt[3] |
|
|
775 | 775 | |
|
|
776 | 776 | def Double_Gauss_fit(self,spc,x,A1,B1,C1,A2,B2,C2,D): |
|
|
777 | 777 | |
|
|
778 | 778 | def double_gaussian(x, a1, b1, c1, a2, b2, c2, d): |
|
|
779 | 779 | val = a1 * numpy.exp(-(x - b1)**2 / (2*c1**2)) + a2 * numpy.exp(-(x - b2)**2 / (2*c2**2)) + d |
|
|
780 | 780 | return val |
|
|
781 | 781 | |
|
|
782 | 782 | |
|
|
783 | 783 | y = spc |
|
|
784 | 784 | |
|
|
785 | 785 | # estimate starting values from the data |
|
|
786 | 786 | a1 = A1 |
|
|
787 | 787 | b1 = B1 |
|
|
788 | 788 | c1 = C1#numpy.std(spc) |
|
|
789 | 789 | |
|
|
790 | 790 | a2 = A2#y.max() |
|
|
791 | 791 | b2 = B2#x[numpy.argmax(y)] |
|
|
792 | 792 | c2 = C2#numpy.std(spc) |
|
|
793 | 793 | d = D |
|
|
794 | 794 | |
|
|
795 | 795 | # define a least squares function to optimize |
|
|
796 | 796 | def minfunc(params): |
|
|
797 | 797 | return sum((y-double_gaussian(x,params[0],params[1],params[2],params[3],params[4],params[5],params[6]))**2) |
|
|
798 | 798 | |
|
|
799 | 799 | # fit |
|
|
800 | 800 | popt = fmin(minfunc,[a1,b1,c1,a2,b2,c2,d],disp=False) |
|
|
801 | 801 | |
|
|
802 | 802 | return double_gaussian(x, popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6]), popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
803 | 803 | |
|
|
804 | 804 | def Double_Gauss_fit_2(self,spc,x,A1,B1,C1,A2,B2,C2,D): |
|
|
805 | 805 | |
|
|
806 | 806 | def double_gaussian(x, a1, b1, c1, a2, b2, c2, d): |
|
|
807 | 807 | val = a1 * numpy.exp(-(x - b1)**2 / (2*c1**2)) + a2 * numpy.exp(-(x - b2)**2 / (2*c2**2)) + d |
|
|
808 | 808 | return val |
|
|
809 | 809 | |
|
|
810 | 810 | |
|
|
811 | 811 | y = spc |
|
|
812 | 812 | |
|
|
813 | 813 | # estimate starting values from the data |
|
|
814 | 814 | a1 = A1 |
|
|
815 | 815 | b1 = B1 |
|
|
816 | 816 | c1 = C1#numpy.std(spc) |
|
|
817 | 817 | |
|
|
818 | 818 | a2 = A2#y.max() |
|
|
819 | 819 | b2 = B2#x[numpy.argmax(y)] |
|
|
820 | 820 | c2 = C2#numpy.std(spc) |
|
|
821 | 821 | d = D |
|
|
822 | 822 | |
|
|
823 | 823 | # fit |
|
|
824 | 824 | |
|
|
825 | 825 | popt,pcov = curve_fit(double_gaussian,x,y,p0=[a1,b1,c1,a2,b2,c2,d]) |
|
|
826 | 826 | |
|
|
827 | 827 | error = numpy.sqrt(numpy.diag(pcov)) |
|
|
828 | 828 | |
|
|
829 | 829 | return popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6], error[0], error[1], error[2], error[3], error[4], error[5], error[6] |
|
|
830 | 830 | |
|
|
831 | 831 | def windowing_double(self,spc,x,A1,B1,C1,A2,B2,C2,D): |
|
|
832 | 832 | from scipy.optimize import curve_fit,fmin |
|
|
833 | 833 | |
|
|
834 | 834 | def R_gaussian(x, a, b, c): |
|
|
835 | 835 | N = int(numpy.shape(x)[0]) |
|
|
836 | 836 | val = a * numpy.exp(-((x)*c*2*2*numpy.pi)**2 / (2))* numpy.exp(1.j*b*x*4*numpy.pi) |
|
|
837 | 837 | return val |
|
|
838 | 838 | |
|
|
839 | 839 | def T(x,N): |
|
|
840 | 840 | T = 1-abs(x)/N |
|
|
841 | 841 | return T |
|
|
842 | 842 | |
|
|
843 | 843 | def R_T_spc_fun(x, a1, b1, c1, a2, b2, c2, d): |
|
|
844 | 844 | |
|
|
845 | 845 | N = int(numpy.shape(x)[0]) |
|
|
846 | 846 | |
|
|
847 | 847 | x_max = x[-1] |
|
|
848 | 848 | |
|
|
849 | 849 | x_pos = x[1600:] |
|
|
850 | 850 | x_neg = x[:1600] |
|
|
851 | 851 | |
|
|
852 | 852 | R_T_neg_1 = R_gaussian(x, a1, b1, c1)[:1600]*T(x_neg,-x[0]) |
|
|
853 | 853 | R_T_pos_1 = R_gaussian(x, a1, b1, c1)[1600:]*T(x_pos,x[-1]) |
|
|
854 | 854 | R_T_sum_1 = R_T_pos_1 + R_T_neg_1 |
|
|
855 | 855 | R_T_spc_1 = numpy.fft.fft(R_T_sum_1).real |
|
|
856 | 856 | R_T_spc_1 = numpy.fft.fftshift(R_T_spc_1) |
|
|
857 | 857 | max_val_1 = numpy.max(R_T_spc_1) |
|
|
858 | 858 | R_T_spc_1 = R_T_spc_1*a1/max_val_1 |
|
|
859 | 859 | |
|
|
860 | 860 | R_T_neg_2 = R_gaussian(x, a2, b2, c2)[:1600]*T(x_neg,-x[0]) |
|
|
861 | 861 | R_T_pos_2 = R_gaussian(x, a2, b2, c2)[1600:]*T(x_pos,x[-1]) |
|
|
862 | 862 | R_T_sum_2 = R_T_pos_2 + R_T_neg_2 |
|
|
863 | 863 | R_T_spc_2 = numpy.fft.fft(R_T_sum_2).real |
|
|
864 | 864 | R_T_spc_2 = numpy.fft.fftshift(R_T_spc_2) |
|
|
865 | 865 | max_val_2 = numpy.max(R_T_spc_2) |
|
|
866 | 866 | R_T_spc_2 = R_T_spc_2*a2/max_val_2 |
|
|
867 | 867 | |
|
|
868 | 868 | R_T_d = d*numpy.fft.fftshift(signal.unit_impulse(N)) |
|
|
869 | 869 | R_T_d_neg = R_T_d[:1600]*T(x_neg,-x[0]) |
|
|
870 | 870 | R_T_d_pos = R_T_d[1600:]*T(x_pos,x[-1]) |
|
|
871 | 871 | R_T_d_sum = R_T_d_pos + R_T_d_neg |
|
|
872 | 872 | R_T_spc_3 = numpy.fft.fft(R_T_d_sum).real |
|
|
873 | 873 | R_T_spc_3 = numpy.fft.fftshift(R_T_spc_3) |
|
|
874 | 874 | |
|
|
875 | 875 | R_T_final = R_T_spc_1 + R_T_spc_2 + R_T_spc_3 |
|
|
876 | 876 | |
|
|
877 | 877 | return R_T_final |
|
|
878 | 878 | |
|
|
879 | 879 | y = spc#gaussian(x, a, meanY, sigmaY) + a*0.1*numpy.random.normal(0, 1, size=len(x)) |
|
|
880 | 880 | |
|
|
881 | 881 | from scipy.stats import norm |
|
|
882 | 882 | mean,std=norm.fit(spc) |
|
|
883 | 883 | |
|
|
884 | 884 | # estimate starting values from the data |
|
|
885 | 885 | a1 = A1 |
|
|
886 | 886 | b1 = B1 |
|
|
887 | 887 | c1 = C1#numpy.std(spc) |
|
|
888 | 888 | |
|
|
889 | 889 | a2 = A2#y.max() |
|
|
890 | 890 | b2 = B2#x[numpy.argmax(y)] |
|
|
891 | 891 | c2 = C2#numpy.std(spc) |
|
|
892 | 892 | d = D |
|
|
893 | 893 | |
|
|
894 | 894 | ippSeconds = 250*20*1.e-6/3 |
|
|
895 | 895 | |
|
|
896 | 896 | x_t = ippSeconds * (numpy.arange(1600) -1600 / 2.) |
|
|
897 | 897 | |
|
|
898 | 898 | x_t = numpy.linspace(x_t[0],x_t[-1],3200) |
|
|
899 | 899 | |
|
|
900 | 900 | x_freq = numpy.fft.fftfreq(1600,d=ippSeconds) |
|
|
901 | 901 | x_freq = numpy.fft.fftshift(x_freq) |
|
|
902 | 902 | |
|
|
903 | 903 | # define a least squares function to optimize |
|
|
904 | 904 | def minfunc(params): |
|
|
905 | 905 | #print(params[2]) |
|
|
906 | 906 | #print(numpy.shape(params[2])) |
|
|
907 | 907 | return sum((y-R_T_spc_fun(x_t,params[0],params[1],params[2],params[3],params[4],params[5],params[6]))**2/1)#y**2) |
|
|
908 | 908 | |
|
|
909 | 909 | # fit |
|
|
910 | 910 | |
|
|
911 | 911 | popt_full = fmin(minfunc,[a1,b1,c1,a2,b2,c2,d],full_output=True) |
|
|
912 | 912 | #print("nIter", popt_full[2]) |
|
|
913 | 913 | popt = popt_full[0] |
|
|
914 | 914 | |
|
|
915 | 915 | #return R_T_spc_fun(x_t,popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6]), popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
916 | 916 | return popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
917 | 917 | |
|
|
918 | 918 | def Double_Gauss_fit_weight(self,spc,x,A1,B1,C1,A2,B2,C2,D): |
|
|
919 | 919 | from scipy.optimize import curve_fit,fmin |
|
|
920 | 920 | |
|
|
921 | 921 | def double_gaussian(x, a1, b1, c1, a2, b2, c2, d): |
|
|
922 | 922 | val = a1 * numpy.exp(-(x - b1)**2 / (2*c1**2)) + a2 * numpy.exp(-(x - b2)**2 / (2*c2**2)) + d |
|
|
923 | 923 | return val |
|
|
924 | 924 | |
|
|
925 | 925 | y = spc |
|
|
926 | 926 | |
|
|
927 | 927 | from scipy.stats import norm |
|
|
928 | 928 | mean,std=norm.fit(spc) |
|
|
929 | 929 | |
|
|
930 | 930 | # estimate starting values from the data |
|
|
931 | 931 | a1 = A1 |
|
|
932 | 932 | b1 = B1 |
|
|
933 | 933 | c1 = C1#numpy.std(spc) |
|
|
934 | 934 | |
|
|
935 | 935 | a2 = A2#y.max() |
|
|
936 | 936 | b2 = B2#x[numpy.argmax(y)] |
|
|
937 | 937 | c2 = C2#numpy.std(spc) |
|
|
938 | 938 | d = D |
|
|
939 | 939 | |
|
|
940 | 940 | y_clean = signal.medfilt(y) |
|
|
941 | 941 | # define a least squares function to optimize |
|
|
942 | 942 | def minfunc(params): |
|
|
943 | 943 | return sum((y-double_gaussian(x,params[0],params[1],params[2],params[3],params[4],params[5],params[6]))**2/(y_clean**2/1)) |
|
|
944 | 944 | |
|
|
945 | 945 | # fit |
|
|
946 | 946 | popt_full = fmin(minfunc,[a1,b1,c1,a2,b2,c2,d], disp =False, full_output=True) |
|
|
947 | 947 | #print("nIter", popt_full[2]) |
|
|
948 | 948 | popt = popt_full[0] |
|
|
949 | 949 | #popt,pcov = curve_fit(double_gaussian,x,y,p0=[a1,b1,c1,a2,b2,c2,d]) |
|
|
950 | 950 | |
|
|
951 | 951 | #return double_gaussian(x, popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6]), popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
952 | 952 | return popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
953 | 953 | |
|
|
954 | 954 | def DH_mode(self,spectra,VelRange): |
|
|
955 | 955 | |
|
|
956 | 956 | from scipy.optimize import curve_fit |
|
|
957 | 957 | |
|
|
958 | 958 | def double_gauss(x, a1,b1,c1, a2,b2,c2, d): |
|
|
959 | 959 | val = a1 * numpy.exp(-(x - b1)**2 / (2*c1**2)) + a2 * numpy.exp(-(x - b2)**2 / (2*c2**2)) + d |
|
|
960 | 960 | return val |
|
|
961 | 961 | |
|
|
962 | 962 | spec = (spectra.copy()).flatten() |
|
|
963 | 963 | amp=spec.max() |
|
|
964 | 964 | params=numpy.array([amp,-400,30,amp/4,-200,150,1.0e7]) |
|
|
965 | 965 | #try: |
|
|
966 | 966 | popt,pcov=curve_fit(double_gauss, VelRange, spec, p0=params,bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf])) |
|
|
967 | 967 | |
|
|
968 | 968 | error = numpy.sqrt(numpy.diag(pcov)) |
|
|
969 | 969 | #doppler_2=popt[4] |
|
|
970 | 970 | #err_2 = numpy.sqrt(pcov[4][4]) |
|
|
971 | 971 | |
|
|
972 | 972 | #except: |
|
|
973 | 973 | #pass |
|
|
974 | 974 | #doppler_2=numpy.NAN |
|
|
975 | 975 | #err_2 = numpy.NAN |
|
|
976 | 976 | |
|
|
977 | 977 | #return doppler_2, err_2 |
|
|
978 | 978 | |
|
|
979 | 979 | return popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6], error[0], error[1], error[2], error[3], error[4], error[5], error[6] |
|
|
980 | 980 | |
|
|
981 | 981 | def Tri_Marco(self,spc,freq,a1,b1,c1,a2,b2,c2,d): |
|
|
982 | 982 | |
|
|
983 | 983 | from scipy.optimize import least_squares |
|
|
984 | 984 | |
|
|
985 | 985 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
986 | 986 | spc_max = numpy.max(spc) |
|
|
987 | 987 | |
|
|
988 | 988 | def tri_gaussian(x, a1, b1, c1, a2, b2, c2, a3, b3, c3, d): |
|
|
989 | 989 | z1 = (x-b1)/c1 |
|
|
990 | 990 | z2 = (x-b2)/c2 |
|
|
991 | 991 | z3 = (x-b3)/c3 |
|
|
992 | 992 | val = a1 * numpy.exp(-z1**2/2) + a2 * numpy.exp(-z2**2/2) + a3 * numpy.exp(-z3**2/2) + d |
|
|
993 | 993 | return val |
|
|
994 | 994 | |
|
|
995 | 995 | from scipy.signal import medfilt |
|
|
996 | 996 | Nincoh = 20 |
|
|
997 | 997 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
998 | 998 | c1 = abs(c1) |
|
|
999 | 999 | c2 = abs(c2) |
|
|
1000 | 1000 | |
|
|
1001 | 1001 | # define a least squares function to optimize |
|
|
1002 | 1002 | def lsq_func(params): |
|
|
1003 | 1003 | return (spc-tri_gaussian(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6],params[7],params[8],params[9]))/spcm |
|
|
1004 | 1004 | |
|
|
1005 | 1005 | # fit |
|
|
1006 | 1006 | #bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf]) |
|
|
1007 | 1007 | bounds=([0,-numpy.inf,0,0,-numpy.inf,0,0,0,0,0],[numpy.inf,-100,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,600,numpy.inf,numpy.inf]) |
|
|
1008 | 1008 | #bounds=([0,-180,0,0,-100,30,0,110,0,0],[numpy.inf,-110,20,numpy.inf,33,80,numpy.inf,150,16,numpy.inf]) |
|
|
1009 | 1009 | #bounds=([0,-540,0,0,-300,100,0,330,0,0],[numpy.inf,-330,60,numpy.inf,100,240,numpy.inf,450,80,numpy.inf]) |
|
|
1010 | 1010 | |
|
|
1011 | 1011 | params_scale = [spc_max,freq_max,freq_max,spc_max,freq_max,freq_max,spc_max,freq_max,freq_max,spc_max] |
|
|
1012 | 1012 | #print(a1,b1,c1,a2,b2,c2,d) |
|
|
1013 | 1013 | popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,a2/4,-b1,c1,d],x_scale=params_scale,bounds=bounds) |
|
|
1014 | 1014 | |
|
|
1015 | 1015 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1016 | 1016 | A2f = popt.x[3]; B2f = popt.x[4]; C2f = popt.x[5] |
|
|
1017 | 1017 | A3f = popt.x[6]; B3f = popt.x[7]; C3f = popt.x[8] |
|
|
1018 | 1018 | Df = popt.x[9] |
|
|
1019 | 1019 | |
|
|
1020 | 1020 | return A1f, B1f, C1f, A2f, B2f, C2f, Df |
|
|
1021 | 1021 | |
|
|
1022 | 1022 | def Tri_Marco(self,spc,freq,a1,b1,c1,a2,b2,c2,d): |
|
|
1023 | 1023 | |
|
|
1024 | 1024 | from scipy.optimize import least_squares |
|
|
1025 | 1025 | |
|
|
1026 | 1026 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1027 | 1027 | spc_max = numpy.max(spc) |
|
|
1028 | 1028 | |
|
|
1029 | 1029 | def duo_gaussian(x, a1, b1, c1, a2, b2, c2, d): |
|
|
1030 | 1030 | z1 = (x-b1)/c1 |
|
|
1031 | 1031 | z2 = (x-b2)/c2 |
|
|
1032 | 1032 | #z3 = (x-b3)/c3 |
|
|
1033 | 1033 | val = a1 * numpy.exp(-z1**2/2) + a2 * numpy.exp(-z2**2/2) + d |
|
|
1034 | 1034 | return val |
|
|
1035 | 1035 | |
|
|
1036 | 1036 | from scipy.signal import medfilt |
|
|
1037 | 1037 | Nincoh = 20 |
|
|
1038 | 1038 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1039 | 1039 | c1 = abs(c1) |
|
|
1040 | 1040 | c2 = abs(c2) |
|
|
1041 | 1041 | |
|
|
1042 | 1042 | # define a least squares function to optimize |
|
|
1043 | 1043 | def lsq_func(params): |
|
|
1044 | 1044 | return (spc-tri_gaussian(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6]))/spcm |
|
|
1045 | 1045 | |
|
|
1046 | 1046 | # fit |
|
|
1047 | 1047 | #bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf]) |
|
|
1048 | 1048 | bounds=([0,-numpy.inf,0,0,-numpy.inf,0,0],[numpy.inf,-100,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf]) |
|
|
1049 | 1049 | #bounds=([0,-180,0,0,-100,30,0,110,0,0],[numpy.inf,-110,20,numpy.inf,33,80,numpy.inf,150,16,numpy.inf]) |
|
|
1050 | 1050 | #bounds=([0,-540,0,0,-300,100,0,330,0,0],[numpy.inf,-330,60,numpy.inf,100,240,numpy.inf,450,80,numpy.inf]) |
|
|
1051 | 1051 | |
|
|
1052 | 1052 | params_scale = [spc_max,freq_max,freq_max,spc_max,freq_max,freq_max,spc_max] |
|
|
1053 | 1053 | #print(a1,b1,c1,a2,b2,c2,d) |
|
|
1054 | 1054 | popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,d],x_scale=params_scale,bounds=bounds) |
|
|
1055 | 1055 | |
|
|
1056 | 1056 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1057 | 1057 | A2f = popt.x[3]; B2f = popt.x[4]; C2f = popt.x[5] |
|
|
1058 | 1058 | #A3f = popt.x[6]; B3f = popt.x[7]; C3f = popt.x[8] |
|
|
1059 | 1059 | Df = popt.x[9] |
|
|
1060 | 1060 | |
|
|
1061 | 1061 | return A1f, B1f, C1f, A2f, B2f, C2f, Df |
|
|
1062 | 1062 | |
|
|
1063 | 1063 | def double_gaussian_skew(self,x, a1, b1, c1, a2, b2, c2, k2, d): |
|
|
1064 | 1064 | #from scipy import special |
|
|
1065 | 1065 | z1 = (x-b1)/c1 |
|
|
1066 | 1066 | z2 = (x-b2)/c2 |
|
|
1067 | 1067 | h2 = 1-k2*z2 |
|
|
1068 | 1068 | h2[h2<0] = 0 |
|
|
1069 | 1069 | y2 = -1/k2*numpy.log(h2) |
|
|
1070 | 1070 | val = a1 * numpy.exp(-z1**2/2) + a2 * numpy.exp(-y2**2/2)/(1-k2*z2) + d |
|
|
1071 | 1071 | return val |
|
|
1072 | 1072 | |
|
|
1073 | 1073 | def gaussian(self, x, a, b, c, d): |
|
|
1074 | 1074 | z = (x-b)/c |
|
|
1075 | 1075 | val = a * numpy.exp(-z**2/2) + d |
|
|
1076 | 1076 | return val |
|
|
1077 | 1077 | |
|
|
1078 | 1078 | def double_gaussian(self, x, a1, b1, c1, a2, b2, c2, d): |
|
|
1079 | 1079 | z1 = (x-b1)/c1 |
|
|
1080 | 1080 | z2 = (x-b2)/c2 |
|
|
1081 | 1081 | val = a1 * numpy.exp(-z1**2/2) + a2 * numpy.exp(-z2**2/2) + d |
|
|
1082 | 1082 | return val |
|
|
1083 | 1083 | |
|
|
1084 | 1084 | def double_gaussian_double_skew(self,x, a1, b1, c1, k1, a2, b2, c2, k2, d): |
|
|
1085 | 1085 | |
|
|
1086 | 1086 | z1 = (x-b1)/c1 |
|
|
1087 | 1087 | h1 = 1-k1*z1 |
|
|
1088 | 1088 | h1[h1<0] = 0 |
|
|
1089 | 1089 | y1 = -1/k1*numpy.log(h1) |
|
|
1090 | 1090 | |
|
|
1091 | 1091 | z2 = (x-b2)/c2 |
|
|
1092 | 1092 | h2 = 1-k2*z2 |
|
|
1093 | 1093 | h2[h2<0] = 0 |
|
|
1094 | 1094 | y2 = -1/k2*numpy.log(h2) |
|
|
1095 | 1095 | |
|
|
1096 | 1096 | val = a1 * numpy.exp(-y1**2/2)/(1-k1*z1) + a2 * numpy.exp(-y2**2/2)/(1-k2*z2) + d |
|
|
1097 | 1097 | return val |
|
|
1098 | 1098 | |
|
|
1099 | 1099 | def gaussian_skew(self,x, a2, b2, c2, k2, d): |
|
|
1100 | 1100 | #from scipy import special |
|
|
1101 | 1101 | z2 = (x-b2)/c2 |
|
|
1102 | 1102 | h2 = 1-k2*z2 |
|
|
1103 | 1103 | h2[h2<0] = 0 |
|
|
1104 | 1104 | y2 = -1/k2*numpy.log(h2) |
|
|
1105 | 1105 | val = a2 * numpy.exp(-y2**2/2)/(1-k2*z2) + d |
|
|
1106 | 1106 | return val |
|
|
1107 | 1107 | |
|
|
1108 | 1108 | def triple_gaussian_skew(self,x, a1, b1, c1, a2, b2, c2, k2, a3, b3, c3, k3, d): |
|
|
1109 | 1109 | #from scipy import special |
|
|
1110 | 1110 | z1 = (x-b1)/c1 |
|
|
1111 | 1111 | z2 = (x-b2)/c2 |
|
|
1112 | 1112 | z3 = (x-b3)/c3 |
|
|
1113 | 1113 | h2 = 1-k2*z2 |
|
|
1114 | 1114 | h2[h2<0] = 0 |
|
|
1115 | 1115 | y2 = -1/k2*numpy.log(h2) |
|
|
1116 | 1116 | h3 = 1-k3*z3 |
|
|
1117 | 1117 | h3[h3<0] = 0 |
|
|
1118 | 1118 | y3 = -1/k3*numpy.log(h3) |
|
|
1119 | 1119 | val = a1 * numpy.exp(-z1**2/2) + a2 * numpy.exp(-y2**2/2)/(1-k2*z2) + a3 * numpy.exp(-y3**2/2)/(1-k3*z3) + d |
|
|
1120 | 1120 | return val |
|
|
1121 | 1121 | |
|
|
1122 | 1122 | def Double_Gauss_Skew_fit_weight_bound_no_inputs(self,spc,freq): |
|
|
1123 | 1123 | |
|
|
1124 | 1124 | from scipy.optimize import least_squares |
|
|
1125 | 1125 | |
|
|
1126 | 1126 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1127 | 1127 | spc_max = numpy.max(spc) |
|
|
1128 | 1128 | |
|
|
1129 | 1129 | from scipy.signal import medfilt |
|
|
1130 | 1130 | Nincoh = 20 |
|
|
1131 | 1131 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1132 | 1132 | |
|
|
1133 | 1133 | # define a least squares function to optimize |
|
|
1134 | 1134 | def lsq_func(params): |
|
|
1135 | 1135 | return (spc-self.double_gaussian_skew(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6],params[7]))/spcm |
|
|
1136 | 1136 | |
|
|
1137 | 1137 | # fit |
|
|
1138 | 1138 | # bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf]) |
|
|
1139 | 1139 | # bounds=([0,-numpy.inf,0,0,-numpy.inf,0,-numpy.inf,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,0,numpy.inf]) |
|
|
1140 | 1140 | #print(a1,b1,c1,a2,b2,c2,k2,d) |
|
|
1141 | 1141 | bounds=([0,-numpy.inf,0,0,-400,0,0,0],[numpy.inf,-340,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1142 | 1142 | #print(bounds) |
|
|
1143 | 1143 | #bounds=([0,-numpy.inf,0,0,-numpy.inf,0,0,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1144 | 1144 | params_scale = [spc_max,freq_max,freq_max,spc_max,freq_max,freq_max,1,spc_max] |
|
|
1145 | 1145 | x0_value = numpy.array([spc_max,-400,30,spc_max/4,-200,150,1,1.0e7]) |
|
|
1146 | 1146 | #popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=1) |
|
|
1147 | 1147 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1148 | 1148 | # popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x_scale=params_scale,verbose=1) |
|
|
1149 | 1149 | |
|
|
1150 | 1150 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1151 | 1151 | A2f = popt.x[3]; B2f = popt.x[4]; C2f = popt.x[5]; K2f = popt.x[6] |
|
|
1152 | 1152 | Df = popt.x[7] |
|
|
1153 | 1153 | |
|
|
1154 | 1154 | aux = self.gaussian_skew(freq, A2f, B2f, C2f, K2f, Df) |
|
|
1155 | 1155 | doppler = freq[numpy.argmax(aux)] |
|
|
1156 | 1156 | |
|
|
1157 | 1157 | #return A1f, B1f, C1f, A2f, B2f, C2f, K2f, Df, doppler |
|
|
1158 | 1158 | return A1f, B1f, C1f, A2f, B2f, C2f, K2f, Df, doppler |
|
|
1159 | 1159 | |
|
|
1160 | 1160 | def Double_Gauss_Double_Skew_fit_weight_bound_no_inputs(self,spc,freq,Nincoh,hei): |
|
|
1161 | 1161 | |
|
|
1162 | 1162 | from scipy.optimize import least_squares |
|
|
1163 | 1163 | |
|
|
1164 | 1164 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1165 | 1165 | spc_max = numpy.max(spc) |
|
|
1166 | 1166 | |
|
|
1167 | 1167 | #from scipy.signal import medfilt |
|
|
1168 | 1168 | #Nincoh = 20 |
|
|
1169 | 1169 | #Nincoh = 80 |
|
|
1170 | 1170 | Nincoh = Nincoh |
|
|
1171 | 1171 | #spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1172 | 1172 | spcm = spc/numpy.sqrt(Nincoh) |
|
|
1173 | 1173 | |
|
|
1174 | 1174 | # define a least squares function to optimize |
|
|
1175 | 1175 | def lsq_func(params): |
|
|
1176 | 1176 | return (spc-self.double_gaussian_double_skew(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6],params[7],params[8]))/spcm |
|
|
1177 | 1177 | |
|
|
1178 | 1178 | # fit |
|
|
1179 | 1179 | # bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf]) |
|
|
1180 | 1180 | # bounds=([0,-numpy.inf,0,0,-numpy.inf,0,-numpy.inf,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,0,numpy.inf]) |
|
|
1181 | 1181 | #print(a1,b1,c1,a2,b2,c2,k2,d) |
|
|
1182 | 1182 | #bounds=([0,-numpy.inf,0,-numpy.inf,0,-400,0,0,0],[numpy.inf,-340,numpy.inf,0,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1183 | 1183 | #bounds=([0,-numpy.inf,0,-numpy.inf,0,-400,0,0,0],[numpy.inf,-140,numpy.inf,0,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1184 | 1184 | bounds=([0,-numpy.inf,0,-5,0,-400,0,0,0],[numpy.inf,-200,numpy.inf,5,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1185 | 1185 | |
|
|
1186 | 1186 | #print(bounds) |
|
|
1187 | 1187 | #bounds=([0,-numpy.inf,0,0,-numpy.inf,0,0,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1188 | 1188 | params_scale = [spc_max,freq_max,freq_max,1,spc_max,freq_max,freq_max,1,spc_max] |
|
|
1189 | 1189 | ####################x0_value = numpy.array([spc_max,-400,30,-.1,spc_max/4,-200,150,1,1.0e7]) |
|
|
1190 | 1190 | |
|
|
1191 | 1191 | dop1_x0 = freq[numpy.argmax(spc)] |
|
|
1192 | 1192 | ####dop1_x0 = freq[numpy.argmax(spcm)] |
|
|
1193 | 1193 | if dop1_x0 < 0: |
|
|
1194 | 1194 | dop2_x0 = dop1_x0 + 100 |
|
|
1195 | 1195 | if dop1_x0 > 0: |
|
|
1196 | 1196 | dop2_x0 = dop1_x0 - 100 |
|
|
1197 | 1197 | |
|
|
1198 | 1198 | ###########x0_value = numpy.array([spc_max,-200.5,30,-.1,spc_max/4,-100.5,150,1,1.0e7]) |
|
|
1199 | 1199 | x0_value = numpy.array([spc_max,dop1_x0,30,-.1,spc_max/4, dop2_x0,150,1,1.0e7]) |
|
|
1200 | 1200 | #x0_value = numpy.array([spc_max,-400.5,30,-.1,spc_max/4,-200.5,150,1,1.0e7]) |
|
|
1201 | 1201 | #popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=1) |
|
|
1202 | 1202 | ''' |
|
|
1203 | 1203 | print("INSIDE 1") |
|
|
1204 | 1204 | print("x0_value: ", x0_value) |
|
|
1205 | 1205 | print("boundaries: ", bounds) |
|
|
1206 | 1206 | import matplotlib.pyplot as plt |
|
|
1207 | 1207 | plt.plot(freq,spc) |
|
|
1208 | 1208 | plt.plot(freq,self.double_gaussian_double_skew(freq,x0_value[0],x0_value[1],x0_value[2],x0_value[3],x0_value[4],x0_value[5],x0_value[6],x0_value[7],x0_value[8])) |
|
|
1209 | 1209 | plt.title(hei) |
|
|
1210 | 1210 | plt.show() |
|
|
1211 | 1211 | ''' |
|
|
1212 | 1212 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1213 | 1213 | # popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x_scale=params_scale,verbose=1) |
|
|
1214 | 1214 | #print(popt) |
|
|
1215 | 1215 | #########print("INSIDE 2") |
|
|
1216 | 1216 | J = popt.jac |
|
|
1217 | 1217 | |
|
|
1218 | 1218 | try: |
|
|
1219 | 1219 | cov = numpy.linalg.inv(J.T.dot(J)) |
|
|
1220 | 1220 | error = numpy.sqrt(numpy.diagonal(cov)) |
|
|
1221 | 1221 | except: |
|
|
1222 | 1222 | error = numpy.ones((9))*numpy.NAN |
|
|
1223 | 1223 | #print("error_inside",error) |
|
|
1224 | 1224 | #exit(1) |
|
|
1225 | 1225 | |
|
|
1226 | 1226 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2]; K1f = popt.x[3] |
|
|
1227 | 1227 | A2f = popt.x[4]; B2f = popt.x[5]; C2f = popt.x[6]; K2f = popt.x[7] |
|
|
1228 | 1228 | Df = popt.x[8] |
|
|
1229 | 1229 | ''' |
|
|
1230 | 1230 | A1f_err = error.x[0]; B1f_err= error.x[1]; C1f_err = error.x[2]; K1f_err = error.x[3] |
|
|
1231 | 1231 | A2f_err = error.x[4]; B2f_err = error.x[5]; C2f_err = error.x[6]; K2f_err = error.x[7] |
|
|
1232 | 1232 | Df_err = error.x[8] |
|
|
1233 | 1233 | ''' |
|
|
1234 | 1234 | aux1 = self.gaussian_skew(freq, A1f, B1f, C1f, K1f, Df) |
|
|
1235 | 1235 | doppler1 = freq[numpy.argmax(aux1)] |
|
|
1236 | 1236 | |
|
|
1237 | 1237 | aux2 = self.gaussian_skew(freq, A2f, B2f, C2f, K2f, Df) |
|
|
1238 | 1238 | doppler2 = freq[numpy.argmax(aux2)] |
|
|
1239 | 1239 | #print("error",error) |
|
|
1240 | 1240 | #exit(1) |
|
|
1241 | 1241 | |
|
|
1242 | 1242 | |
|
|
1243 | 1243 | return A1f, B1f, C1f, K1f, A2f, B2f, C2f, K2f, Df, doppler1, doppler2, error |
|
|
1244 | 1244 | |
|
|
1245 | 1245 | def Double_Gauss_fit_weight_bound_no_inputs(self,spc,freq,Nincoh): |
|
|
1246 | 1246 | |
|
|
1247 | 1247 | from scipy.optimize import least_squares |
|
|
1248 | 1248 | |
|
|
1249 | 1249 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1250 | 1250 | spc_max = numpy.max(spc) |
|
|
1251 | 1251 | |
|
|
1252 | 1252 | from scipy.signal import medfilt |
|
|
1253 | 1253 | Nincoh = 20 |
|
|
1254 | 1254 | Nincoh = 80 |
|
|
1255 | 1255 | Nincoh = Nincoh |
|
|
1256 | 1256 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1257 | 1257 | |
|
|
1258 | 1258 | # define a least squares function to optimize |
|
|
1259 | 1259 | def lsq_func(params): |
|
|
1260 | 1260 | return (spc-self.double_gaussian(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6]))/spcm |
|
|
1261 | 1261 | |
|
|
1262 | 1262 | # fit |
|
|
1263 | 1263 | # bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf]) |
|
|
1264 | 1264 | # bounds=([0,-numpy.inf,0,0,-numpy.inf,0,-numpy.inf,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,0,numpy.inf]) |
|
|
1265 | 1265 | #print(a1,b1,c1,a2,b2,c2,k2,d) |
|
|
1266 | 1266 | |
|
|
1267 | 1267 | dop1_x0 = freq[numpy.argmax(spcm)] |
|
|
1268 | 1268 | |
|
|
1269 | 1269 | #####bounds=([0,-numpy.inf,0,0,-400,0,0],[numpy.inf,-340,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf]) |
|
|
1270 | 1270 | #####bounds=([0,-numpy.inf,0,0,dop1_x0-50,0,0],[numpy.inf,-340,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf]) |
|
|
1271 | 1271 | bounds=([0,-numpy.inf,0,0,dop1_x0-50,0,0],[numpy.inf,-300,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf]) |
|
|
1272 | 1272 | #####bounds=([0,-numpy.inf,0,0,-500,0,0],[numpy.inf,-340,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf]) |
|
|
1273 | 1273 | #bounds=([0,-numpy.inf,0,-numpy.inf,0,-500,0,0,0],[numpy.inf,-240,numpy.inf,0,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1274 | 1274 | #print(bounds) |
|
|
1275 | 1275 | #bounds=([0,-numpy.inf,0,0,-numpy.inf,0,0,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1276 | 1276 | params_scale = [spc_max,freq_max,freq_max,spc_max,freq_max,freq_max,spc_max] |
|
|
1277 | 1277 | #x0_value = numpy.array([spc_max,-400.5,30,spc_max/4,-200.5,150,1.0e7]) |
|
|
1278 | 1278 | x0_value = numpy.array([spc_max,-400.5,30,spc_max/4,dop1_x0,150,1.0e7]) |
|
|
1279 | 1279 | #x0_value = numpy.array([spc_max,-420.5,30,-.1,spc_max/4,-50,150,.1,numpy.mean(spc[-50:])]) |
|
|
1280 | 1280 | #print("before popt") |
|
|
1281 | 1281 | #print(x0_value) |
|
|
1282 | 1282 | #print("freq: ",freq) |
|
|
1283 | 1283 | #popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=1) |
|
|
1284 | 1284 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1285 | 1285 | # popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x_scale=params_scale,verbose=1) |
|
|
1286 | 1286 | #print("after popt") |
|
|
1287 | 1287 | J = popt.jac |
|
|
1288 | 1288 | |
|
|
1289 | 1289 | try: |
|
|
1290 | 1290 | cov = numpy.linalg.inv(J.T.dot(J)) |
|
|
1291 | 1291 | error = numpy.sqrt(numpy.diagonal(cov)) |
|
|
1292 | 1292 | except: |
|
|
1293 | 1293 | error = numpy.ones((7))*numpy.NAN |
|
|
1294 | 1294 | |
|
|
1295 | 1295 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1296 | 1296 | A2f = popt.x[3]; B2f = popt.x[4]; C2f = popt.x[5] |
|
|
1297 | 1297 | Df = popt.x[6] |
|
|
1298 | 1298 | #print("before return") |
|
|
1299 | 1299 | return A1f, B1f, C1f, A2f, B2f, C2f, Df, error |
|
|
1300 | 1300 | |
|
|
1301 | 1301 | def Double_Gauss_Double_Skew_fit_weight_bound_with_inputs(self, spc, freq, a1, b1, c1, a2, b2, c2, k2, d): |
|
|
1302 | 1302 | |
|
|
1303 | 1303 | from scipy.optimize import least_squares |
|
|
1304 | 1304 | |
|
|
1305 | 1305 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1306 | 1306 | spc_max = numpy.max(spc) |
|
|
1307 | 1307 | |
|
|
1308 | 1308 | from scipy.signal import medfilt |
|
|
1309 | 1309 | Nincoh = dataOut.nIncohInt |
|
|
1310 | 1310 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1311 | 1311 | |
|
|
1312 | 1312 | # define a least squares function to optimize |
|
|
1313 | 1313 | def lsq_func(params): |
|
|
1314 | 1314 | return (spc-self.double_gaussian_double_skew(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6],params[7],params[8]))/spcm |
|
|
1315 | 1315 | |
|
|
1316 | 1316 | |
|
|
1317 | 1317 | bounds=([0,-numpy.inf,0,-numpy.inf,0,-400,0,0,0],[numpy.inf,-340,numpy.inf,0,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1318 | 1318 | |
|
|
1319 | 1319 | params_scale = [spc_max,freq_max,freq_max,1,spc_max,freq_max,freq_max,1,spc_max] |
|
|
1320 | 1320 | |
|
|
1321 | 1321 | x0_value = numpy.array([a1,b1,c1,-.1,a2,b2,c2,k2,d]) |
|
|
1322 | 1322 | |
|
|
1323 | 1323 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1324 | 1324 | |
|
|
1325 | 1325 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2]; K1f = popt.x[3] |
|
|
1326 | 1326 | A2f = popt.x[4]; B2f = popt.x[5]; C2f = popt.x[6]; K2f = popt.x[7] |
|
|
1327 | 1327 | Df = popt.x[8] |
|
|
1328 | 1328 | |
|
|
1329 | 1329 | aux = self.gaussian_skew(freq, A2f, B2f, C2f, K2f, Df) |
|
|
1330 | 1330 | doppler = x[numpy.argmax(aux)] |
|
|
1331 | 1331 | |
|
|
1332 | 1332 | return A1f, B1f, C1f, K1f, A2f, B2f, C2f, K2f, Df, doppler |
|
|
1333 | 1333 | |
|
|
1334 | 1334 | def Triple_Gauss_Skew_fit_weight_bound_no_inputs(self,spc,freq): |
|
|
1335 | 1335 | |
|
|
1336 | 1336 | from scipy.optimize import least_squares |
|
|
1337 | 1337 | |
|
|
1338 | 1338 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1339 | 1339 | spc_max = numpy.max(spc) |
|
|
1340 | 1340 | |
|
|
1341 | 1341 | from scipy.signal import medfilt |
|
|
1342 | 1342 | Nincoh = 20 |
|
|
1343 | 1343 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1344 | 1344 | |
|
|
1345 | 1345 | # define a least squares function to optimize |
|
|
1346 | 1346 | def lsq_func(params): |
|
|
1347 | 1347 | return (spc-self.triple_gaussian_skew(freq,params[0],params[1],params[2],params[3],params[4],params[5],params[6],params[7],params[8],params[9],params[10],params[11]))/spcm |
|
|
1348 | 1348 | |
|
|
1349 | 1349 | # fit |
|
|
1350 | 1350 | # bounds=([0,-460,0,0,-400,120,0],[numpy.inf,-340,50,numpy.inf,0,250,numpy.inf]) |
|
|
1351 | 1351 | # bounds=([0,-numpy.inf,0,0,-numpy.inf,0,-numpy.inf,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,0,numpy.inf]) |
|
|
1352 | 1352 | #print(a1,b1,c1,a2,b2,c2,k2,d) |
|
|
1353 | 1353 | bounds=([0,-numpy.inf,0,0,-400,0,0,0,0,0,0,0],[numpy.inf,-340,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf,numpy.inf,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1354 | 1354 | #print(bounds) |
|
|
1355 | 1355 | #bounds=([0,-numpy.inf,0,0,-numpy.inf,0,0,0],[numpy.inf,-200,numpy.inf,numpy.inf,0,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1356 | 1356 | params_scale = [spc_max,freq_max,freq_max,spc_max,freq_max,freq_max,1,spc_max,freq_max,freq_max,1,spc_max] |
|
|
1357 | 1357 | x0_value = numpy.array([spc_max,-400,30,spc_max/4,-200,150,1,spc_max/4,400,150,1,1.0e7]) |
|
|
1358 | 1358 | #popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=1) |
|
|
1359 | 1359 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1360 | 1360 | # popt = least_squares(lsq_func,[a1,b1,c1,a2,b2,c2,k2,d],x_scale=params_scale,verbose=1) |
|
|
1361 | 1361 | |
|
|
1362 | 1362 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1363 | 1363 | A2f = popt.x[3]; B2f = popt.x[4]; C2f = popt.x[5]; K2f = popt.x[6] |
|
|
1364 | 1364 | A3f = popt.x[7]; B3f = popt.x[8]; C3f = popt.x[9]; K3f = popt.x[10] |
|
|
1365 | 1365 | Df = popt.x[11] |
|
|
1366 | 1366 | |
|
|
1367 | 1367 | aux = self.gaussian_skew(freq, A2f, B2f, C2f, K2f, Df) |
|
|
1368 | 1368 | doppler = freq[numpy.argmax(aux)] |
|
|
1369 | 1369 | |
|
|
1370 | 1370 | return A1f, B1f, C1f, A2f, B2f, C2f, K2f, A3f, B3f, C3f, K3f, Df, doppler |
|
|
1371 | 1371 | |
|
|
1372 | 1372 | def CEEJ_Skew_fit_weight_bound_no_inputs(self,spc,freq,Nincoh): |
|
|
1373 | 1373 | |
|
|
1374 | 1374 | from scipy.optimize import least_squares |
|
|
1375 | 1375 | |
|
|
1376 | 1376 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1377 | 1377 | spc_max = numpy.max(spc) |
|
|
1378 | 1378 | |
|
|
1379 | 1379 | from scipy.signal import medfilt |
|
|
1380 | 1380 | Nincoh = 20 |
|
|
1381 | 1381 | Nincoh = 80 |
|
|
1382 | 1382 | Nincoh = Nincoh |
|
|
1383 | 1383 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1384 | 1384 | |
|
|
1385 | 1385 | # define a least squares function to optimize |
|
|
1386 | 1386 | def lsq_func(params): |
|
|
1387 | 1387 | return (spc-self.gaussian_skew(freq,params[0],params[1],params[2],params[3],params[4]))#/spcm |
|
|
1388 | 1388 | |
|
|
1389 | 1389 | |
|
|
1390 | 1390 | bounds=([0,0,0,-numpy.inf,0],[numpy.inf,numpy.inf,numpy.inf,0,numpy.inf]) |
|
|
1391 | 1391 | |
|
|
1392 | 1392 | params_scale = [spc_max,freq_max,freq_max,1,spc_max] |
|
|
1393 | 1393 | |
|
|
1394 | 1394 | x0_value = numpy.array([spc_max,freq[numpy.argmax(spc)],30,-.1,numpy.mean(spc[:50])]) |
|
|
1395 | 1395 | |
|
|
1396 | 1396 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1397 | 1397 | |
|
|
1398 | 1398 | J = popt.jac |
|
|
1399 | 1399 | |
|
|
1400 | 1400 | try: |
|
|
1401 | 1401 | error = numpy.ones((9))*numpy.NAN |
|
|
1402 | 1402 | cov = numpy.linalg.inv(J.T.dot(J)) |
|
|
1403 | 1403 | error[:4] = numpy.sqrt(numpy.diagonal(cov))[:4] |
|
|
1404 | 1404 | error[-1] = numpy.sqrt(numpy.diagonal(cov))[-1] |
|
|
1405 | 1405 | except: |
|
|
1406 | 1406 | error = numpy.ones((9))*numpy.NAN |
|
|
1407 | 1407 | |
|
|
1408 | 1408 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2]; K1f = popt.x[3] |
|
|
1409 | 1409 | Df = popt.x[4] |
|
|
1410 | 1410 | |
|
|
1411 | 1411 | aux1 = self.gaussian_skew(freq, A1f, B1f, C1f, K1f, Df) |
|
|
1412 | 1412 | doppler1 = freq[numpy.argmax(aux1)] |
|
|
1413 | 1413 | #print("CEEJ ERROR:",error) |
|
|
1414 | 1414 | |
|
|
1415 | 1415 | return A1f, B1f, C1f, K1f, numpy.NAN, numpy.NAN, numpy.NAN, numpy.NAN, Df, doppler1, numpy.NAN, error |
|
|
1416 | 1416 | |
|
|
1417 | 1417 | def CEEJ_fit_weight_bound_no_inputs(self,spc,freq,Nincoh): |
|
|
1418 | 1418 | |
|
|
1419 | 1419 | from scipy.optimize import least_squares |
|
|
1420 | 1420 | |
|
|
1421 | 1421 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1422 | 1422 | spc_max = numpy.max(spc) |
|
|
1423 | 1423 | |
|
|
1424 | 1424 | from scipy.signal import medfilt |
|
|
1425 | 1425 | Nincoh = 20 |
|
|
1426 | 1426 | Nincoh = 80 |
|
|
1427 | 1427 | Nincoh = Nincoh |
|
|
1428 | 1428 | spcm = medfilt(spc,11)/numpy.sqrt(Nincoh) |
|
|
1429 | 1429 | |
|
|
1430 | 1430 | # define a least squares function to optimize |
|
|
1431 | 1431 | def lsq_func(params): |
|
|
1432 | 1432 | return (spc-self.gaussian(freq,params[0],params[1],params[2],params[3]))#/spcm |
|
|
1433 | 1433 | |
|
|
1434 | 1434 | |
|
|
1435 | 1435 | bounds=([0,0,0,0],[numpy.inf,numpy.inf,numpy.inf,numpy.inf]) |
|
|
1436 | 1436 | |
|
|
1437 | 1437 | params_scale = [spc_max,freq_max,freq_max,spc_max] |
|
|
1438 | 1438 | |
|
|
1439 | 1439 | x0_value = numpy.array([spc_max,freq[numpy.argmax(spcm)],30,numpy.mean(spc[:50])]) |
|
|
1440 | 1440 | |
|
|
1441 | 1441 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1442 | 1442 | |
|
|
1443 | 1443 | J = popt.jac |
|
|
1444 | 1444 | |
|
|
1445 | 1445 | try: |
|
|
1446 | 1446 | error = numpy.ones((4))*numpy.NAN |
|
|
1447 | 1447 | cov = numpy.linalg.inv(J.T.dot(J)) |
|
|
1448 | 1448 | error = numpy.sqrt(numpy.diagonal(cov)) |
|
|
1449 | 1449 | except: |
|
|
1450 | 1450 | error = numpy.ones((4))*numpy.NAN |
|
|
1451 | 1451 | |
|
|
1452 | 1452 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1453 | 1453 | Df = popt.x[3] |
|
|
1454 | 1454 | |
|
|
1455 | 1455 | return A1f, B1f, C1f, Df, error |
|
|
1456 | 1456 | |
|
|
1457 | 1457 | def Simple_fit_bound(self,spc,freq,Nincoh): |
|
|
1458 | 1458 | |
|
|
1459 | 1459 | freq_max = numpy.max(numpy.abs(freq)) |
|
|
1460 | 1460 | spc_max = numpy.max(spc) |
|
|
1461 | 1461 | |
|
|
1462 | 1462 | Nincoh = Nincoh |
|
|
1463 | 1463 | |
|
|
1464 | 1464 | def lsq_func(params): |
|
|
1465 | 1465 | return (spc-self.gaussian(freq,params[0],params[1],params[2],params[3])) |
|
|
1466 | 1466 | |
|
|
1467 | 1467 | bounds=([0,-50,0,0],[numpy.inf,+50,numpy.inf,numpy.inf]) |
|
|
1468 | 1468 | |
|
|
1469 | 1469 | params_scale = [spc_max,freq_max,freq_max,spc_max] |
|
|
1470 | 1470 | |
|
|
1471 | 1471 | x0_value = numpy.array([spc_max,-20.5,5,1.0e7]) |
|
|
1472 | 1472 | |
|
|
1473 | 1473 | popt = least_squares(lsq_func,x0=x0_value,x_scale=params_scale,bounds=bounds,verbose=0) |
|
|
1474 | 1474 | |
|
|
1475 | 1475 | J = popt.jac |
|
|
1476 | 1476 | |
|
|
1477 | 1477 | try: |
|
|
1478 | 1478 | cov = numpy.linalg.inv(J.T.dot(J)) |
|
|
1479 | 1479 | error = numpy.sqrt(numpy.diagonal(cov)) |
|
|
1480 | 1480 | except: |
|
|
1481 | 1481 | error = numpy.ones((4))*numpy.NAN |
|
|
1482 | 1482 | |
|
|
1483 | 1483 | A1f = popt.x[0]; B1f = popt.x[1]; C1f = popt.x[2] |
|
|
1484 | 1484 | Df = popt.x[3] |
|
|
1485 | 1485 | |
|
|
1486 | 1486 | return A1f, B1f, C1f, Df, error |
|
|
1487 | 1487 | |
|
|
1488 | 1488 | def clean_outliers(self,param): |
|
|
1489 | 1489 | |
|
|
1490 | 1490 | threshold = 700 |
|
|
1491 | 1491 | |
|
|
1492 | 1492 | param = numpy.where(param < -threshold, numpy.nan, param) |
|
|
1493 | 1493 | param = numpy.where(param > +threshold, numpy.nan, param) |
|
|
1494 | 1494 | |
|
|
1495 | 1495 | return param |
|
|
1496 | 1496 | |
|
|
1497 | 1497 | def windowing_single(self,spc,x,A,B,C,D,nFFTPoints): |
|
|
1498 | 1498 | from scipy.optimize import curve_fit,fmin |
|
|
1499 | 1499 | |
|
|
1500 | 1500 | def R_gaussian(x, a, b, c): |
|
|
1501 | 1501 | N = int(numpy.shape(x)[0]) |
|
|
1502 | 1502 | val = a * numpy.exp(-((x)*c*2*2*numpy.pi)**2 / (2))* numpy.exp(1.j*b*x*4*numpy.pi) |
|
|
1503 | 1503 | return val |
|
|
1504 | 1504 | |
|
|
1505 | 1505 | def T(x,N): |
|
|
1506 | 1506 | T = 1-abs(x)/N |
|
|
1507 | 1507 | return T |
|
|
1508 | 1508 | |
|
|
1509 | 1509 | def R_T_spc_fun(x, a, b, c, d, nFFTPoints): |
|
|
1510 | 1510 | |
|
|
1511 | 1511 | N = int(numpy.shape(x)[0]) |
|
|
1512 | 1512 | |
|
|
1513 | 1513 | x_max = x[-1] |
|
|
1514 | 1514 | |
|
|
1515 | 1515 | x_pos = x[int(nFFTPoints/2):] |
|
|
1516 | 1516 | x_neg = x[:int(nFFTPoints/2)] |
|
|
1517 | 1517 | |
|
|
1518 | 1518 | R_T_neg_1 = R_gaussian(x, a, b, c)[:int(nFFTPoints/2)]*T(x_neg,-x[0]) |
|
|
1519 | 1519 | R_T_pos_1 = R_gaussian(x, a, b, c)[int(nFFTPoints/2):]*T(x_pos,x[-1]) |
|
|
1520 | 1520 | R_T_sum_1 = R_T_pos_1 + R_T_neg_1 |
|
|
1521 | 1521 | R_T_spc_1 = numpy.fft.fft(R_T_sum_1).real |
|
|
1522 | 1522 | R_T_spc_1 = numpy.fft.fftshift(R_T_spc_1) |
|
|
1523 | 1523 | max_val_1 = numpy.max(R_T_spc_1) |
|
|
1524 | 1524 | R_T_spc_1 = R_T_spc_1*a/max_val_1 |
|
|
1525 | 1525 | |
|
|
1526 | 1526 | R_T_d = d*numpy.fft.fftshift(signal.unit_impulse(N)) |
|
|
1527 | 1527 | R_T_d_neg = R_T_d[:int(nFFTPoints/2)]*T(x_neg,-x[0]) |
|
|
1528 | 1528 | R_T_d_pos = R_T_d[int(nFFTPoints/2):]*T(x_pos,x[-1]) |
|
|
1529 | 1529 | R_T_d_sum = R_T_d_pos + R_T_d_neg |
|
|
1530 | 1530 | R_T_spc_3 = numpy.fft.fft(R_T_d_sum).real |
|
|
1531 | 1531 | R_T_spc_3 = numpy.fft.fftshift(R_T_spc_3) |
|
|
1532 | 1532 | |
|
|
1533 | 1533 | R_T_final = R_T_spc_1 + R_T_spc_3 |
|
|
1534 | 1534 | |
|
|
1535 | 1535 | return R_T_final |
|
|
1536 | 1536 | |
|
|
1537 | 1537 | y = spc#gaussian(x, a, meanY, sigmaY) + a*0.1*numpy.random.normal(0, 1, size=len(x)) |
|
|
1538 | 1538 | |
|
|
1539 | 1539 | from scipy.stats import norm |
|
|
1540 | 1540 | mean,std=norm.fit(spc) |
|
|
1541 | 1541 | |
|
|
1542 | 1542 | # estimate starting values from the data |
|
|
1543 | 1543 | a = A |
|
|
1544 | 1544 | b = B |
|
|
1545 | 1545 | c = C#numpy.std(spc) |
|
|
1546 | 1546 | d = D |
|
|
1547 | 1547 | ''' |
|
|
1548 | 1548 | ippSeconds = 250*20*1.e-6/3 |
|
|
1549 | 1549 | |
|
|
1550 | 1550 | x_t = ippSeconds * (numpy.arange(1600) -1600 / 2.) |
|
|
1551 | 1551 | |
|
|
1552 | 1552 | x_t = numpy.linspace(x_t[0],x_t[-1],3200) |
|
|
1553 | 1553 | |
|
|
1554 | 1554 | x_freq = numpy.fft.fftfreq(1600,d=ippSeconds) |
|
|
1555 | 1555 | x_freq = numpy.fft.fftshift(x_freq) |
|
|
1556 | 1556 | ''' |
|
|
1557 | 1557 | # define a least squares function to optimize |
|
|
1558 | 1558 | def minfunc(params): |
|
|
1559 | 1559 | return sum((y-R_T_spc_fun(x,params[0],params[1],params[2],params[3],params[4],params[5],params[6]))**2/1)#y**2) |
|
|
1560 | 1560 | |
|
|
1561 | 1561 | # fit |
|
|
1562 | 1562 | |
|
|
1563 | 1563 | popt_full = fmin(minfunc,[a,b,c,d],full_output=True) |
|
|
1564 | 1564 | #print("nIter", popt_full[2]) |
|
|
1565 | 1565 | popt = popt_full[0] |
|
|
1566 | 1566 | |
|
|
1567 | 1567 | #return R_T_spc_fun(x_t,popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6]), popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
1568 | 1568 | return popt[0], popt[1], popt[2], popt[3] |
|
|
1569 | 1569 | |
|
|
1570 | 1570 | def run(self, dataOut, mode = 0, Hmin1 = None, Hmax1 = None, Hmin2 = None, Hmax2 = None, Dop = 'Shift'): |
|
|
1571 | 1571 | |
|
|
1572 | 1572 | pwcode = 1 |
|
|
1573 | 1573 | |
|
|
1574 | 1574 | if dataOut.flagDecodeData: |
|
|
1575 | 1575 | pwcode = numpy.sum(dataOut.code[0]**2) |
|
|
1576 | 1576 | #normFactor = min(self.nFFTPoints,self.nProfiles)*self.nIncohInt*self.nCohInt*pwcode*self.windowOfFilter |
|
|
1577 | 1577 | normFactor = dataOut.nProfiles * dataOut.nIncohInt * dataOut.nCohInt * pwcode * dataOut.windowOfFilter |
|
|
1578 | 1578 | factor = normFactor |
|
|
1579 | 1579 | z = dataOut.data_spc / factor |
|
|
1580 | 1580 | z = numpy.where(numpy.isfinite(z), z, numpy.NAN) |
|
|
1581 | 1581 | dataOut.power = numpy.average(z, axis=1) |
|
|
1582 | 1582 | dataOut.powerdB = 10 * numpy.log10(dataOut.power) |
|
|
1583 | 1583 | |
|
|
1584 | 1584 | x = dataOut.getVelRange(0) |
|
|
1585 | 1585 | |
|
|
1586 | 1586 | dataOut.Oblique_params = numpy.ones((1,7,dataOut.nHeights))*numpy.NAN |
|
|
1587 | 1587 | dataOut.Oblique_param_errors = numpy.ones((1,7,dataOut.nHeights))*numpy.NAN |
|
|
1588 | 1588 | dataOut.dplr_2_u = numpy.ones((1,1,dataOut.nHeights))*numpy.NAN |
|
|
1589 | 1589 | |
|
|
1590 | 1590 | if mode == 6: |
|
|
1591 | 1591 | dataOut.Oblique_params = numpy.ones((1,9,dataOut.nHeights))*numpy.NAN |
|
|
1592 | 1592 | elif mode == 7: |
|
|
1593 | 1593 | dataOut.Oblique_params = numpy.ones((1,13,dataOut.nHeights))*numpy.NAN |
|
|
1594 | 1594 | elif mode == 8: |
|
|
1595 | 1595 | dataOut.Oblique_params = numpy.ones((1,10,dataOut.nHeights))*numpy.NAN |
|
|
1596 | 1596 | elif mode == 9: |
|
|
1597 | 1597 | dataOut.Oblique_params = numpy.ones((1,11,dataOut.nHeights))*numpy.NAN |
|
|
1598 | 1598 | dataOut.Oblique_param_errors = numpy.ones((1,9,dataOut.nHeights))*numpy.NAN |
|
|
1599 | 1599 | elif mode == 11: |
|
|
1600 | 1600 | dataOut.Oblique_params = numpy.ones((1,7,dataOut.nHeights))*numpy.NAN |
|
|
1601 | 1601 | dataOut.Oblique_param_errors = numpy.ones((1,7,dataOut.nHeights))*numpy.NAN |
|
|
1602 | 1602 | elif mode == 10: #150 km |
|
|
1603 | 1603 | dataOut.Oblique_params = numpy.ones((1,4,dataOut.nHeights))*numpy.NAN |
|
|
1604 | 1604 | dataOut.Oblique_param_errors = numpy.ones((1,4,dataOut.nHeights))*numpy.NAN |
|
|
1605 | 1605 | dataOut.snr_log10 = numpy.ones((1,dataOut.nHeights))*numpy.NAN |
|
|
1606 | 1606 | |
|
|
1607 | 1607 | dataOut.VelRange = x |
|
|
1608 | 1608 | |
|
|
1609 | 1609 | |
|
|
1610 | 1610 | |
|
|
1611 | 1611 | #l1=range(22,36) #+62 |
|
|
1612 | 1612 | #l1=range(32,36) |
|
|
1613 | 1613 | #l2=range(58,99) #+62 |
|
|
1614 | 1614 | |
|
|
1615 | 1615 | #if Hmin1 == None or Hmax1 == None or Hmin2 == None or Hmax2 == None: |
|
|
1616 | 1616 | |
|
|
1617 | 1617 | minHei1 = 105. |
|
|
1618 | 1618 | maxHei1 = 122.5 |
|
|
1619 | 1619 | maxHei1 = 130.5 |
|
|
1620 | 1620 | |
|
|
1621 | 1621 | if mode == 10: #150 km |
|
|
1622 | 1622 | minHei1 = 100 |
|
|
1623 | 1623 | maxHei1 = 100 |
|
|
1624 | 1624 | |
|
|
1625 | 1625 | inda1 = numpy.where(dataOut.heightList >= minHei1) |
|
|
1626 | 1626 | indb1 = numpy.where(dataOut.heightList <= maxHei1) |
|
|
1627 | 1627 | |
|
|
1628 | 1628 | minIndex1 = inda1[0][0] |
|
|
1629 | 1629 | maxIndex1 = indb1[0][-1] |
|
|
1630 | 1630 | |
|
|
1631 | 1631 | minHei2 = 150. |
|
|
1632 | 1632 | maxHei2 = 201.25 |
|
|
1633 | 1633 | maxHei2 = 225.3 |
|
|
1634 | 1634 | |
|
|
1635 | 1635 | if mode == 10: #150 km |
|
|
1636 | 1636 | minHei2 = 110 |
|
|
1637 | 1637 | maxHei2 = 165 |
|
|
1638 | 1638 | |
|
|
1639 | 1639 | inda2 = numpy.where(dataOut.heightList >= minHei2) |
|
|
1640 | 1640 | indb2 = numpy.where(dataOut.heightList <= maxHei2) |
|
|
1641 | 1641 | |
|
|
1642 | 1642 | minIndex2 = inda2[0][0] |
|
|
1643 | 1643 | maxIndex2 = indb2[0][-1] |
|
|
1644 | 1644 | |
|
|
1645 | 1645 | l1=range(minIndex1,maxIndex1) |
|
|
1646 | 1646 | l2=range(minIndex2,maxIndex2) |
|
|
1647 | 1647 | |
|
|
1648 | 1648 | if mode == 4: |
|
|
1649 | 1649 | ''' |
|
|
1650 | 1650 | for ind in range(dataOut.nHeights): |
|
|
1651 | 1651 | if(dataOut.heightList[ind]>=168 and dataOut.heightList[ind]<188): |
|
|
1652 | 1652 | try: |
|
|
1653 | 1653 | dataOut.Oblique_params[0,0,ind],dataOut.Oblique_params[0,1,ind],dataOut.Oblique_params[0,2,ind],dataOut.Oblique_params[0,3,ind],dataOut.Oblique_params[0,4,ind],dataOut.Oblique_params[0,5,ind],dataOut.Oblique_params[0,6,ind],dataOut.Oblique_param_errors[0,0,ind],dataOut.Oblique_param_errors[0,1,ind],dataOut.Oblique_param_errors[0,2,ind],dataOut.Oblique_param_errors[0,3,ind],dataOut.Oblique_param_errors[0,4,ind],dataOut.Oblique_param_errors[0,5,ind],dataOut.Oblique_param_errors[0,6,ind] = self.DH_mode(dataOut.data_spc[0,:,ind],dataOut.VelRange) |
|
|
1654 | 1654 | except: |
|
|
1655 | 1655 | pass |
|
|
1656 | 1656 | ''' |
|
|
1657 | 1657 | for ind in itertools.chain(l1, l2): |
|
|
1658 | 1658 | |
|
|
1659 | 1659 | try: |
|
|
1660 | 1660 | dataOut.Oblique_params[0,0,ind],dataOut.Oblique_params[0,1,ind],dataOut.Oblique_params[0,2,ind],dataOut.Oblique_params[0,3,ind],dataOut.Oblique_params[0,4,ind],dataOut.Oblique_params[0,5,ind],dataOut.Oblique_params[0,6,ind],dataOut.Oblique_param_errors[0,0,ind],dataOut.Oblique_param_errors[0,1,ind],dataOut.Oblique_param_errors[0,2,ind],dataOut.Oblique_param_errors[0,3,ind],dataOut.Oblique_param_errors[0,4,ind],dataOut.Oblique_param_errors[0,5,ind],dataOut.Oblique_param_errors[0,6,ind] = self.DH_mode(dataOut.data_spc[0,:,ind],dataOut.VelRange) |
|
|
1661 | 1661 | dataOut.dplr_2_u[0,0,ind] = dataOut.Oblique_params[0,4,ind]/numpy.sin(numpy.arccos(102/dataOut.heightList[ind])) |
|
|
1662 | 1662 | except: |
|
|
1663 | 1663 | pass |
|
|
1664 | 1664 | |
|
|
1665 | 1665 | else: |
|
|
1666 | 1666 | #print("After: ", dataOut.data_snr[0]) |
|
|
1667 | 1667 | #######import matplotlib.pyplot as plt |
|
|
1668 | 1668 | #######plt.plot(dataOut.data_snr[0],dataOut.heightList,marker='*',linestyle='--') |
|
|
1669 | 1669 | #######plt.show() |
|
|
1670 | 1670 | #print("l1: ", dataOut.heightList[l1]) |
|
|
1671 | 1671 | #print("l2: ", dataOut.heightList[l2]) |
|
|
1672 | 1672 | for hei in itertools.chain(l1, l2): |
|
|
1673 | 1673 | #for hei in range(79,81): |
|
|
1674 | 1674 | #if numpy.isnan(dataOut.data_snr[0,hei]) or numpy.isnan(numpy.log10(dataOut.data_snr[0,hei])): |
|
|
1675 | 1675 | if numpy.isnan(dataOut.snl[0,hei]):# or dataOut.snl[0,hei]<.0: |
|
|
1676 | 1676 | |
|
|
1677 | 1677 | continue #Avoids the analysis when there is only noise |
|
|
1678 | 1678 | |
|
|
1679 | 1679 | try: |
|
|
1680 | 1680 | spc = dataOut.data_spc[0,:,hei] |
|
|
1681 | 1681 | |
|
|
1682 | 1682 | if mode == 6: #Skew Weighted Bounded |
|
|
1683 | 1683 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_params[0,7,hei],dataOut.Oblique_params[0,8,hei] = self.Double_Gauss_Skew_fit_weight_bound_no_inputs(spc,x) |
|
|
1684 | 1684 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,8,hei]/numpy.sin(numpy.arccos(100./dataOut.heightList[hei])) |
|
|
1685 | 1685 | |
|
|
1686 | 1686 | elif mode == 7: #Triple Skew Weighted Bounded |
|
|
1687 | 1687 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_params[0,7,hei],dataOut.Oblique_params[0,8,hei],dataOut.Oblique_params[0,9,hei],dataOut.Oblique_params[0,10,hei],dataOut.Oblique_params[0,11,hei],dataOut.Oblique_params[0,12,hei] = self.Triple_Gauss_Skew_fit_weight_bound_no_inputs(spc,x) |
|
|
1688 | 1688 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,12,hei]/numpy.sin(numpy.arccos(100./dataOut.heightList[hei])) |
|
|
1689 | 1689 | |
|
|
1690 | 1690 | elif mode == 8: #Double Skewed Weighted Bounded with inputs |
|
|
1691 | 1691 | a1, b1, c1, a2, b2, c2, k2, d, dopp = self.Double_Gauss_Skew_fit_weight_bound_no_inputs(spc,x) |
|
|
1692 | 1692 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_params[0,7,hei],dataOut.Oblique_params[0,8,hei],dataOut.Oblique_params[0,9,hei] = self.Double_Gauss_Skew_fit_weight_bound_no_inputs(spc,x, a1, b1, c1, a2, b2, c2, k2, d) |
|
|
1693 | 1693 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,9,hei]/numpy.sin(numpy.arccos(100./dataOut.heightList[hei])) |
|
|
1694 | 1694 | |
|
|
1695 | 1695 | elif mode == 9: #Double Skewed Weighted Bounded no inputs |
|
|
1696 | 1696 | #if numpy.max(spc) <= 0: |
|
|
1697 | 1697 | from scipy.signal import medfilt |
|
|
1698 | 1698 | spcm = medfilt(spc,11) |
|
|
1699 | 1699 | if x[numpy.argmax(spcm)] <= 0: |
|
|
1700 | 1700 | #print("EEJ", dataOut.heightList[hei], hei) |
|
|
1701 | 1701 | #if hei != 70: |
|
|
1702 | 1702 | #continue |
|
|
1703 | 1703 | #else: |
|
|
1704 | 1704 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_params[0,7,hei],dataOut.Oblique_params[0,8,hei],dataOut.Oblique_params[0,9,hei],dataOut.Oblique_params[0,10,hei],dataOut.Oblique_param_errors[0,:,hei] = self.Double_Gauss_Double_Skew_fit_weight_bound_no_inputs(spcm,x,dataOut.nIncohInt,dataOut.heightList[hei]) |
|
|
1705 | 1705 | #if dataOut.Oblique_params[0,-2,hei] < -500 or dataOut.Oblique_params[0,-2,hei] > 500 or dataOut.Oblique_params[0,-1,hei] < -500 or dataOut.Oblique_params[0,-1,hei] > 500: |
|
|
1706 | 1706 | # dataOut.Oblique_params[0,:,hei] *= numpy.NAN |
|
|
1707 | 1707 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,10,hei]/numpy.sin(numpy.arccos(100./dataOut.heightList[hei])) |
|
|
1708 | 1708 | |
|
|
1709 | 1709 | else: |
|
|
1710 | 1710 | #print("CEEJ") |
|
|
1711 | 1711 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_params[0,7,hei],dataOut.Oblique_params[0,8,hei],dataOut.Oblique_params[0,9,hei],dataOut.Oblique_params[0,10,hei],dataOut.Oblique_param_errors[0,:,hei] = self.CEEJ_Skew_fit_weight_bound_no_inputs(spcm,x,dataOut.nIncohInt) |
|
|
1712 | 1712 | #if dataOut.Oblique_params[0,-2,hei] < -500 or dataOut.Oblique_params[0,-2,hei] > 500 or dataOut.Oblique_params[0,-1,hei] < -500 or dataOut.Oblique_params[0,-1,hei] > 500: |
|
|
1713 | 1713 | # dataOut.Oblique_params[0,:,hei] *= numpy.NAN |
|
|
1714 | 1714 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,10,hei]/numpy.sin(numpy.arccos(100./dataOut.heightList[hei])) |
|
|
1715 | 1715 | elif mode == 11: #Double Weighted Bounded no inputs |
|
|
1716 | 1716 | #if numpy.max(spc) <= 0: |
|
|
1717 | 1717 | from scipy.signal import medfilt |
|
|
1718 | 1718 | spcm = medfilt(spc,11) |
|
|
1719 | 1719 | |
|
|
1720 | 1720 | if x[numpy.argmax(spcm)] <= 0: |
|
|
1721 | 1721 | #print("EEJ") |
|
|
1722 | 1722 | #print("EEJ",dataOut.heightList[hei]) |
|
|
1723 | 1723 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_param_errors[0,:,hei] = self.Double_Gauss_fit_weight_bound_no_inputs(spc,x,dataOut.nIncohInt) |
|
|
1724 | 1724 | #if dataOut.Oblique_params[0,-2,hei] < -500 or dataOut.Oblique_params[0,-2,hei] > 500 or dataOut.Oblique_params[0,-1,hei] < -500 or dataOut.Oblique_params[0,-1,hei] > 500: |
|
|
1725 | 1725 | # dataOut.Oblique_params[0,:,hei] *= numpy.NAN |
|
|
1726 | 1726 | else: |
|
|
1727 | 1727 | #print("CEEJ",dataOut.heightList[hei]) |
|
|
1728 | 1728 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_param_errors[0,:,hei] = self.CEEJ_fit_weight_bound_no_inputs(spc,x,dataOut.nIncohInt) |
|
|
1729 | 1729 | |
|
|
1730 | 1730 | elif mode == 10: #150km |
|
|
1731 | 1731 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_param_errors[0,:,hei] = self.Simple_fit_bound(spc,x,dataOut.nIncohInt) |
|
|
1732 | 1732 | snr = (dataOut.power[0,hei]*factor - dataOut.Oblique_params[0,3,hei])/dataOut.Oblique_params[0,3,hei] |
|
|
1733 | 1733 | dataOut.snr_log10[0,hei] = numpy.log10(snr) |
|
|
1734 | 1734 | |
|
|
1735 | 1735 | else: |
|
|
1736 | 1736 | spc_fit, A1, B1, C1, D1 = self.Gauss_fit_2(spc,x,'first') |
|
|
1737 | 1737 | |
|
|
1738 | 1738 | spc_diff = spc - spc_fit |
|
|
1739 | 1739 | spc_diff[spc_diff < 0] = 0 |
|
|
1740 | 1740 | |
|
|
1741 | 1741 | spc_fit_diff, A2, B2, C2, D2 = self.Gauss_fit_2(spc_diff,x,'second') |
|
|
1742 | 1742 | |
|
|
1743 | 1743 | D = (D1+D2) |
|
|
1744 | 1744 | |
|
|
1745 | 1745 | if mode == 0: #Double Fit |
|
|
1746 | 1746 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei],dataOut.Oblique_param_errors[0,0,hei],dataOut.Oblique_param_errors[0,1,hei],dataOut.Oblique_param_errors[0,2,hei],dataOut.Oblique_param_errors[0,3,hei],dataOut.Oblique_param_errors[0,4,hei],dataOut.Oblique_param_errors[0,5,hei],dataOut.Oblique_param_errors[0,6,hei] = self.Double_Gauss_fit_2(spc,x,A1,B1,C1,A2,B2,C2,D) |
|
|
1747 | 1747 | #spc_double_fit,dataOut.Oblique_params = self.Double_Gauss_fit(spc,x,A1,B1,C1,A2,B2,C2,D) |
|
|
1748 | 1748 | |
|
|
1749 | 1749 | elif mode == 1: #Double Fit Windowed |
|
|
1750 | 1750 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei] = self.windowing_double(spc,dataOut.getFreqRange(0),A1,B1,C1,A2,B2,C2,D) |
|
|
1751 | 1751 | |
|
|
1752 | 1752 | elif mode == 2: #Double Fit Weight |
|
|
1753 | 1753 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei] = self.Double_Gauss_fit_weight(spc,x,A1,B1,C1,A2,B2,C2,D) |
|
|
1754 | 1754 | |
|
|
1755 | 1755 | elif mode == 3: #Simple Fit |
|
|
1756 | 1756 | dataOut.Oblique_params[0,0,hei] = A1 |
|
|
1757 | 1757 | dataOut.Oblique_params[0,1,hei] = B1 |
|
|
1758 | 1758 | dataOut.Oblique_params[0,2,hei] = C1 |
|
|
1759 | 1759 | dataOut.Oblique_params[0,3,hei] = A2 |
|
|
1760 | 1760 | dataOut.Oblique_params[0,4,hei] = B2 |
|
|
1761 | 1761 | dataOut.Oblique_params[0,5,hei] = C2 |
|
|
1762 | 1762 | dataOut.Oblique_params[0,6,hei] = D |
|
|
1763 | 1763 | |
|
|
1764 | 1764 | elif mode == 5: #Triple Fit Weight |
|
|
1765 | 1765 | if hei in l1: |
|
|
1766 | 1766 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei] = self.duo_Marco(spc,x,A1,B1,C1,A2,B2,C2,D) |
|
|
1767 | 1767 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,4,hei]/numpy.sin(numpy.arccos(102/dataOut.heightList[hei])) |
|
|
1768 | 1768 | #print(dataOut.Oblique_params[0,0,hei]) |
|
|
1769 | 1769 | #print(dataOut.dplr_2_u[0,0,hei]) |
|
|
1770 | 1770 | else: |
|
|
1771 | 1771 | dataOut.Oblique_params[0,0,hei],dataOut.Oblique_params[0,1,hei],dataOut.Oblique_params[0,2,hei],dataOut.Oblique_params[0,3,hei],dataOut.Oblique_params[0,4,hei],dataOut.Oblique_params[0,5,hei],dataOut.Oblique_params[0,6,hei] = self.Double_Gauss_fit_weight(spc,x,A1,B1,C1,A2,B2,C2,D) |
|
|
1772 | 1772 | dataOut.dplr_2_u[0,0,hei] = dataOut.Oblique_params[0,4,hei]/numpy.sin(numpy.arccos(102/dataOut.heightList[hei])) |
|
|
1773 | 1773 | |
|
|
1774 | 1774 | |
|
|
1775 | 1775 | except: |
|
|
1776 | 1776 | ###dataOut.Oblique_params[0,:,hei] = dataOut.Oblique_params[0,:,hei]*numpy.NAN |
|
|
1777 | 1777 | pass |
|
|
1778 | 1778 | |
|
|
1779 | 1779 | #exit(1) |
|
|
1780 | 1780 | dataOut.paramInterval = dataOut.nProfiles*dataOut.nCohInt*dataOut.ippSeconds |
|
|
1781 | 1781 | dataOut.lat=-11.95 |
|
|
1782 | 1782 | dataOut.lon=-76.87 |
|
|
1783 | 1783 | ''' |
|
|
1784 | 1784 | dataOut.Oblique_params = numpy.where(dataOut.Oblique_params<-700, numpy.nan, dop_t1) |
|
|
1785 | 1785 | dataOut.Oblique_params = numpy.where(dataOut.Oblique_params<+700, numpy.nan, dop_t1) |
|
|
1786 | 1786 | AquΓ debo exceptuar las amplitudes |
|
|
1787 | 1787 | ''' |
|
|
1788 | 1788 | if mode == 9: #Double Skew Gaussian |
|
|
1789 | 1789 | #dataOut.Dop_EEJ_T1 = dataOut.Oblique_params[:,-2,:] #Pos[Max_value] |
|
|
1790 | 1790 | #dataOut.Dop_EEJ_T1 = dataOut.Oblique_params[:,1,:] #Shift |
|
|
1791 | 1791 | dataOut.Spec_W_T1 = dataOut.Oblique_params[:,2,:] |
|
|
1792 | 1792 | #dataOut.Dop_EEJ_T2 = dataOut.Oblique_params[:,-1,:] #Pos[Max_value] |
|
|
1793 | 1793 | #dataOut.Dop_EEJ_T2 = dataOut.Oblique_params[:,5,:] #Shift |
|
|
1794 | 1794 | dataOut.Spec_W_T2 = dataOut.Oblique_params[:,6,:] |
|
|
1795 | 1795 | if Dop == 'Shift': |
|
|
1796 | 1796 | dataOut.Dop_EEJ_T1 = dataOut.Oblique_params[:,1,:] #Shift |
|
|
1797 | 1797 | dataOut.Dop_EEJ_T2 = dataOut.Oblique_params[:,5,:] #Shift |
|
|
1798 | 1798 | elif Dop == 'Max': |
|
|
1799 | 1799 | dataOut.Dop_EEJ_T1 = dataOut.Oblique_params[:,-2,:] #Pos[Max_value] |
|
|
1800 | 1800 | dataOut.Dop_EEJ_T2 = dataOut.Oblique_params[:,-1,:] #Pos[Max_value] |
|
|
1801 | 1801 | |
|
|
1802 | 1802 | dataOut.Err_Dop_EEJ_T1 = dataOut.Oblique_param_errors[:,1,:] #En realidad este es el error? |
|
|
1803 | 1803 | dataOut.Err_Spec_W_T1 = dataOut.Oblique_param_errors[:,2,:] |
|
|
1804 | 1804 | dataOut.Err_Dop_EEJ_T2 = dataOut.Oblique_param_errors[:,5,:] #En realidad este es el error? |
|
|
1805 | 1805 | dataOut.Err_Spec_W_T2 = dataOut.Oblique_param_errors[:,6,:] |
|
|
1806 | 1806 | |
|
|
1807 | 1807 | elif mode == 11: #Double Gaussian |
|
|
1808 | 1808 | dataOut.Dop_EEJ_T1 = dataOut.Oblique_params[:,1,:] |
|
|
1809 | 1809 | dataOut.Spec_W_T1 = dataOut.Oblique_params[:,2,:] |
|
|
1810 | 1810 | dataOut.Dop_EEJ_T2 = dataOut.Oblique_params[:,4,:] |
|
|
1811 | 1811 | dataOut.Spec_W_T2 = dataOut.Oblique_params[:,5,:] |
|
|
1812 | 1812 | |
|
|
1813 | 1813 | dataOut.Err_Dop_EEJ_T1 = dataOut.Oblique_param_errors[:,1,:] |
|
|
1814 | 1814 | dataOut.Err_Spec_W_T1 = dataOut.Oblique_param_errors[:,2,:] |
|
|
1815 | 1815 | dataOut.Err_Dop_EEJ_T2 = dataOut.Oblique_param_errors[:,4,:] |
|
|
1816 | 1816 | dataOut.Err_Spec_W_T2 = dataOut.Oblique_param_errors[:,5,:] |
|
|
1817 | 1817 | |
|
|
1818 | 1818 | #print("Before: ", dataOut.Dop_EEJ_T2) |
|
|
1819 | 1819 | dataOut.Spec_W_T1 = self.clean_outliers(dataOut.Spec_W_T1) |
|
|
1820 | 1820 | dataOut.Spec_W_T2 = self.clean_outliers(dataOut.Spec_W_T2) |
|
|
1821 | 1821 | dataOut.Dop_EEJ_T1 = self.clean_outliers(dataOut.Dop_EEJ_T1) |
|
|
1822 | 1822 | dataOut.Dop_EEJ_T2 = self.clean_outliers(dataOut.Dop_EEJ_T2) |
|
|
1823 | 1823 | #print("After: ", dataOut.Dop_EEJ_T2) |
|
|
1824 | 1824 | dataOut.Err_Spec_W_T1 = self.clean_outliers(dataOut.Err_Spec_W_T1) |
|
|
1825 | 1825 | dataOut.Err_Spec_W_T2 = self.clean_outliers(dataOut.Err_Spec_W_T2) |
|
|
1826 | 1826 | dataOut.Err_Dop_EEJ_T1 = self.clean_outliers(dataOut.Err_Dop_EEJ_T1) |
|
|
1827 | 1827 | dataOut.Err_Dop_EEJ_T2 = self.clean_outliers(dataOut.Err_Dop_EEJ_T2) |
|
|
1828 | 1828 | #print("Before data_snr: ", dataOut.data_snr) |
|
|
1829 | 1829 | #dataOut.data_snr = numpy.where(numpy.isnan(dataOut.Dop_EEJ_T1), numpy.nan, dataOut.data_snr) |
|
|
1830 | 1830 | dataOut.snl = numpy.where(numpy.isnan(dataOut.Dop_EEJ_T1), numpy.nan, dataOut.snl) |
|
|
1831 | 1831 | |
|
|
1832 | 1832 | #print("After data_snr: ", dataOut.data_snr) |
|
|
1833 | 1833 | dataOut.mode = mode |
|
|
1834 | 1834 | dataOut.flagNoData = numpy.all(numpy.isnan(dataOut.Dop_EEJ_T1)) #Si todos los valores son NaN no se prosigue |
|
|
1835 | 1835 | ###dataOut.flagNoData = False #Descomentar solo para ploteo sino mantener comentado (para guardado) |
|
|
1836 | 1836 | |
|
|
1837 | 1837 | return dataOut |
|
|
1838 | 1838 | |
|
|
1839 | 1839 | class Gaussian_Windowed(Operation): |
|
|
1840 | 1840 | ''' |
|
|
1841 | 1841 | Written by R. Flores |
|
|
1842 | 1842 | ''' |
|
|
1843 | 1843 | def __init__(self): |
|
|
1844 | 1844 | Operation.__init__(self) |
|
|
1845 | 1845 | |
|
|
1846 | 1846 | def windowing_single(self,spc,x,A,B,C,D,nFFTPoints): |
|
|
1847 | 1847 | from scipy.optimize import curve_fit,fmin |
|
|
1848 | 1848 | |
|
|
1849 | 1849 | def gaussian(x, a, b, c, d): |
|
|
1850 | 1850 | val = a * numpy.exp(-(x - b)**2 / (2*c**2)) + d |
|
|
1851 | 1851 | return val |
|
|
1852 | 1852 | |
|
|
1853 | 1853 | def R_gaussian(x, a, b, c): |
|
|
1854 | 1854 | N = int(numpy.shape(x)[0]) |
|
|
1855 | 1855 | val = a * numpy.exp(-((x)*c*2*2*numpy.pi)**2 / (2))* numpy.exp(1.j*b*x*4*numpy.pi) |
|
|
1856 | 1856 | return val |
|
|
1857 | 1857 | |
|
|
1858 | 1858 | def T(x,N): |
|
|
1859 | 1859 | T = 1-abs(x)/N |
|
|
1860 | 1860 | return T |
|
|
1861 | 1861 | |
|
|
1862 | 1862 | def R_T_spc_fun(x, a, b, c, d, nFFTPoints): |
|
|
1863 | 1863 | |
|
|
1864 | 1864 | N = int(numpy.shape(x)[0]) |
|
|
1865 | 1865 | |
|
|
1866 | 1866 | x_max = x[-1] |
|
|
1867 | 1867 | |
|
|
1868 | 1868 | x_pos = x[nFFTPoints:] |
|
|
1869 | 1869 | x_neg = x[:nFFTPoints] |
|
|
1870 | 1870 | #print([int(nFFTPoints/2)) |
|
|
1871 | 1871 | #print("x: ", x) |
|
|
1872 | 1872 | #print("x_neg: ", x_neg) |
|
|
1873 | 1873 | #print("x_pos: ", x_pos) |
|
|
1874 | 1874 | |
|
|
1875 | 1875 | |
|
|
1876 | 1876 | R_T_neg_1 = R_gaussian(x, a, b, c)[:nFFTPoints]*T(x_neg,-x[0]) |
|
|
1877 | 1877 | R_T_pos_1 = R_gaussian(x, a, b, c)[nFFTPoints:]*T(x_pos,x[-1]) |
|
|
1878 | 1878 | #print(T(x_pos,x[-1]),x_pos,x[-1]) |
|
|
1879 | 1879 | #print(R_T_neg_1.shape,R_T_pos_1.shape) |
|
|
1880 | 1880 | R_T_sum_1 = R_T_pos_1 + R_T_neg_1 |
|
|
1881 | 1881 | R_T_spc_1 = numpy.fft.fft(R_T_sum_1).real |
|
|
1882 | 1882 | R_T_spc_1 = numpy.fft.fftshift(R_T_spc_1) |
|
|
1883 | 1883 | max_val_1 = numpy.max(R_T_spc_1) |
|
|
1884 | 1884 | R_T_spc_1 = R_T_spc_1*a/max_val_1 |
|
|
1885 | 1885 | |
|
|
1886 | 1886 | R_T_d = d*numpy.fft.fftshift(signal.unit_impulse(N)) |
|
|
1887 | 1887 | R_T_d_neg = R_T_d[:nFFTPoints]*T(x_neg,-x[0]) |
|
|
1888 | 1888 | R_T_d_pos = R_T_d[nFFTPoints:]*T(x_pos,x[-1]) |
|
|
1889 | 1889 | R_T_d_sum = R_T_d_pos + R_T_d_neg |
|
|
1890 | 1890 | R_T_spc_3 = numpy.fft.fft(R_T_d_sum).real |
|
|
1891 | 1891 | R_T_spc_3 = numpy.fft.fftshift(R_T_spc_3) |
|
|
1892 | 1892 | |
|
|
1893 | 1893 | R_T_final = R_T_spc_1 + R_T_spc_3 |
|
|
1894 | 1894 | |
|
|
1895 | 1895 | return R_T_final |
|
|
1896 | 1896 | |
|
|
1897 | 1897 | y = spc#gaussian(x, a, meanY, sigmaY) + a*0.1*numpy.random.normal(0, 1, size=len(x)) |
|
|
1898 | 1898 | |
|
|
1899 | 1899 | from scipy.stats import norm |
|
|
1900 | 1900 | mean,std=norm.fit(spc) |
|
|
1901 | 1901 | |
|
|
1902 | 1902 | # estimate starting values from the data |
|
|
1903 | 1903 | a = A |
|
|
1904 | 1904 | b = B |
|
|
1905 | 1905 | c = C#numpy.std(spc) |
|
|
1906 | 1906 | d = D |
|
|
1907 | 1907 | #''' |
|
|
1908 | 1908 | #ippSeconds = 250*20*1.e-6/3 |
|
|
1909 | 1909 | |
|
|
1910 | 1910 | #x_t = ippSeconds * (numpy.arange(nFFTPoints) - nFFTPoints / 2.) |
|
|
1911 | 1911 | |
|
|
1912 | 1912 | #x_t = numpy.linspace(x_t[0],x_t[-1],3200) |
|
|
1913 | 1913 | #print("x_t: ", x_t) |
|
|
1914 | 1914 | #print("nFFTPoints: ", nFFTPoints) |
|
|
1915 | 1915 | x_vel = numpy.linspace(x[0],x[-1],int(2*nFFTPoints)) |
|
|
1916 | 1916 | #print("x_vel: ", x_vel) |
|
|
1917 | 1917 | #x_freq = numpy.fft.fftfreq(1600,d=ippSeconds) |
|
|
1918 | 1918 | #x_freq = numpy.fft.fftshift(x_freq) |
|
|
1919 | 1919 | #''' |
|
|
1920 | 1920 | # define a least squares function to optimize |
|
|
1921 | 1921 | def minfunc(params): |
|
|
1922 | 1922 | #print("y.shape: ", numpy.shape(y)) |
|
|
1923 | 1923 | return sum((y-R_T_spc_fun(x_vel,params[0],params[1],params[2],params[3],nFFTPoints))**2/1)#y**2) |
|
|
1924 | 1924 | |
|
|
1925 | 1925 | # fit |
|
|
1926 | 1926 | |
|
|
1927 | 1927 | popt_full = fmin(minfunc,[a,b,c,d], disp=False) |
|
|
1928 | 1928 | #print("nIter", popt_full[2]) |
|
|
1929 | 1929 | popt = popt_full#[0] |
|
|
1930 | 1930 | |
|
|
1931 | 1931 | fun = gaussian(x, popt[0], popt[1], popt[2], popt[3]) |
|
|
1932 | 1932 | |
|
|
1933 | 1933 | #return R_T_spc_fun(x_t,popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6]), popt[0], popt[1], popt[2], popt[3], popt[4], popt[5], popt[6] |
|
|
1934 | 1934 | return fun, popt[0], popt[1], popt[2], popt[3] |
|
|
1935 | 1935 | |
|
|
1936 | 1936 | def run(self, dataOut): |
|
|
1937 | 1937 | |
|
|
1938 | 1938 | from scipy.signal import medfilt |
|
|
1939 | 1939 | import matplotlib.pyplot as plt |
|
|
1940 | 1940 | dataOut.moments = numpy.ones((dataOut.nChannels,4,dataOut.nHeights))*numpy.NAN |
|
|
1941 | 1941 | dataOut.VelRange = dataOut.getVelRange(0) |
|
|
1942 | 1942 | for nChannel in range(dataOut.nChannels): |
|
|
1943 | 1943 | for hei in range(dataOut.heightList.shape[0]): |
|
|
1944 | 1944 | #print("ipp: ", dataOut.ippSeconds) |
|
|
1945 | 1945 | spc = numpy.copy(dataOut.data_spc[nChannel,:,hei]) |
|
|
1946 | 1946 | |
|
|
1947 | 1947 | #print(VelRange) |
|
|
1948 | 1948 | #print(dataOut.getFreqRange(64)) |
|
|
1949 | 1949 | spcm = medfilt(spc,11) |
|
|
1950 | 1950 | spc_max = numpy.max(spcm) |
|
|
1951 | 1951 | dop1_x0 = dataOut.VelRange[numpy.argmax(spcm)] |
|
|
1952 | 1952 | D = numpy.min(spcm) |
|
|
1953 | 1953 | |
|
|
1954 | 1954 | fun, A, B, C, D = self.windowing_single(spc,dataOut.VelRange,spc_max,dop1_x0,abs(dop1_x0),D,dataOut.nFFTPoints) |
|
|
1955 | 1955 | dataOut.moments[nChannel,0,hei] = A |
|
|
1956 | 1956 | dataOut.moments[nChannel,1,hei] = B |
|
|
1957 | 1957 | dataOut.moments[nChannel,2,hei] = C |
|
|
1958 | 1958 | dataOut.moments[nChannel,3,hei] = D |
|
|
1959 | 1959 | ''' |
|
|
1960 | 1960 | plt.figure() |
|
|
1961 | 1961 | plt.plot(VelRange,spc,marker='*',linestyle='') |
|
|
1962 | 1962 | plt.plot(VelRange,fun) |
|
|
1963 | 1963 | plt.title(dataOut.heightList[hei]) |
|
|
1964 | 1964 | plt.show() |
|
|
1965 | 1965 | ''' |
|
|
1966 | 1966 | |
|
|
1967 | 1967 | return dataOut |
|
|
1968 | 1968 | |
|
|
1969 | 1969 | class PrecipitationProc(Operation): |
|
|
1970 | 1970 | |
|
|
1971 | 1971 | ''' |
|
|
1972 | 1972 | Operator that estimates Reflectivity factor (Z), and estimates rainfall Rate (R) |
|
|
1973 | 1973 | |
|
|
1974 | 1974 | Input: |
|
|
1975 | 1975 | self.dataOut.data_pre : SelfSpectra |
|
|
1976 | 1976 | |
|
|
1977 | 1977 | Output: |
|
|
1978 | 1978 | |
|
|
1979 | 1979 | self.dataOut.data_output : Reflectivity factor, rainfall Rate |
|
|
1980 | 1980 | |
|
|
1981 | 1981 | |
|
|
1982 | 1982 | Parameters affected: |
|
|
1983 | 1983 | ''' |
|
|
1984 | 1984 | |
|
|
1985 | 1985 | def __init__(self): |
|
|
1986 | 1986 | Operation.__init__(self) |
|
|
1987 | 1987 | self.i=0 |
|
|
1988 | 1988 | |
|
|
1989 | 1989 | def run(self, dataOut, radar=None, Pt=5000, Gt=295.1209, Gr=70.7945, Lambda=0.6741, aL=2.5118, |
|
|
1990 | 1990 | tauW=4e-06, ThetaT=0.1656317, ThetaR=0.36774087, Km2 = 0.93, Altitude=3350, SNRdBlimit=-30, |
|
|
1991 | 1991 | channel=None): |
|
|
1992 | 1992 | |
|
|
1993 | 1993 | # print ('Entering PrecepitationProc ... ') |
|
|
1994 | 1994 | |
|
|
1995 | 1995 | if radar == "MIRA35C" : |
|
|
1996 | 1996 | |
|
|
1997 | 1997 | self.spc = dataOut.data_pre[0].copy() |
|
|
1998 | 1998 | self.Num_Hei = self.spc.shape[2] |
|
|
1999 | 1999 | self.Num_Bin = self.spc.shape[1] |
|
|
2000 | 2000 | self.Num_Chn = self.spc.shape[0] |
|
|
2001 | 2001 | Ze = self.dBZeMODE2(dataOut) |
|
|
2002 | 2002 | |
|
|
2003 | 2003 | else: |
|
|
2004 | 2004 | |
|
|
2005 | 2005 | self.spc = dataOut.data_pre[0].copy() |
|
|
2006 | 2006 | |
|
|
2007 | 2007 | #NOTA SE DEBE REMOVER EL RANGO DEL PULSO TX |
|
|
2008 | 2008 | self.spc[:,:,0:7]= numpy.NaN |
|
|
2009 | 2009 | |
|
|
2010 | 2010 | self.Num_Hei = self.spc.shape[2] |
|
|
2011 | 2011 | self.Num_Bin = self.spc.shape[1] |
|
|
2012 | 2012 | self.Num_Chn = self.spc.shape[0] |
|
|
2013 | 2013 | |
|
|
2014 | 2014 | VelRange = dataOut.spc_range[2] |
|
|
2015 | 2015 | |
|
|
2016 | 2016 | ''' Se obtiene la constante del RADAR ''' |
|
|
2017 | 2017 | |
|
|
2018 | 2018 | self.Pt = Pt |
|
|
2019 | 2019 | self.Gt = Gt |
|
|
2020 | 2020 | self.Gr = Gr |
|
|
2021 | 2021 | self.Lambda = Lambda |
|
|
2022 | 2022 | self.aL = aL |
|
|
2023 | 2023 | self.tauW = tauW |
|
|
2024 | 2024 | self.ThetaT = ThetaT |
|
|
2025 | 2025 | self.ThetaR = ThetaR |
|
|
2026 | 2026 | self.GSys = 10**(36.63/10) # Ganancia de los LNA 36.63 dB |
|
|
2027 | 2027 | self.lt = 10**(1.67/10) # Perdida en cables Tx 1.67 dB |
|
|
2028 | 2028 | self.lr = 10**(5.73/10) # Perdida en cables Rx 5.73 dB |
|
|
2029 | 2029 | |
|
|
2030 | 2030 | Numerator = ( (4*numpy.pi)**3 * aL**2 * 16 * numpy.log(2) ) |
|
|
2031 | 2031 | Denominator = ( Pt * Gt * Gr * Lambda**2 * SPEED_OF_LIGHT * tauW * numpy.pi * ThetaT * ThetaR) |
|
|
2032 | 2032 | RadarConstant = 10e-26 * Numerator / Denominator # |
|
|
2033 | 2033 | ExpConstant = 10**(40/10) #Constante Experimental |
|
|
2034 | 2034 | |
|
|
2035 | 2035 | SignalPower = numpy.zeros([self.Num_Chn,self.Num_Bin,self.Num_Hei]) |
|
|
2036 | 2036 | for i in range(self.Num_Chn): |
|
|
2037 | 2037 | SignalPower[i,:,:] = self.spc[i,:,:] - dataOut.noise[i] |
|
|
2038 | 2038 | SignalPower[numpy.where(SignalPower < 0)] = 1e-20 |
|
|
2039 | 2039 | |
|
|
2040 | 2040 | if channel is None: |
|
|
2041 | 2041 | SPCmean = numpy.mean(SignalPower, 0) |
|
|
2042 | 2042 | else: |
|
|
2043 | 2043 | SPCmean = SignalPower[channel] |
|
|
2044 | 2044 | Pr = SPCmean[:,:]/dataOut.normFactor |
|
|
2045 | 2045 | |
|
|
2046 | 2046 | # Declaring auxiliary variables |
|
|
2047 | 2047 | Range = dataOut.heightList*1000. #Range in m |
|
|
2048 | 2048 | # replicate the heightlist to obtain a matrix [Num_Bin,Num_Hei] |
|
|
2049 | 2049 | rMtrx = numpy.transpose(numpy.transpose([dataOut.heightList*1000.] * self.Num_Bin)) |
|
|
2050 | 2050 | zMtrx = rMtrx+Altitude |
|
|
2051 | 2051 | # replicate the VelRange to obtain a matrix [Num_Bin,Num_Hei] |
|
|
2052 | 2052 | VelMtrx = numpy.transpose(numpy.tile(VelRange[:-1], (self.Num_Hei,1))) |
|
|
2053 | 2053 | |
|
|
2054 | 2054 | # height dependence to air density Foote and Du Toit (1969) |
|
|
2055 | 2055 | delv_z = 1 + 3.68e-5 * zMtrx + 1.71e-9 * zMtrx**2 |
|
|
2056 | 2056 | VMtrx = VelMtrx / delv_z #Normalized velocity |
|
|
2057 | 2057 | VMtrx[numpy.where(VMtrx> 9.6)] = numpy.NaN |
|
|
2058 | 2058 | # Diameter is related to the fall speed of falling drops |
|
|
2059 | 2059 | D_Vz = -1.667 * numpy.log( 0.9369 - 0.097087 * VMtrx ) # D in [mm] |
|
|
2060 | 2060 | # Only valid for D>= 0.16 mm |
|
|
2061 | 2061 | D_Vz[numpy.where(D_Vz < 0.16)] = numpy.NaN |
|
|
2062 | 2062 | |
|
|
2063 | 2063 | #Calculate Radar Reflectivity ETAn |
|
|
2064 | 2064 | ETAn = (RadarConstant *ExpConstant) * Pr * rMtrx**2 #Reflectivity (ETA) |
|
|
2065 | 2065 | ETAd = ETAn * 6.18 * exp( -0.6 * D_Vz ) * delv_z |
|
|
2066 | 2066 | # Radar Cross Section |
|
|
2067 | 2067 | sigmaD = Km2 * (D_Vz * 1e-3 )**6 * numpy.pi**5 / Lambda**4 |
|
|
2068 | 2068 | # Drop Size Distribution |
|
|
2069 | 2069 | DSD = ETAn / sigmaD |
|
|
2070 | 2070 | # Equivalente Reflectivy |
|
|
2071 | 2071 | Ze_eqn = numpy.nansum( DSD * D_Vz**6 ,axis=0) |
|
|
2072 | 2072 | Ze_org = numpy.nansum(ETAn * Lambda**4, axis=0) / (1e-18*numpy.pi**5 * Km2) # [mm^6 /m^3] |
|
|
2073 | 2073 | # RainFall Rate |
|
|
2074 | 2074 | RR = 0.0006*numpy.pi * numpy.nansum( D_Vz**3 * DSD * VelMtrx ,0) #mm/hr |
|
|
2075 | 2075 | |
|
|
2076 | 2076 | # Censoring the data |
|
|
2077 | 2077 | # Removing data with SNRth < 0dB se debe considerar el SNR por canal |
|
|
2078 | 2078 | SNRth = 10**(SNRdBlimit/10) #-30dB |
|
|
2079 | 2079 | novalid = numpy.where((dataOut.data_snr[0,:] <SNRth) | (dataOut.data_snr[1,:] <SNRth) | (dataOut.data_snr[2,:] <SNRth)) # AND condition. Maybe OR condition better |
|
|
2080 | 2080 | W = numpy.nanmean(dataOut.data_dop,0) |
|
|
2081 | 2081 | W[novalid] = numpy.NaN |
|
|
2082 | 2082 | Ze_org[novalid] = numpy.NaN |
|
|
2083 | 2083 | RR[novalid] = numpy.NaN |
|
|
2084 | 2084 | |
|
|
2085 | 2085 | dataOut.data_output = RR[8] |
|
|
2086 | 2086 | dataOut.data_param = numpy.ones([3,self.Num_Hei]) |
|
|
2087 | 2087 | dataOut.channelList = [0,1,2] |
|
|
2088 | 2088 | |
|
|
2089 | 2089 | dataOut.data_param[0]=10*numpy.log10(Ze_org) |
|
|
2090 | 2090 | dataOut.data_param[1]=-W |
|
|
2091 | 2091 | dataOut.data_param[2]=RR |
|
|
2092 | 2092 | |
|
|
2093 | 2093 | # print ('Leaving PrecepitationProc ... ') |
|
|
2094 | 2094 | return dataOut |
|
|
2095 | 2095 | |
|
|
2096 | 2096 | def dBZeMODE2(self, dataOut): # Processing for MIRA35C |
|
|
2097 | 2097 | |
|
|
2098 | 2098 | NPW = dataOut.NPW |
|
|
2099 | 2099 | COFA = dataOut.COFA |
|
|
2100 | 2100 | |
|
|
2101 | 2101 | SNR = numpy.array([self.spc[0,:,:] / NPW[0]]) #, self.spc[1,:,:] / NPW[1]]) |
|
|
2102 | 2102 | RadarConst = dataOut.RadarConst |
|
|
2103 | 2103 | #frequency = 34.85*10**9 |
|
|
2104 | 2104 | |
|
|
2105 | 2105 | ETA = numpy.zeros(([self.Num_Chn ,self.Num_Hei])) |
|
|
2106 | 2106 | data_output = numpy.ones([self.Num_Chn , self.Num_Hei])*numpy.NaN |
|
|
2107 | 2107 | |
|
|
2108 | 2108 | ETA = numpy.sum(SNR,1) |
|
|
2109 | 2109 | |
|
|
2110 | 2110 | ETA = numpy.where(ETA != 0. , ETA, numpy.NaN) |
|
|
2111 | 2111 | |
|
|
2112 | 2112 | Ze = numpy.ones([self.Num_Chn, self.Num_Hei] ) |
|
|
2113 | 2113 | |
|
|
2114 | 2114 | for r in range(self.Num_Hei): |
|
|
2115 | 2115 | |
|
|
2116 | 2116 | Ze[0,r] = ( ETA[0,r] ) * COFA[0,r][0] * RadarConst * ((r/5000.)**2) |
|
|
2117 | 2117 | #Ze[1,r] = ( ETA[1,r] ) * COFA[1,r][0] * RadarConst * ((r/5000.)**2) |
|
|
2118 | 2118 | |
|
|
2119 | 2119 | return Ze |
|
|
2120 | 2120 | |
|
|
2121 | 2121 | # def GetRadarConstant(self): |
|
|
2122 | 2122 | # |
|
|
2123 | 2123 | # """ |
|
|
2124 | 2124 | # Constants: |
|
|
2125 | 2125 | # |
|
|
2126 | 2126 | # Pt: Transmission Power dB 5kW 5000 |
|
|
2127 | 2127 | # Gt: Transmission Gain dB 24.7 dB 295.1209 |
|
|
2128 | 2128 | # Gr: Reception Gain dB 18.5 dB 70.7945 |
|
|
2129 | 2129 | # Lambda: Wavelenght m 0.6741 m 0.6741 |
|
|
2130 | 2130 | # aL: Attenuation loses dB 4dB 2.5118 |
|
|
2131 | 2131 | # tauW: Width of transmission pulse s 4us 4e-6 |
|
|
2132 | 2132 | # ThetaT: Transmission antenna bean angle rad 0.1656317 rad 0.1656317 |
|
|
2133 | 2133 | # ThetaR: Reception antenna beam angle rad 0.36774087 rad 0.36774087 |
|
|
2134 | 2134 | # |
|
|
2135 | 2135 | # """ |
|
|
2136 | 2136 | # |
|
|
2137 | 2137 | # Numerator = ( (4*numpy.pi)**3 * aL**2 * 16 * numpy.log(2) ) |
|
|
2138 | 2138 | # Denominator = ( Pt * Gt * Gr * Lambda**2 * SPEED_OF_LIGHT * TauW * numpy.pi * ThetaT * TheraR) |
|
|
2139 | 2139 | # RadarConstant = Numerator / Denominator |
|
|
2140 | 2140 | # |
|
|
2141 | 2141 | # return RadarConstant |
|
|
2142 | 2142 | |
|
|
2143 | 2143 | |
|
|
2144 | 2144 | class FullSpectralAnalysis(Operation): |
|
|
2145 | 2145 | |
|
|
2146 | 2146 | """ |
|
|
2147 | 2147 | Function that implements Full Spectral Analysis technique. |
|
|
2148 | 2148 | |
|
|
2149 | 2149 | Input: |
|
|
2150 | 2150 | self.dataOut.data_pre : SelfSpectra and CrossSpectra data |
|
|
2151 | 2151 | self.dataOut.groupList : Pairlist of channels |
|
|
2152 | 2152 | self.dataOut.ChanDist : Physical distance between receivers |
|
|
2153 | 2153 | |
|
|
2154 | 2154 | |
|
|
2155 | 2155 | Output: |
|
|
2156 | 2156 | |
|
|
2157 | 2157 | self.dataOut.data_output : Zonal wind, Meridional wind, and Vertical wind |
|
|
2158 | 2158 | |
|
|
2159 | 2159 | |
|
|
2160 | 2160 | Parameters affected: Winds, height range, SNR |
|
|
2161 | 2161 | |
|
|
2162 | 2162 | """ |
|
|
2163 | 2163 | def run(self, dataOut, Xi01=None, Xi02=None, Xi12=None, Eta01=None, Eta02=None, Eta12=None, SNRdBlimit=-30, |
|
|
2164 | 2164 | minheight=None, maxheight=None, NegativeLimit=None, PositiveLimit=None): |
|
|
2165 | 2165 | |
|
|
2166 | 2166 | spc = dataOut.data_pre[0].copy() |
|
|
2167 | 2167 | cspc = dataOut.data_pre[1] |
|
|
2168 | 2168 | nHeights = spc.shape[2] |
|
|
2169 | 2169 | |
|
|
2170 | 2170 | # first_height = 0.75 #km (ref: data header 20170822) |
|
|
2171 | 2171 | # resolution_height = 0.075 #km |
|
|
2172 | 2172 | ''' |
|
|
2173 | 2173 | finding height range. check this when radar parameters are changed! |
|
|
2174 | 2174 | ''' |
|
|
2175 | 2175 | if maxheight is not None: |
|
|
2176 | 2176 | # range_max = math.ceil((maxheight - first_height) / resolution_height) # theoretical |
|
|
2177 | 2177 | range_max = math.ceil(13.26 * maxheight - 3) # empirical, works better |
|
|
2178 | 2178 | else: |
|
|
2179 | 2179 | range_max = nHeights |
|
|
2180 | 2180 | if minheight is not None: |
|
|
2181 | 2181 | # range_min = int((minheight - first_height) / resolution_height) # theoretical |
|
|
2182 | 2182 | range_min = int(13.26 * minheight - 5) # empirical, works better |
|
|
2183 | 2183 | if range_min < 0: |
|
|
2184 | 2184 | range_min = 0 |
|
|
2185 | 2185 | else: |
|
|
2186 | 2186 | range_min = 0 |
|
|
2187 | 2187 | |
|
|
2188 | 2188 | pairsList = dataOut.groupList |
|
|
2189 | 2189 | if dataOut.ChanDist is not None : |
|
|
2190 | 2190 | ChanDist = dataOut.ChanDist |
|
|
2191 | 2191 | else: |
|
|
2192 | 2192 | ChanDist = numpy.array([[Xi01, Eta01],[Xi02,Eta02],[Xi12,Eta12]]) |
|
|
2193 | 2193 | |
|
|
2194 | 2194 | # 4 variables: zonal, meridional, vertical, and average SNR |
|
|
2195 | 2195 | data_param = numpy.zeros([4,nHeights]) * numpy.NaN |
|
|
2196 | 2196 | velocityX = numpy.zeros([nHeights]) * numpy.NaN |
|
|
2197 | 2197 | velocityY = numpy.zeros([nHeights]) * numpy.NaN |
|
|
2198 | 2198 | velocityZ = numpy.zeros([nHeights]) * numpy.NaN |
|
|
2199 | 2199 | |
|
|
2200 | 2200 | dbSNR = 10*numpy.log10(numpy.average(dataOut.data_snr,0)) |
|
|
2201 | 2201 | |
|
|
2202 | 2202 | '''***********************************************WIND ESTIMATION**************************************''' |
|
|
2203 | 2203 | for Height in range(nHeights): |
|
|
2204 | 2204 | |
|
|
2205 | 2205 | if Height >= range_min and Height < range_max: |
|
|
2206 | 2206 | # error_code will be useful in future analysis |
|
|
2207 | 2207 | [Vzon,Vmer,Vver, error_code] = self.WindEstimation(spc[:,:,Height], cspc[:,:,Height], pairsList, |
|
|
2208 | 2208 | ChanDist, Height, dataOut.noise, dataOut.spc_range, dbSNR[Height], SNRdBlimit, NegativeLimit, PositiveLimit,dataOut.frequency) |
|
|
2209 | 2209 | |
|
|
2210 | 2210 | if abs(Vzon) < 100. and abs(Vmer) < 100.: |
|
|
2211 | 2211 | velocityX[Height] = Vzon |
|
|
2212 | 2212 | velocityY[Height] = -Vmer |
|
|
2213 | 2213 | velocityZ[Height] = Vver |
|
|
2214 | 2214 | |
|
|
2215 | 2215 | # Censoring data with SNR threshold |
|
|
2216 | 2216 | dbSNR [dbSNR < SNRdBlimit] = numpy.NaN |
|
|
2217 | 2217 | |
|
|
2218 | 2218 | data_param[0] = velocityX |
|
|
2219 | 2219 | data_param[1] = velocityY |
|
|
2220 | 2220 | data_param[2] = velocityZ |
|
|
2221 | 2221 | data_param[3] = dbSNR |
|
|
2222 | 2222 | dataOut.data_param = data_param |
|
|
2223 | 2223 | return dataOut |
|
|
2224 | 2224 | |
|
|
2225 | 2225 | def moving_average(self,x, N=2): |
|
|
2226 | 2226 | """ convolution for smoothenig data. note that last N-1 values are convolution with zeroes """ |
|
|
2227 | 2227 | return numpy.convolve(x, numpy.ones((N,))/N)[(N-1):] |
|
|
2228 | 2228 | |
|
|
2229 | 2229 | def gaus(self,xSamples,Amp,Mu,Sigma): |
|
|
2230 | 2230 | return Amp * numpy.exp(-0.5*((xSamples - Mu)/Sigma)**2) |
|
|
2231 | 2231 | |
|
|
2232 | 2232 | def Moments(self, ySamples, xSamples): |
|
|
2233 | 2233 | Power = numpy.nanmean(ySamples) # Power, 0th Moment |
|
|
2234 | 2234 | yNorm = ySamples / numpy.nansum(ySamples) |
|
|
2235 | 2235 | RadVel = numpy.nansum(xSamples * yNorm) # Radial Velocity, 1st Moment |
|
|
2236 | 2236 | Sigma2 = numpy.nansum(yNorm * (xSamples - RadVel)**2) # Spectral Width, 2nd Moment |
|
|
2237 | 2237 | StdDev = numpy.sqrt(numpy.abs(Sigma2)) # Desv. Estandar, Ancho espectral |
|
|
2238 | 2238 | return numpy.array([Power,RadVel,StdDev]) |
|
|
2239 | 2239 | |
|
|
2240 | 2240 | def StopWindEstimation(self, error_code): |
|
|
2241 | 2241 | Vzon = numpy.NaN |
|
|
2242 | 2242 | Vmer = numpy.NaN |
|
|
2243 | 2243 | Vver = numpy.NaN |
|
|
2244 | 2244 | return Vzon, Vmer, Vver, error_code |
|
|
2245 | 2245 | |
|
|
2246 | 2246 | def AntiAliasing(self, interval, maxstep): |
|
|
2247 | 2247 | """ |
|
|
2248 | 2248 | function to prevent errors from aliased values when computing phaseslope |
|
|
2249 | 2249 | """ |
|
|
2250 | 2250 | antialiased = numpy.zeros(len(interval)) |
|
|
2251 | 2251 | copyinterval = interval.copy() |
|
|
2252 | 2252 | |
|
|
2253 | 2253 | antialiased[0] = copyinterval[0] |
|
|
2254 | 2254 | |
|
|
2255 | 2255 | for i in range(1,len(antialiased)): |
|
|
2256 | 2256 | step = interval[i] - interval[i-1] |
|
|
2257 | 2257 | if step > maxstep: |
|
|
2258 | 2258 | copyinterval -= 2*numpy.pi |
|
|
2259 | 2259 | antialiased[i] = copyinterval[i] |
|
|
2260 | 2260 | elif step < maxstep*(-1): |
|
|
2261 | 2261 | copyinterval += 2*numpy.pi |
|
|
2262 | 2262 | antialiased[i] = copyinterval[i] |
|
|
2263 | 2263 | else: |
|
|
2264 | 2264 | antialiased[i] = copyinterval[i].copy() |
|
|
2265 | 2265 | |
|
|
2266 | 2266 | return antialiased |
|
|
2267 | 2267 | |
|
|
2268 | 2268 | def WindEstimation(self, spc, cspc, pairsList, ChanDist, Height, noise, AbbsisaRange, dbSNR, SNRlimit, NegativeLimit, PositiveLimit, radfreq): |
|
|
2269 | 2269 | """ |
|
|
2270 | 2270 | Function that Calculates Zonal, Meridional and Vertical wind velocities. |
|
|
2271 | 2271 | Initial Version by E. Bocanegra updated by J. Zibell until Nov. 2019. |
|
|
2272 | 2272 | |
|
|
2273 | 2273 | Input: |
|
|
2274 | 2274 | spc, cspc : self spectra and cross spectra data. In Briggs notation something like S_i*(S_i)_conj, (S_j)_conj respectively. |
|
|
2275 | 2275 | pairsList : Pairlist of channels |
|
|
2276 | 2276 | ChanDist : array of xi_ij and eta_ij |
|
|
2277 | 2277 | Height : height at which data is processed |
|
|
2278 | 2278 | noise : noise in [channels] format for specific height |
|
|
2279 | 2279 | Abbsisarange : range of the frequencies or velocities |
|
|
2280 | 2280 | dbSNR, SNRlimit : signal to noise ratio in db, lower limit |
|
|
2281 | 2281 | |
|
|
2282 | 2282 | Output: |
|
|
2283 | 2283 | Vzon, Vmer, Vver : wind velocities |
|
|
2284 | 2284 | error_code : int that states where code is terminated |
|
|
2285 | 2285 | |
|
|
2286 | 2286 | 0 : no error detected |
|
|
2287 | 2287 | 1 : Gaussian of mean spc exceeds widthlimit |
|
|
2288 | 2288 | 2 : no Gaussian of mean spc found |
|
|
2289 | 2289 | 3 : SNR to low or velocity to high -> prec. e.g. |
|
|
2290 | 2290 | 4 : at least one Gaussian of cspc exceeds widthlimit |
|
|
2291 | 2291 | 5 : zero out of three cspc Gaussian fits converged |
|
|
2292 | 2292 | 6 : phase slope fit could not be found |
|
|
2293 | 2293 | 7 : arrays used to fit phase have different length |
|
|
2294 | 2294 | 8 : frequency range is either too short (len <= 5) or very long (> 30% of cspc) |
|
|
2295 | 2295 | |
|
|
2296 | 2296 | """ |
|
|
2297 | 2297 | |
|
|
2298 | 2298 | error_code = 0 |
|
|
2299 | 2299 | |
|
|
2300 | 2300 | nChan = spc.shape[0] |
|
|
2301 | 2301 | nProf = spc.shape[1] |
|
|
2302 | 2302 | nPair = cspc.shape[0] |
|
|
2303 | 2303 | |
|
|
2304 | 2304 | SPC_Samples = numpy.zeros([nChan, nProf]) # for normalized spc values for one height |
|
|
2305 | 2305 | CSPC_Samples = numpy.zeros([nPair, nProf], dtype=numpy.complex_) # for normalized cspc values |
|
|
2306 | 2306 | phase = numpy.zeros([nPair, nProf]) # phase between channels |
|
|
2307 | 2307 | PhaseSlope = numpy.zeros(nPair) # slope of the phases, channelwise |
|
|
2308 | 2308 | PhaseInter = numpy.zeros(nPair) # intercept to the slope of the phases, channelwise |
|
|
2309 | 2309 | xFrec = AbbsisaRange[0][:-1] # frequency range |
|
|
2310 | 2310 | xVel = AbbsisaRange[2][:-1] # velocity range |
|
|
2311 | 2311 | xSamples = xFrec # the frequency range is taken |
|
|
2312 | 2312 | delta_x = xSamples[1] - xSamples[0] # delta_f or delta_x |
|
|
2313 | 2313 | |
|
|
2314 | 2314 | # only consider velocities with in NegativeLimit and PositiveLimit |
|
|
2315 | 2315 | if (NegativeLimit is None): |
|
|
2316 | 2316 | NegativeLimit = numpy.min(xVel) |
|
|
2317 | 2317 | if (PositiveLimit is None): |
|
|
2318 | 2318 | PositiveLimit = numpy.max(xVel) |
|
|
2319 | 2319 | xvalid = numpy.where((xVel > NegativeLimit) & (xVel < PositiveLimit)) |
|
|
2320 | 2320 | xSamples_zoom = xSamples[xvalid] |
|
|
2321 | 2321 | |
|
|
2322 | 2322 | '''Getting Eij and Nij''' |
|
|
2323 | 2323 | Xi01, Xi02, Xi12 = ChanDist[:,0] |
|
|
2324 | 2324 | Eta01, Eta02, Eta12 = ChanDist[:,1] |
|
|
2325 | 2325 | |
|
|
2326 | 2326 | # spwd limit - updated by D. ScipiΓ³n 30.03.2021 |
|
|
2327 | 2327 | widthlimit = 10 |
|
|
2328 | 2328 | '''************************* SPC is normalized ********************************''' |
|
|
2329 | 2329 | spc_norm = spc.copy() |
|
|
2330 | 2330 | # For each channel |
|
|
2331 | 2331 | for i in range(nChan): |
|
|
2332 | 2332 | spc_sub = spc_norm[i,:] - noise[i] # only the signal power |
|
|
2333 | 2333 | SPC_Samples[i] = spc_sub / (numpy.nansum(spc_sub) * delta_x) |
|
|
2334 | 2334 | |
|
|
2335 | 2335 | '''********************** FITTING MEAN SPC GAUSSIAN **********************''' |
|
|
2336 | 2336 | |
|
|
2337 | 2337 | """ the gaussian of the mean: first subtract noise, then normalize. this is legal because |
|
|
2338 | 2338 | you only fit the curve and don't need the absolute value of height for calculation, |
|
|
2339 | 2339 | only for estimation of width. for normalization of cross spectra, you need initial, |
|
|
2340 | 2340 | unnormalized self-spectra With noise. |
|
|
2341 | 2341 | |
|
|
2342 | 2342 | Technically, you don't even need to normalize the self-spectra, as you only need the |
|
|
2343 | 2343 | width of the peak. However, it was left this way. Note that the normalization has a flaw: |
|
|
2344 | 2344 | due to subtraction of the noise, some values are below zero. Raw "spc" values should be |
|
|
2345 | 2345 | >= 0, as it is the modulus squared of the signals (complex * it's conjugate) |
|
|
2346 | 2346 | """ |
|
|
2347 | 2347 | # initial conditions |
|
|
2348 | 2348 | popt = [1e-10,0,1e-10] |
|
|
2349 | 2349 | # Spectra average |
|
|
2350 | 2350 | SPCMean = numpy.average(SPC_Samples,0) |
|
|
2351 | 2351 | # Moments in frequency |
|
|
2352 | 2352 | SPCMoments = self.Moments(SPCMean[xvalid], xSamples_zoom) |
|
|
2353 | 2353 | |
|
|
2354 | 2354 | # Gauss Fit SPC in frequency domain |
|
|
2355 | 2355 | if dbSNR > SNRlimit: # only if SNR > SNRth |
|
|
2356 | 2356 | try: |
|
|
2357 | 2357 | popt,pcov = curve_fit(self.gaus,xSamples_zoom,SPCMean[xvalid],p0=SPCMoments) |
|
|
2358 | 2358 | if popt[2] <= 0 or popt[2] > widthlimit: # CONDITION |
|
|
2359 | 2359 | return self.StopWindEstimation(error_code = 1) |
|
|
2360 | 2360 | FitGauss = self.gaus(xSamples_zoom,*popt) |
|
|
2361 | 2361 | except :#RuntimeError: |
|
|
2362 | 2362 | return self.StopWindEstimation(error_code = 2) |
|
|
2363 | 2363 | else: |
|
|
2364 | 2364 | return self.StopWindEstimation(error_code = 3) |
|
|
2365 | 2365 | |
|
|
2366 | 2366 | '''***************************** CSPC Normalization ************************* |
|
|
2367 | 2367 | The Spc spectra are used to normalize the crossspectra. Peaks from precipitation |
|
|
2368 | 2368 | influence the norm which is not desired. First, a range is identified where the |
|
|
2369 | 2369 | wind peak is estimated -> sum_wind is sum of those frequencies. Next, the area |
|
|
2370 | 2370 | around it gets cut off and values replaced by mean determined by the boundary |
|
|
2371 | 2371 | data -> sum_noise (spc is not normalized here, thats why the noise is important) |
|
|
2372 | 2372 | |
|
|
2373 | 2373 | The sums are then added and multiplied by range/datapoints, because you need |
|
|
2374 | 2374 | an integral and not a sum for normalization. |
|
|
2375 | 2375 | |
|
|
2376 | 2376 | A norm is found according to Briggs 92. |
|
|
2377 | 2377 | ''' |
|
|
2378 | 2378 | # for each pair |
|
|
2379 | 2379 | for i in range(nPair): |
|
|
2380 | 2380 | cspc_norm = cspc[i,:].copy() |
|
|
2381 | 2381 | chan_index0 = pairsList[i][0] |
|
|
2382 | 2382 | chan_index1 = pairsList[i][1] |
|
|
2383 | 2383 | CSPC_Samples[i] = cspc_norm / (numpy.sqrt(numpy.nansum(spc_norm[chan_index0])*numpy.nansum(spc_norm[chan_index1])) * delta_x) |
|
|
2384 | 2384 | phase[i] = numpy.arctan2(CSPC_Samples[i].imag, CSPC_Samples[i].real) |
|
|
2385 | 2385 | |
|
|
2386 | 2386 | CSPCmoments = numpy.vstack([self.Moments(numpy.abs(CSPC_Samples[0,xvalid]), xSamples_zoom), |
|
|
2387 | 2387 | self.Moments(numpy.abs(CSPC_Samples[1,xvalid]), xSamples_zoom), |
|
|
2388 | 2388 | self.Moments(numpy.abs(CSPC_Samples[2,xvalid]), xSamples_zoom)]) |
|
|
2389 | 2389 | |
|
|
2390 | 2390 | popt01, popt02, popt12 = [1e-10,0,1e-10], [1e-10,0,1e-10] ,[1e-10,0,1e-10] |
|
|
2391 | 2391 | FitGauss01, FitGauss02, FitGauss12 = numpy.zeros(len(xSamples)), numpy.zeros(len(xSamples)), numpy.zeros(len(xSamples)) |
|
|
2392 | 2392 | |
|
|
2393 | 2393 | '''*******************************FIT GAUSS CSPC************************************''' |
|
|
2394 | 2394 | try: |
|
|
2395 | 2395 | popt01,pcov = curve_fit(self.gaus,xSamples_zoom,numpy.abs(CSPC_Samples[0][xvalid]),p0=CSPCmoments[0]) |
|
|
2396 | 2396 | if popt01[2] > widthlimit: # CONDITION |
|
|
2397 | 2397 | return self.StopWindEstimation(error_code = 4) |
|
|
2398 | 2398 | popt02,pcov = curve_fit(self.gaus,xSamples_zoom,numpy.abs(CSPC_Samples[1][xvalid]),p0=CSPCmoments[1]) |
|
|
2399 | 2399 | if popt02[2] > widthlimit: # CONDITION |
|
|
2400 | 2400 | return self.StopWindEstimation(error_code = 4) |
|
|
2401 | 2401 | popt12,pcov = curve_fit(self.gaus,xSamples_zoom,numpy.abs(CSPC_Samples[2][xvalid]),p0=CSPCmoments[2]) |
|
|
2402 | 2402 | if popt12[2] > widthlimit: # CONDITION |
|
|
2403 | 2403 | return self.StopWindEstimation(error_code = 4) |
|
|
2404 | 2404 | |
|
|
2405 | 2405 | FitGauss01 = self.gaus(xSamples_zoom, *popt01) |
|
|
2406 | 2406 | FitGauss02 = self.gaus(xSamples_zoom, *popt02) |
|
|
2407 | 2407 | FitGauss12 = self.gaus(xSamples_zoom, *popt12) |
|
|
2408 | 2408 | except: |
|
|
2409 | 2409 | return self.StopWindEstimation(error_code = 5) |
|
|
2410 | 2410 | |
|
|
2411 | 2411 | |
|
|
2412 | 2412 | '''************* Getting Fij ***************''' |
|
|
2413 | 2413 | # x-axis point of the gaussian where the center is located from GaussFit of spectra |
|
|
2414 | 2414 | GaussCenter = popt[1] |
|
|
2415 | 2415 | ClosestCenter = xSamples_zoom[numpy.abs(xSamples_zoom-GaussCenter).argmin()] |
|
|
2416 | 2416 | PointGauCenter = numpy.where(xSamples_zoom==ClosestCenter)[0][0] |
|
|
2417 | 2417 | |
|
|
2418 | 2418 | # Point where e^-1 is located in the gaussian |
|
|
2419 | 2419 | PeMinus1 = numpy.max(FitGauss) * numpy.exp(-1) |
|
|
2420 | 2420 | FijClosest = FitGauss[numpy.abs(FitGauss-PeMinus1).argmin()] # The closest point to"Peminus1" in "FitGauss" |
|
|
2421 | 2421 | PointFij = numpy.where(FitGauss==FijClosest)[0][0] |
|
|
2422 | 2422 | Fij = numpy.abs(xSamples_zoom[PointFij] - xSamples_zoom[PointGauCenter]) |
|
|
2423 | 2423 | |
|
|
2424 | 2424 | '''********** Taking frequency ranges from mean SPCs **********''' |
|
|
2425 | 2425 | GauWidth = popt[2] * 3/2 # Bandwidth of Gau01 |
|
|
2426 | 2426 | Range = numpy.empty(2) |
|
|
2427 | 2427 | Range[0] = GaussCenter - GauWidth |
|
|
2428 | 2428 | Range[1] = GaussCenter + GauWidth |
|
|
2429 | 2429 | # Point in x-axis where the bandwidth is located (min:max) |
|
|
2430 | 2430 | ClosRangeMin = xSamples_zoom[numpy.abs(xSamples_zoom-Range[0]).argmin()] |
|
|
2431 | 2431 | ClosRangeMax = xSamples_zoom[numpy.abs(xSamples_zoom-Range[1]).argmin()] |
|
|
2432 | 2432 | PointRangeMin = numpy.where(xSamples_zoom==ClosRangeMin)[0][0] |
|
|
2433 | 2433 | PointRangeMax = numpy.where(xSamples_zoom==ClosRangeMax)[0][0] |
|
|
2434 | 2434 | Range = numpy.array([ PointRangeMin, PointRangeMax ]) |
|
|
2435 | 2435 | FrecRange = xSamples_zoom[ Range[0] : Range[1] ] |
|
|
2436 | 2436 | |
|
|
2437 | 2437 | '''************************** Getting Phase Slope ***************************''' |
|
|
2438 | 2438 | for i in range(nPair): |
|
|
2439 | 2439 | if len(FrecRange) > 5: |
|
|
2440 | 2440 | PhaseRange = phase[i, xvalid[0][Range[0]:Range[1]]].copy() |
|
|
2441 | 2441 | mask = ~numpy.isnan(FrecRange) & ~numpy.isnan(PhaseRange) |
|
|
2442 | 2442 | if len(FrecRange) == len(PhaseRange): |
|
|
2443 | 2443 | try: |
|
|
2444 | 2444 | slope, intercept, _, _, _ = stats.linregress(FrecRange[mask], self.AntiAliasing(PhaseRange[mask], 4.5)) |
|
|
2445 | 2445 | PhaseSlope[i] = slope |
|
|
2446 | 2446 | PhaseInter[i] = intercept |
|
|
2447 | 2447 | except: |
|
|
2448 | 2448 | return self.StopWindEstimation(error_code = 6) |
|
|
2449 | 2449 | else: |
|
|
2450 | 2450 | return self.StopWindEstimation(error_code = 7) |
|
|
2451 | 2451 | else: |
|
|
2452 | 2452 | return self.StopWindEstimation(error_code = 8) |
|
|
2453 | 2453 | |
|
|
2454 | 2454 | '''*** Constants A-H correspond to the convention as in Briggs and Vincent 1992 ***''' |
|
|
2455 | 2455 | |
|
|
2456 | 2456 | '''Getting constant C''' |
|
|
2457 | 2457 | cC=(Fij*numpy.pi)**2 |
|
|
2458 | 2458 | |
|
|
2459 | 2459 | '''****** Getting constants F and G ******''' |
|
|
2460 | 2460 | MijEijNij = numpy.array([[Xi02,Eta02], [Xi12,Eta12]]) |
|
|
2461 | 2461 | # MijEijNij = numpy.array([[Xi01,Eta01], [Xi02,Eta02], [Xi12,Eta12]]) |
|
|
2462 | 2462 | # MijResult0 = (-PhaseSlope[0] * cC) / (2*numpy.pi) |
|
|
2463 | 2463 | MijResult1 = (-PhaseSlope[1] * cC) / (2*numpy.pi) |
|
|
2464 | 2464 | MijResult2 = (-PhaseSlope[2] * cC) / (2*numpy.pi) |
|
|
2465 | 2465 | # MijResults = numpy.array([MijResult0, MijResult1, MijResult2]) |
|
|
2466 | 2466 | MijResults = numpy.array([MijResult1, MijResult2]) |
|
|
2467 | 2467 | (cF,cG) = numpy.linalg.solve(MijEijNij, MijResults) |
|
|
2468 | 2468 | |
|
|
2469 | 2469 | '''****** Getting constants A, B and H ******''' |
|
|
2470 | 2470 | W01 = numpy.nanmax( FitGauss01 ) |
|
|
2471 | 2471 | W02 = numpy.nanmax( FitGauss02 ) |
|
|
2472 | 2472 | W12 = numpy.nanmax( FitGauss12 ) |
|
|
2473 | 2473 | |
|
|
2474 | 2474 | WijResult01 = ((cF * Xi01 + cG * Eta01)**2)/cC - numpy.log(W01 / numpy.sqrt(numpy.pi / cC)) |
|
|
2475 | 2475 | WijResult02 = ((cF * Xi02 + cG * Eta02)**2)/cC - numpy.log(W02 / numpy.sqrt(numpy.pi / cC)) |
|
|
2476 | 2476 | WijResult12 = ((cF * Xi12 + cG * Eta12)**2)/cC - numpy.log(W12 / numpy.sqrt(numpy.pi / cC)) |
|
|
2477 | 2477 | WijResults = numpy.array([WijResult01, WijResult02, WijResult12]) |
|
|
2478 | 2478 | |
|
|
2479 | 2479 | WijEijNij = numpy.array([ [Xi01**2, Eta01**2, 2*Xi01*Eta01] , [Xi02**2, Eta02**2, 2*Xi02*Eta02] , [Xi12**2, Eta12**2, 2*Xi12*Eta12] ]) |
|
|
2480 | 2480 | (cA,cB,cH) = numpy.linalg.solve(WijEijNij, WijResults) |
|
|
2481 | 2481 | |
|
|
2482 | 2482 | VxVy = numpy.array([[cA,cH],[cH,cB]]) |
|
|
2483 | 2483 | VxVyResults = numpy.array([-cF,-cG]) |
|
|
2484 | 2484 | (Vmer,Vzon) = numpy.linalg.solve(VxVy, VxVyResults) |
|
|
2485 | 2485 | Vver = -SPCMoments[1]*SPEED_OF_LIGHT/(2*radfreq) |
|
|
2486 | 2486 | error_code = 0 |
|
|
2487 | 2487 | |
|
|
2488 | 2488 | return Vzon, Vmer, Vver, error_code |
|
|
2489 | 2489 | |
|
|
2490 | 2490 | class SpectralMoments(Operation): |
|
|
2491 | 2491 | |
|
|
2492 | 2492 | ''' |
|
|
2493 | 2493 | Function SpectralMoments() |
|
|
2494 | 2494 | |
|
|
2495 | 2495 | Calculates moments (power, mean, standard deviation) and SNR of the signal |
|
|
2496 | 2496 | |
|
|
2497 | 2497 | Type of dataIn: Spectra |
|
|
2498 | 2498 | |
|
|
2499 | 2499 | Configuration Parameters: |
|
|
2500 | 2500 | |
|
|
2501 | 2501 | proc_type : (0) First spectral moments routine (Default), |
|
|
2502 | 2502 | (1) Spectral moment routine similar to JULIA. |
|
|
2503 | 2503 | mode_fit : (0) No gaussian fit |
|
|
2504 | 2504 | (1) One gaussian fit for 150Km processing. |
|
|
2505 | 2505 | |
|
|
2506 | 2506 | exp : '150EEJ' To select 128 points window |
|
|
2507 | 2507 | 'ESF_EW' To select full window. |
|
|
2508 | 2508 | |
|
|
2509 | 2509 | Input: |
|
|
2510 | 2510 | channelList : simple channel list to select e.g. [2,3,7] |
|
|
2511 | 2511 | self.dataOut.data_pre : Spectral data |
|
|
2512 | 2512 | self.dataOut.abscissaList : List of frequencies |
|
|
2513 | 2513 | self.dataOut.noise : Noise level per channel |
|
|
2514 | 2514 | |
|
|
2515 | 2515 | Affected: |
|
|
2516 | 2516 | self.dataOut.moments : Parameters per channel |
|
|
2517 | 2517 | self.dataOut.data_snr : SNR per channel |
|
|
2518 | 2518 | |
|
|
2519 | 2519 | ''' |
|
|
2520 | 2520 | |
|
|
2521 | 2521 | def __calculateMoments(self, oldspec, oldfreq, n0, |
|
|
2522 | 2522 | nicoh = None, graph = None, smooth = None, type1 = None, fwindow = None, \ |
|
|
2523 | 2523 | snrth = None, dc = None, aliasing = None, oldfd = None, wwauto = None, \ |
|
|
2524 | 2524 | vers= None, Hei= None, debug=False, dbg_hei=None, ymax=0.1, curr_ch=0, sel_ch=[0,1]): |
|
|
2525 | 2525 | |
|
|
2526 | 2526 | def __GAUSSWINFIT1(A, flagPDER=0): |
|
|
2527 | 2527 | nonlocal truex, xvalid |
|
|
2528 | 2528 | nparams = 4 |
|
|
2529 | 2529 | M=truex.size |
|
|
2530 | 2530 | mm=numpy.arange(M,dtype='f4') |
|
|
2531 | 2531 | delta = numpy.zeros(M,dtype='f4') |
|
|
2532 | 2532 | delta[0] = 1.0 |
|
|
2533 | 2533 | Ts = numpy.array([1.0/(2*truex[0])],dtype='f4')[0] |
|
|
2534 | 2534 | jj = -1j |
|
|
2535 | 2535 | #if self.winauto is None: self.winauto = (1.0 - mm/M) |
|
|
2536 | 2536 | winauto = (1.0 - mm/M) |
|
|
2537 | 2537 | winauto = winauto/winauto.max() # Normalized to 1 |
|
|
2538 | 2538 | #ON_ERROR,2 # IDL sentence: Return to caller if an error occurs |
|
|
2539 | 2539 | A[0] = numpy.abs(A[0]) |
|
|
2540 | 2540 | A[2] = numpy.abs(A[2]) |
|
|
2541 | 2541 | A[3] = numpy.abs(A[3]) |
|
|
2542 | 2542 | pi=numpy.array([numpy.pi],dtype='f4')[0] |
|
|
2543 | 2543 | if A[2] != 0: |
|
|
2544 | 2544 | Z = numpy.exp(-2*numpy.power((pi*A[2]*mm*Ts),2,dtype='f4')+jj*2*pi*A[1]*mm*Ts, dtype='c8') # Get Z |
|
|
2545 | 2545 | else: |
|
|
2546 | 2546 | Z = mm*0.0 |
|
|
2547 | 2547 | A[0] = 0.0 |
|
|
2548 | 2548 | junkF = numpy.roll(2*fft(winauto*(A[0]*Z+A[3]*delta)).real - \ |
|
|
2549 | 2549 | winauto[0]*(A[0]+A[3]), M//2) # *M scale for fft not needed in python |
|
|
2550 | 2550 | F = junkF[xvalid] |
|
|
2551 | 2551 | if flagPDER == 0: #NEED PARTIAL? |
|
|
2552 | 2552 | return F |
|
|
2553 | 2553 | PDER = numpy.zeros((M,nparams)) #YES, MAKE ARRAY. |
|
|
2554 | 2554 | PDER[:,0] = numpy.shift(2*(fft(winauto*Z)*M) - winauto[0], M/2) |
|
|
2555 | 2555 | PDER[:,1] = numpy.shift(2*(fft(winauto*jj*2*numpy.pi*mm*Ts*A[0]*Z)*M), M/2) |
|
|
2556 | 2556 | PDER[:,2] = numpy.shift(2*(fft(winauto*(-4*numpy.power(numpy.pi*mm*Ts,2)*A[2]*A[0]*Z))*M), M/2) |
|
|
2557 | 2557 | PDER[:,3] = numpy.shift(2*(fft(winauto*delta)*M) - winauto[0], M/2) |
|
|
2558 | 2558 | PDER = PDER[xvalid,:] |
|
|
2559 | 2559 | return F, PDER |
|
|
2560 | 2560 | |
|
|
2561 | 2561 | def __curvefit_koki(y, a, Weights, FlagNoDerivative=1, |
|
|
2562 | 2562 | itmax=20, tol=None): |
|
|
2563 | 2563 | #ON_ERROR,2 IDL SENTENCE: RETURN TO THE CALLER IF ERROR |
|
|
2564 | 2564 | if tol == None: |
|
|
2565 | 2565 | tol = numpy.array([1.e-3],dtype='f4')[0] |
|
|
2566 | 2566 | typ=a.dtype |
|
|
2567 | 2567 | double = 1 if typ == numpy.float64 else 0 |
|
|
2568 | 2568 | if typ != numpy.float32: |
|
|
2569 | 2569 | a=a.astype(numpy.float32) #Make params floating |
|
|
2570 | 2570 | # if we will be estimating partial derivates then compute machine precision |
|
|
2571 | 2571 | if FlagNoDerivative == 1: |
|
|
2572 | 2572 | res=numpy.MachAr(float_conv=numpy.float32) |
|
|
2573 | 2573 | eps=numpy.sqrt(res.eps) |
|
|
2574 | 2574 | |
|
|
2575 | 2575 | nterms = a.size # Number of parameters |
|
|
2576 | 2576 | nfree=numpy.array([numpy.size(y) - nterms],dtype='f4')[0] # Degrees of freedom |
|
|
2577 | 2577 | if nfree <= 0: print('Curvefit - not enough data points.') |
|
|
2578 | 2578 | flambda= numpy.array([0.001],dtype='f4')[0] # Initial lambda |
|
|
2579 | 2579 | #diag=numpy.arange(nterms)*(nterms+1) # Subscripta of diagonal elements |
|
|
2580 | 2580 | # Use diag method in python |
|
|
2581 | 2581 | converge=1 |
|
|
2582 | 2582 | |
|
|
2583 | 2583 | #Define the partial derivative array |
|
|
2584 | 2584 | PDER = numpy.zeros((nterms,numpy.size(y)),dtype='f8') if double == 1 else numpy.zeros((nterms,numpy.size(y)),dtype='f4') |
|
|
2585 | 2585 | |
|
|
2586 | 2586 | for Niter in range(itmax): #Iteration loop |
|
|
2587 | 2587 | |
|
|
2588 | 2588 | if FlagNoDerivative == 1: |
|
|
2589 | 2589 | #Evaluate function and estimate partial derivatives |
|
|
2590 | 2590 | yfit = __GAUSSWINFIT1(a) |
|
|
2591 | 2591 | for term in range(nterms): |
|
|
2592 | 2592 | p=a.copy() # Copy current parameters |
|
|
2593 | 2593 | #Increment size for forward difference derivative |
|
|
2594 | 2594 | inc = eps * abs(p[term]) |
|
|
2595 | 2595 | if inc == 0: inc = eps |
|
|
2596 | 2596 | p[term] = p[term] + inc |
|
|
2597 | 2597 | yfit1 = __GAUSSWINFIT1(p) |
|
|
2598 | 2598 | PDER[term,:] = (yfit1-yfit)/inc |
|
|
2599 | 2599 | else: |
|
|
2600 | 2600 | #The user's procedure will return partial derivatives |
|
|
2601 | 2601 | yfit,PDER=__GAUSSWINFIT1(a, flagPDER=1) |
|
|
2602 | 2602 | |
|
|
2603 | 2603 | beta = numpy.dot(PDER,(y-yfit)*Weights) |
|
|
2604 | 2604 | alpha = numpy.dot(PDER * numpy.tile(Weights,(nterms,1)), numpy.transpose(PDER)) |
|
|
2605 | 2605 | # save current values of return parameters |
|
|
2606 | 2606 | sigma1 = numpy.sqrt( 1.0 / numpy.diag(alpha) ) # Current sigma. |
|
|
2607 | 2607 | sigma = sigma1 |
|
|
2608 | 2608 | |
|
|
2609 | 2609 | chisq1 = numpy.sum(Weights*numpy.power(y-yfit,2,dtype='f4'),dtype='f4')/nfree # Current chi squared. |
|
|
2610 | 2610 | chisq = chisq1 |
|
|
2611 | 2611 | yfit1 = yfit |
|
|
2612 | 2612 | elev7=numpy.array([1.0e7],dtype='f4')[0] |
|
|
2613 | 2613 | compara =numpy.sum(abs(y))/elev7/nfree |
|
|
2614 | 2614 | done_early = chisq1 < compara |
|
|
2615 | 2615 | |
|
|
2616 | 2616 | if done_early: |
|
|
2617 | 2617 | chi2 = chisq # Return chi-squared (chi2 obsolete-still works) |
|
|
2618 | 2618 | if done_early: Niter -= 1 |
|
|
2619 | 2619 | #save_tp(chisq,Niter,yfit) |
|
|
2620 | 2620 | return yfit, a, converge, sigma, chisq # return result |
|
|
2621 | 2621 | #c = numpy.dot(c, c) # this operator implemented at the next lines |
|
|
2622 | 2622 | c_tmp = numpy.sqrt(numpy.diag(alpha)) |
|
|
2623 | 2623 | siz=len(c_tmp) |
|
|
2624 | 2624 | c=numpy.dot(c_tmp.reshape(siz,1),c_tmp.reshape(1,siz)) |
|
|
2625 | 2625 | lambdaCount = 0 |
|
|
2626 | 2626 | while True: |
|
|
2627 | 2627 | lambdaCount += 1 |
|
|
2628 | 2628 | # Normalize alpha to have unit diagonal. |
|
|
2629 | 2629 | array = alpha / c |
|
|
2630 | 2630 | # Augment the diagonal. |
|
|
2631 | 2631 | one=numpy.array([1.],dtype='f4')[0] |
|
|
2632 | 2632 | numpy.fill_diagonal(array,numpy.diag(array)*(one+flambda)) |
|
|
2633 | 2633 | # Invert modified curvature matrix to find new parameters. |
|
|
2634 | 2634 | |
|
|
2635 | 2635 | try: |
|
|
2636 | 2636 | array = (1.0/array) if array.size == 1 else numpy.linalg.inv(array) |
|
|
2637 | 2637 | except Exception as e: |
|
|
2638 | 2638 | print(e) |
|
|
2639 | 2639 | array[:]=numpy.NaN |
|
|
2640 | 2640 | |
|
|
2641 | 2641 | b = a + numpy.dot(numpy.transpose(beta),array/c) # New params |
|
|
2642 | 2642 | yfit = __GAUSSWINFIT1(b) # Evaluate function |
|
|
2643 | 2643 | chisq = numpy.sum(Weights*numpy.power(y-yfit,2,dtype='f4'),dtype='f4')/nfree # New chisq |
|
|
2644 | 2644 | sigma = numpy.sqrt(numpy.diag(array)/numpy.diag(alpha)) # New sigma |
|
|
2645 | 2645 | if (numpy.isfinite(chisq) == 0) or \ |
|
|
2646 | 2646 | (lambdaCount > 30 and chisq >= chisq1): |
|
|
2647 | 2647 | # Reject changes made this iteration, use old values. |
|
|
2648 | 2648 | yfit = yfit1 |
|
|
2649 | 2649 | sigma = sigma1 |
|
|
2650 | 2650 | chisq = chisq1 |
|
|
2651 | 2651 | converge = 0 |
|
|
2652 | 2652 | #print('Failed to converge.') |
|
|
2653 | 2653 | chi2 = chisq # Return chi-squared (chi2 obsolete-still works) |
|
|
2654 | 2654 | if done_early: Niter -= 1 |
|
|
2655 | 2655 | return yfit, a, converge, sigma, chisq, chi2 # return result |
|
|
2656 | 2656 | ten=numpy.array([10.0],dtype='f4')[0] |
|
|
2657 | 2657 | flambda *= ten # Assume fit got worse |
|
|
2658 | 2658 | if chisq <= chisq1: |
|
|
2659 | 2659 | break |
|
|
2660 | 2660 | hundred=numpy.array([100.0],dtype='f4')[0] |
|
|
2661 | 2661 | flambda /= hundred |
|
|
2662 | 2662 | |
|
|
2663 | 2663 | a=b # Save new parameter estimate. |
|
|
2664 | 2664 | if ((chisq1-chisq)/chisq1) <= tol: # Finished? |
|
|
2665 | 2665 | chi2 = chisq # Return chi-squared (chi2 obsolete-still works) |
|
|
2666 | 2666 | if done_early: Niter -= 1 |
|
|
2667 | 2667 | return yfit, a, converge, sigma, chisq, chi2 # return result |
|
|
2668 | 2668 | converge = 0 |
|
|
2669 | 2669 | chi2 = chisq |
|
|
2670 | 2670 | #print('Failed to converge.') |
|
|
2671 | 2671 | return yfit, a, converge, sigma, chisq, chi2 |
|
|
2672 | 2672 | |
|
|
2673 | 2673 | |
|
|
2674 | 2674 | def spectral_cut(Hei, ind, dbg_hei, freq, fd, snr, n1, w, ymax, spec, spec2, n0, max_spec, ss1, m, bb0, curr_ch, sel_ch): |
|
|
2675 | 2675 | if Hei[ind] > dbg_hei[0] and Hei[ind] < dbg_hei[1] and (curr_ch in sel_ch): |
|
|
2676 | 2676 | nsa=len(freq) |
|
|
2677 | 2677 | aux='H=%iKm, dop: %4.1f, snr: %4.1f, noise: %4.1f, sw: %4.1f'%(Hei[ind],fd, 10*numpy.log10(snr),10*numpy.log10(n1), w) |
|
|
2678 | 2678 | plt.subplots() |
|
|
2679 | 2679 | plt.ylim(0,ymax) |
|
|
2680 | 2680 | plt.plot(freq,spec,'b-',freq,spec2,'b--', freq,numpy.repeat(n1, nsa),'k--', freq,numpy.repeat(n0, nsa),'k-', freq,numpy.repeat(max_spec, nsa),'y.-', numpy.repeat(fd, nsa),numpy.linspace(0,ymax,nsa),'r--', numpy.repeat(freq[ss1], nsa),numpy.linspace(0,ymax,nsa),'g-.', numpy.repeat(freq[m + bb0], nsa),numpy.linspace(0,ymax,nsa),'g-.') |
|
|
2681 | 2681 | plt.title(aux) |
|
|
2682 | 2682 | plt.show() |
|
|
2683 | 2683 | |
|
|
2684 | 2684 | |
|
|
2685 | 2685 | if (nicoh is None): nicoh = 1 |
|
|
2686 | 2686 | if (smooth is None): smooth = 0 |
|
|
2687 | 2687 | if (type1 is None): type1 = 0 |
|
|
2688 | 2688 | if (vers is None): vers = 0 |
|
|
2689 | 2689 | if (fwindow is None): fwindow = numpy.zeros(oldfreq.size) + 1 |
|
|
2690 | 2690 | if (snrth is None): snrth = -20.0 |
|
|
2691 | 2691 | if (dc is None): dc = 0 |
|
|
2692 | 2692 | if (aliasing is None): aliasing = 0 |
|
|
2693 | 2693 | if (oldfd is None): oldfd = 0 |
|
|
2694 | 2694 | if (wwauto is None): wwauto = 0 |
|
|
2695 | 2695 | |
|
|
2696 | 2696 | if (n0 < 1.e-20): n0 = 1.e-20 |
|
|
2697 | 2697 | |
|
|
2698 | 2698 | xvalid = numpy.where(fwindow == 1)[0] |
|
|
2699 | 2699 | freq = oldfreq |
|
|
2700 | 2700 | truex = oldfreq |
|
|
2701 | 2701 | vec_power = numpy.zeros(oldspec.shape[1]) |
|
|
2702 | 2702 | vec_fd = numpy.zeros(oldspec.shape[1]) |
|
|
2703 | 2703 | vec_w = numpy.zeros(oldspec.shape[1]) |
|
|
2704 | 2704 | vec_snr = numpy.zeros(oldspec.shape[1]) |
|
|
2705 | 2705 | vec_n1 = numpy.empty(oldspec.shape[1]) |
|
|
2706 | 2706 | vec_fp = numpy.empty(oldspec.shape[1]) |
|
|
2707 | 2707 | vec_sigma_fd = numpy.empty(oldspec.shape[1]) |
|
|
2708 | 2708 | |
|
|
2709 | 2709 | for ind in range(oldspec.shape[1]): |
|
|
2710 | 2710 | spec = oldspec[:,ind] |
|
|
2711 | 2711 | if (smooth == 0): |
|
|
2712 | 2712 | spec2 = spec |
|
|
2713 | 2713 | else: |
|
|
2714 | 2714 | spec2 = scipy.ndimage.filters.uniform_filter1d(spec,size=smooth) |
|
|
2715 | 2715 | |
|
|
2716 | 2716 | aux = spec2*fwindow |
|
|
2717 | 2717 | max_spec = aux.max() |
|
|
2718 | 2718 | m = aux.tolist().index(max_spec) |
|
|
2719 | 2719 | |
|
|
2720 | 2720 | if m > 2 and m < oldfreq.size - 3: |
|
|
2721 | 2721 | newindex = m + numpy.array([-2,-1,0,1,2]) |
|
|
2722 | 2722 | newfreq = numpy.arange(20)/20.0*(numpy.max(freq[newindex])-numpy.min(freq[newindex]))+numpy.min(freq[newindex]) |
|
|
2723 | 2723 | tck = interpolate.splrep(freq[newindex], spec2[newindex]) |
|
|
2724 | 2724 | peakspec = interpolate.splev(newfreq, tck) |
|
|
2725 | 2725 | max_spec = numpy.max(peakspec) |
|
|
2726 | 2726 | mnew = numpy.argmax(peakspec) |
|
|
2727 | 2727 | fp = newfreq[mnew] |
|
|
2728 | 2728 | else: |
|
|
2729 | 2729 | fp = freq[m] |
|
|
2730 | 2730 | |
|
|
2731 | 2731 | if vers ==0: |
|
|
2732 | 2732 | |
|
|
2733 | 2733 | # Moments Estimation |
|
|
2734 | 2734 | bb = spec2[numpy.arange(m,spec2.size)] |
|
|
2735 | 2735 | bb = (bb<n0).nonzero() |
|
|
2736 | 2736 | bb = bb[0] |
|
|
2737 | 2737 | |
|
|
2738 | 2738 | ss = spec2[numpy.arange(0,m + 1)] |
|
|
2739 | 2739 | ss = (ss<n0).nonzero() |
|
|
2740 | 2740 | ss = ss[0] |
|
|
2741 | 2741 | |
|
|
2742 | 2742 | if (bb.size == 0): |
|
|
2743 | 2743 | bb0 = spec.size - 1 - m |
|
|
2744 | 2744 | else: |
|
|
2745 | 2745 | bb0 = bb[0] - 1 |
|
|
2746 | 2746 | if (bb0 < 0): |
|
|
2747 | 2747 | bb0 = 0 |
|
|
2748 | 2748 | |
|
|
2749 | 2749 | if (ss.size == 0): |
|
|
2750 | 2750 | ss1 = 1 |
|
|
2751 | 2751 | else: |
|
|
2752 | 2752 | ss1 = max(ss) + 1 |
|
|
2753 | 2753 | |
|
|
2754 | 2754 | if (ss1 > m): |
|
|
2755 | 2755 | ss1 = m |
|
|
2756 | 2756 | |
|
|
2757 | 2757 | valid = numpy.arange(int(m + bb0 - ss1 + 1)) + ss1 |
|
|
2758 | 2758 | |
|
|
2759 | 2759 | signal_power = ((spec2[valid] - n0) * fwindow[valid]).mean() # D. ScipiΓ³n added with correct definition |
|
|
2760 | 2760 | total_power = (spec2[valid] * fwindow[valid]).mean() # D. ScipiΓ³n added with correct definition |
|
|
2761 | 2761 | power = ((spec2[valid] - n0) * fwindow[valid]).sum() |
|
|
2762 | 2762 | fd = ((spec2[valid]- n0)*freq[valid] * fwindow[valid]).sum() / power |
|
|
2763 | 2763 | w = numpy.sqrt(((spec2[valid] - n0)*fwindow[valid]*(freq[valid]- fd)**2).sum() / power) |
|
|
2764 | 2764 | snr = (spec2.mean()-n0)/n0 |
|
|
2765 | 2765 | if (snr < 1.e-20): snr = 1.e-20 |
|
|
2766 | 2766 | |
|
|
2767 | 2767 | vec_power[ind] = total_power |
|
|
2768 | 2768 | vec_fd[ind] = fd |
|
|
2769 | 2769 | vec_w[ind] = w |
|
|
2770 | 2770 | vec_snr[ind] = snr |
|
|
2771 | 2771 | else: |
|
|
2772 | 2772 | # Noise by heights |
|
|
2773 | 2773 | n1, stdv = self.__get_noise2(spec, nicoh) |
|
|
2774 | 2774 | # Moments Estimation |
|
|
2775 | 2775 | bb = spec2[numpy.arange(m,spec2.size)] |
|
|
2776 | 2776 | bb = (bb<n1).nonzero() |
|
|
2777 | 2777 | bb = bb[0] |
|
|
2778 | 2778 | |
|
|
2779 | 2779 | ss = spec2[numpy.arange(0,m + 1)] |
|
|
2780 | 2780 | ss = (ss<n1).nonzero() |
|
|
2781 | 2781 | ss = ss[0] |
|
|
2782 | 2782 | |
|
|
2783 | 2783 | if (bb.size == 0): |
|
|
2784 | 2784 | bb0 = spec.size - 1 - m |
|
|
2785 | 2785 | else: |
|
|
2786 | 2786 | bb0 = bb[0] - 1 |
|
|
2787 | 2787 | if (bb0 < 0): |
|
|
2788 | 2788 | bb0 = 0 |
|
|
2789 | 2789 | |
|
|
2790 | 2790 | if (ss.size == 0): |
|
|
2791 | 2791 | ss1 = 1 |
|
|
2792 | 2792 | else: |
|
|
2793 | 2793 | ss1 = max(ss) + 1 |
|
|
2794 | 2794 | |
|
|
2795 | 2795 | if (ss1 > m): |
|
|
2796 | 2796 | ss1 = m |
|
|
2797 | 2797 | |
|
|
2798 | 2798 | valid = numpy.arange(int(m + bb0 - ss1 + 1)) + ss1 |
|
|
2799 | 2799 | |
|
|
2800 | 2800 | power = ((spec[valid] - n1)*fwindow[valid]).sum() |
|
|
2801 | 2801 | fd = ((spec[valid]- n1)*freq[valid]*fwindow[valid]).sum()/power |
|
|
2802 | 2802 | try: |
|
|
2803 | 2803 | w = numpy.sqrt(((spec[valid] - n1)*fwindow[valid]*(freq[valid]- fd)**2).sum()/power) |
|
|
2804 | 2804 | except: |
|
|
2805 | 2805 | w = float("NaN") |
|
|
2806 | 2806 | snr = power/(n0*fwindow.sum()) |
|
|
2807 | 2807 | |
|
|
2808 | 2808 | if debug: |
|
|
2809 | 2809 | spectral_cut(Hei, ind, dbg_hei, freq, fd, snr, n1, w, ymax, spec, spec2, n0, max_spec, ss1, m, bb0, curr_ch, sel_ch) |
|
|
2810 | 2810 | |
|
|
2811 | 2811 | if snr < 1.e-20: snr = 1.e-20 |
|
|
2812 | 2812 | |
|
|
2813 | 2813 | # Here start gaussean adjustment |
|
|
2814 | 2814 | |
|
|
2815 | 2815 | if type1 == 1 and snr > numpy.power(10,0.1*snrth): |
|
|
2816 | 2816 | |
|
|
2817 | 2817 | a = numpy.zeros(4,dtype='f4') |
|
|
2818 | 2818 | a[0] = snr * n0 |
|
|
2819 | 2819 | a[1] = fd |
|
|
2820 | 2820 | a[2] = w |
|
|
2821 | 2821 | a[3] = n0 |
|
|
2822 | 2822 | |
|
|
2823 | 2823 | np = spec.size |
|
|
2824 | 2824 | aold = a.copy() |
|
|
2825 | 2825 | spec2 = spec.copy() |
|
|
2826 | 2826 | oldxvalid = xvalid.copy() |
|
|
2827 | 2827 | |
|
|
2828 | 2828 | for i in range(2): |
|
|
2829 | 2829 | |
|
|
2830 | 2830 | ww = 1.0/(numpy.power(spec2,2)/nicoh) |
|
|
2831 | 2831 | ww[np//2] = 0.0 |
|
|
2832 | 2832 | |
|
|
2833 | 2833 | a = aold.copy() |
|
|
2834 | 2834 | xvalid = oldxvalid.copy() |
|
|
2835 | 2835 | #self.show_var(xvalid) |
|
|
2836 | 2836 | |
|
|
2837 | 2837 | gaussfn = __curvefit_koki(spec[xvalid], a, ww[xvalid]) |
|
|
2838 | 2838 | a = gaussfn[1] |
|
|
2839 | 2839 | converge = gaussfn[2] |
|
|
2840 | 2840 | |
|
|
2841 | 2841 | xvalid = numpy.arange(np) |
|
|
2842 | 2842 | spec2 = __GAUSSWINFIT1(a) |
|
|
2843 | 2843 | |
|
|
2844 | 2844 | xvalid = oldxvalid.copy() |
|
|
2845 | 2845 | power = a[0] * np |
|
|
2846 | 2846 | fd = a[1] |
|
|
2847 | 2847 | sigma_fd = gaussfn[3][1] |
|
|
2848 | 2848 | snr = max(power/ (max(a[3],n0) * len(oldxvalid)) * converge, 1e-20) |
|
|
2849 | 2849 | w = numpy.abs(a[2]) |
|
|
2850 | 2850 | n1 = max(a[3], n0) |
|
|
2851 | 2851 | |
|
|
2852 | 2852 | #gauss_adj=[fd,w,snr,n1,fp,sigma_fd] |
|
|
2853 | 2853 | else: |
|
|
2854 | 2854 | sigma_fd=numpy.nan # to avoid UnboundLocalError: local variable 'sigma_fd' referenced before assignment |
|
|
2855 | 2855 | |
|
|
2856 | 2856 | vec_fd[ind] = fd |
|
|
2857 | 2857 | vec_w[ind] = w |
|
|
2858 | 2858 | vec_snr[ind] = snr |
|
|
2859 | 2859 | vec_n1[ind] = n1 |
|
|
2860 | 2860 | vec_fp[ind] = fp |
|
|
2861 | 2861 | vec_sigma_fd[ind] = sigma_fd |
|
|
2862 | 2862 | vec_power[ind] = power # to compare with type 0 proccessing |
|
|
2863 | 2863 | |
|
|
2864 | 2864 | if vers==1: |
|
|
2865 | 2865 | return numpy.vstack((vec_snr, vec_w, vec_fd, vec_n1, vec_fp, vec_sigma_fd, vec_power)) |
|
|
2866 | 2866 | else: |
|
|
2867 | 2867 | return numpy.vstack((vec_snr, vec_power, vec_fd, vec_w)) |
|
|
2868 | 2868 | |
|
|
2869 | 2869 | def __get_noise2(self,POWER, fft_avg, TALK=0): |
|
|
2870 | 2870 | ''' |
|
|
2871 | 2871 | Rutina para cΓ‘lculo de ruido por alturas(n1). Similar a IDL |
|
|
2872 | 2872 | ''' |
|
|
2873 | 2873 | SPECT_PTS = len(POWER) |
|
|
2874 | 2874 | fft_avg = fft_avg*1.0 |
|
|
2875 | 2875 | NOMIT = 0 |
|
|
2876 | 2876 | NN = SPECT_PTS - NOMIT |
|
|
2877 | 2877 | N = NN//2 |
|
|
2878 | 2878 | ARR = numpy.concatenate((POWER[0:N+1],POWER[N+NOMIT+1:SPECT_PTS])) |
|
|
2879 | 2879 | ARR = numpy.sort(ARR) |
|
|
2880 | 2880 | NUMS_MIN = (SPECT_PTS+7)//8 |
|
|
2881 | 2881 | RTEST = (1.0+1.0/fft_avg) |
|
|
2882 | 2882 | SUM = 0.0 |
|
|
2883 | 2883 | SUMSQ = 0.0 |
|
|
2884 | 2884 | J = 0 |
|
|
2885 | 2885 | for I in range(NN): |
|
|
2886 | 2886 | J = J + 1 |
|
|
2887 | 2887 | SUM = SUM + ARR[I] |
|
|
2888 | 2888 | SUMSQ = SUMSQ + ARR[I]*ARR[I] |
|
|
2889 | 2889 | AVE = SUM*1.0/J |
|
|
2890 | 2890 | if J > NUMS_MIN: |
|
|
2891 | 2891 | if (SUMSQ*J <= RTEST*SUM*SUM): RNOISE = AVE |
|
|
2892 | 2892 | else: |
|
|
2893 | 2893 | if J == NUMS_MIN: RNOISE = AVE |
|
|
2894 | 2894 | if TALK == 1: print('Noise Power (2):%4.4f' %RNOISE) |
|
|
2895 | 2895 | stdv = numpy.sqrt(SUMSQ/J - numpy.power(SUM/J,2)) |
|
|
2896 | 2896 | return RNOISE, stdv |
|
|
2897 | 2897 | |
|
|
2898 | 2898 | def __get_noise1(self, power, fft_avg, TALK=0): |
|
|
2899 | 2899 | ''' |
|
|
2900 | 2900 | Rutina para cΓ‘lculo de ruido por alturas(n0). Similar a IDL |
|
|
2901 | 2901 | ''' |
|
|
2902 | 2902 | num_pts = numpy.size(power) |
|
|
2903 | 2903 | fft_avg = fft_avg*1.0 |
|
|
2904 | 2904 | ind = numpy.argsort(power, axis=None, kind='stable') |
|
|
2905 | 2905 | ARR = numpy.reshape(power,-1)[ind] |
|
|
2906 | 2906 | NUMS_MIN = num_pts//10 |
|
|
2907 | 2907 | RTEST = (1.0+1.0/fft_avg) |
|
|
2908 | 2908 | SUM = 0.0 |
|
|
2909 | 2909 | SUMSQ = 0.0 |
|
|
2910 | 2910 | J = 0 |
|
|
2911 | 2911 | cont = 1 |
|
|
2912 | 2912 | while cont == 1 and J < num_pts: |
|
|
2913 | 2913 | |
|
|
2914 | 2914 | SUM = SUM + ARR[J] |
|
|
2915 | 2915 | SUMSQ = SUMSQ + ARR[J]*ARR[J] |
|
|
2916 | 2916 | J = J + 1 |
|
|
2917 | 2917 | |
|
|
2918 | 2918 | if J > NUMS_MIN: |
|
|
2919 | 2919 | if (SUMSQ*J <= RTEST*SUM*SUM): |
|
|
2920 | 2920 | LNOISE = SUM*1.0/J |
|
|
2921 | 2921 | else: |
|
|
2922 | 2922 | J = J - 1 |
|
|
2923 | 2923 | SUM = SUM - ARR[J] |
|
|
2924 | 2924 | SUMSQ = SUMSQ - ARR[J]*ARR[J] |
|
|
2925 | 2925 | cont = 0 |
|
|
2926 | 2926 | else: |
|
|
2927 | 2927 | if J == NUMS_MIN: LNOISE = SUM*1.0/J |
|
|
2928 | 2928 | if TALK == 1: print('Noise Power (1):%8.8f' %LNOISE) |
|
|
2929 | 2929 | stdv = numpy.sqrt(SUMSQ/J - numpy.power(SUM/J,2)) |
|
|
2930 | 2930 | return LNOISE, stdv |
|
|
2931 | 2931 | |
|
|
2932 | 2932 | def __NoiseByChannel(self, num_prof, num_incoh, spectra,talk=0): |
|
|
2933 | 2933 | |
|
|
2934 | 2934 | val_frq = numpy.arange(num_prof-2)+1 |
|
|
2935 | 2935 | val_frq[(num_prof-2)//2:] = val_frq[(num_prof-2)//2:] + 1 |
|
|
2936 | 2936 | junkspc = numpy.sum(spectra[val_frq,:], axis=1) |
|
|
2937 | 2937 | junkid = numpy.argsort(junkspc) |
|
|
2938 | 2938 | noisezone = val_frq[junkid[0:num_prof//2]] |
|
|
2939 | 2939 | specnoise = spectra[noisezone,:] |
|
|
2940 | 2940 | noise, stdvnoise = self.__get_noise1(specnoise,num_incoh) |
|
|
2941 | 2941 | |
|
|
2942 | 2942 | if talk: |
|
|
2943 | 2943 | print('noise =', noise) |
|
|
2944 | 2944 | return noise, stdvnoise |
|
|
2945 | 2945 | |
|
|
2946 | 2946 | def run(self, dataOut, proc_type=0, mode_fit=0, exp='150EEJ', debug=False, dbg_hei=None, ymax=1, sel_ch=[0,1]): |
|
|
2947 | 2947 | |
|
|
2948 | 2948 | absc = dataOut.abscissaList[:-1] |
|
|
2949 | 2949 | nChannel = dataOut.data_pre[0].shape[0] |
|
|
2950 | 2950 | nHei = dataOut.data_pre[0].shape[2] |
|
|
2951 | 2951 | Hei=dataOut.heightList |
|
|
2952 | 2952 | data_param = numpy.zeros((nChannel, 4 + proc_type*3, nHei)) |
|
|
2953 | 2953 | nProfiles = dataOut.nProfiles |
|
|
2954 | 2954 | nCohInt = dataOut.nCohInt |
|
|
2955 | 2955 | nIncohInt = dataOut.nIncohInt |
|
|
2956 | 2956 | M = numpy.power(numpy.array(1/(nProfiles * nCohInt) ,dtype='float32'),2) |
|
|
2957 | 2957 | N = numpy.array(M / nIncohInt,dtype='float32') |
|
|
2958 | 2958 | |
|
|
2959 | 2959 | if proc_type == 1: |
|
|
2960 | 2960 | type1 = mode_fit |
|
|
2961 | 2961 | fwindow = numpy.zeros(absc.size) + 1 |
|
|
2962 | 2962 | if exp == '150EEJ': |
|
|
2963 | 2963 | b=64 |
|
|
2964 | 2964 | fwindow[0:absc.size//2 - b] = 0 |
|
|
2965 | 2965 | fwindow[absc.size//2 + b:] = 0 |
|
|
2966 | 2966 | vers = 1 # new |
|
|
2967 | 2967 | |
|
|
2968 | 2968 | data = dataOut.data_pre[0] * N |
|
|
2969 | 2969 | |
|
|
2970 | 2970 | noise = numpy.zeros(nChannel) |
|
|
2971 | 2971 | stdvnoise = numpy.zeros(nChannel) |
|
|
2972 | 2972 | for ind in range(nChannel): |
|
|
2973 | 2973 | noise[ind], stdvnoise[ind] = self.__NoiseByChannel(nProfiles, nIncohInt, data[ind,:,:]) |
|
|
2974 | 2974 | smooth=3 |
|
|
2975 | 2975 | else: |
|
|
2976 | 2976 | data = dataOut.data_pre[0] |
|
|
2977 | 2977 | noise = dataOut.noise |
|
|
2978 | 2978 | fwindow = None |
|
|
2979 | 2979 | type1 = None |
|
|
2980 | 2980 | vers = 0 # old |
|
|
2981 | 2981 | nIncohInt = None |
|
|
2982 | 2982 | smooth=None |
|
|
2983 | 2983 | |
|
|
2984 | 2984 | for ind in range(nChannel): |
|
|
2985 | 2985 | data_param[ind,:,:] = self.__calculateMoments(data[ind,:,:] , absc , noise[ind], nicoh=nIncohInt, smooth=smooth, type1=type1, fwindow=fwindow, vers=vers, Hei=Hei, debug=debug, dbg_hei=dbg_hei, ymax=ymax, curr_ch=ind, sel_ch=sel_ch) |
|
|
2986 | 2986 | #data_param[ind,:,:] = self.__calculateMoments(data[ind,:,:] , absc , noise[ind], nicoh=nIncohInt, smooth=smooth, type1=type1, fwindow=fwindow, vers=vers, Hei=Hei, debug=debug) |
|
|
2987 | 2987 | if exp == 'ESF_EW': |
|
|
2988 | 2988 | data_param[ind,0,:]*=(noise[ind]/stdvnoise[ind]) |
|
|
2989 | 2989 | data_param[ind,3,:]*=(1.0/M) |
|
|
2990 | 2990 | |
|
|
2991 | 2991 | if proc_type == 1: |
|
|
2992 | 2992 | dataOut.moments = data_param[:,1:,:] |
|
|
2993 | 2993 | dataOut.data_dop = data_param[:,2] |
|
|
2994 | 2994 | dataOut.data_width = data_param[:,1] |
|
|
2995 | 2995 | dataOut.data_snr = data_param[:,0] |
|
|
2996 | 2996 | dataOut.data_pow = data_param[:,6] # to compare with type0 proccessing |
|
|
2997 | 2997 | dataOut.spcpar=numpy.stack((dataOut.data_dop,dataOut.data_width,dataOut.data_snr, data_param[:,3], data_param[:,4],data_param[:,5]),axis=2) |
|
|
2998 | 2998 | |
|
|
2999 | 2999 | if exp == 'ESF_EW': |
|
|
3000 | 3000 | spc=dataOut.data_pre[0]* N |
|
|
3001 | 3001 | cspc=dataOut.data_pre[1]* N |
|
|
3002 | 3002 | nHei=dataOut.data_pre[1].shape[2] |
|
|
3003 | 3003 | cross_pairs=dataOut.pairsList |
|
|
3004 | 3004 | nDiffIncohInt = dataOut.nDiffIncohInt |
|
|
3005 | 3005 | N2=numpy.array(1 / nDiffIncohInt,dtype='float32') |
|
|
3006 | 3006 | diffcspectra=dataOut.data_diffcspc.copy()* N2 * M * M |
|
|
3007 | 3007 | num_pairs=len(dataOut.pairsList) |
|
|
3008 | 3008 | |
|
|
3009 | 3009 | if num_pairs >= 0: |
|
|
3010 | 3010 | fbinv=numpy.where(absc != 0)[0] |
|
|
3011 | 3011 | ccf=numpy.sum(cspc[:,fbinv,:], axis=1) |
|
|
3012 | 3012 | jvpower=numpy.sum(spc[:,fbinv,:], axis=1) |
|
|
3013 | 3013 | coh=ccf/numpy.sqrt(jvpower[cross_pairs[0][0],:]*jvpower[cross_pairs[0][1],:]) |
|
|
3014 | 3014 | dccf=numpy.sum(diffcspectra[:,fbinv,:], axis=1) |
|
|
3015 | 3015 | dataOut.ccfpar = numpy.zeros((num_pairs,nHei,3)) |
|
|
3016 | 3016 | dataOut.ccfpar[:,:,0]=numpy.abs(coh) |
|
|
3017 | 3017 | dataOut.ccfpar[:,:,1]=numpy.arctan(numpy.imag(coh)/numpy.real(coh)) |
|
|
3018 | 3018 | dataOut.ccfpar[:,:,2]=numpy.arctan(numpy.imag(dccf)/numpy.real(dccf)) |
|
|
3019 | 3019 | else: |
|
|
3020 | 3020 | dataOut.moments = data_param[:,1:,:] |
|
|
3021 | 3021 | dataOut.data_snr = data_param[:,0] |
|
|
3022 | 3022 | dataOut.data_pow = data_param[:,1] |
|
|
3023 | 3023 | dataOut.data_dop = data_param[:,2] |
|
|
3024 | 3024 | dataOut.data_width = data_param[:,3] |
|
|
3025 | 3025 | dataOut.spcpar=numpy.stack((dataOut.data_dop,dataOut.data_width,dataOut.data_snr, dataOut.data_pow),axis=2) |
|
|
3026 | 3026 | |
|
|
3027 | 3027 | return dataOut |
|
|
3028 | 3028 | |
|
|
3029 | 3029 | |
|
|
3030 | 3030 | class JULIA_DayVelocities(Operation): |
|
|
3031 | 3031 | ''' |
|
|
3032 | 3032 | Function SpectralMoments() |
|
|
3033 | 3033 | |
|
|
3034 | 3034 | From espectral parameters calculates: |
|
|
3035 | 3035 | |
|
|
3036 | 3036 | 1. Signal to noise level (SNL) |
|
|
3037 | 3037 | 2. Vertical velocity |
|
|
3038 | 3038 | 3. Zonal velocity |
|
|
3039 | 3039 | 4. Vertical velocity error |
|
|
3040 | 3040 | 5. Zonal velocity error. |
|
|
3041 | 3041 | |
|
|
3042 | 3042 | Type of dataIn: SpectralMoments |
|
|
3043 | 3043 | |
|
|
3044 | 3044 | Configuration Parameters: |
|
|
3045 | 3045 | |
|
|
3046 | 3046 | zenith : Pairs of angles corresponding to the two beams related to the perpendicular to B from the center of the antenna. |
|
|
3047 | 3047 | zenithCorrection : Adjustment angle for the zenith. Default 0. |
|
|
3048 | 3048 | heights : Range to process 150kM echoes. By default [125,185]. |
|
|
3049 | 3049 | nchan : To process 2 or 1 channel. 2 by default. |
|
|
3050 | 3050 | chan : If nchan = 1, chan indicates which of the 2 channels to process. |
|
|
3051 | 3051 | clean : 2nd cleaning processing (Graphical). Default False |
|
|
3052 | 3052 | driftstdv_th : Diferencia mΓ‘xima entre valores promedio consecutivos de vertical. |
|
|
3053 | 3053 | zonalstdv_th : Diferencia mΓ‘xima entre valores promedio consecutivos de zonal. |
|
|
3054 | 3054 | |
|
|
3055 | 3055 | Input: |
|
|
3056 | 3056 | |
|
|
3057 | 3057 | Affected: |
|
|
3058 | 3058 | |
|
|
3059 | 3059 | ''' |
|
|
3060 | 3060 | |
|
|
3061 | 3061 | def __init__(self): |
|
|
3062 | 3062 | Operation.__init__(self) |
|
|
3063 | 3063 | self.old_drift=None |
|
|
3064 | 3064 | self.old_zonal=None |
|
|
3065 | 3065 | self.count_drift=0 |
|
|
3066 | 3066 | self.count_zonal=0 |
|
|
3067 | 3067 | self.oldTime_drift=None |
|
|
3068 | 3068 | self.oldTime_zonal=None |
|
|
3069 | 3069 | |
|
|
3070 | 3070 | def newtotal(self, data): |
|
|
3071 | 3071 | return numpy.nansum(data) |
|
|
3072 | 3072 | |
|
|
3073 | 3073 | def data_filter(self, parm, snrth=-20, swth=20, wErrth=500): |
|
|
3074 | 3074 | |
|
|
3075 | 3075 | Sz0 = parm.shape # Sz0: h,p |
|
|
3076 | 3076 | drift = parm[:,0] |
|
|
3077 | 3077 | sw = 2*parm[:,1] |
|
|
3078 | 3078 | snr = 10*numpy.log10(parm[:,2]) |
|
|
3079 | 3079 | Sz = drift.shape # Sz: h |
|
|
3080 | 3080 | mask = numpy.ones((Sz[0])) |
|
|
3081 | 3081 | th=0 |
|
|
3082 | 3082 | valid=numpy.where(numpy.isfinite(snr)) |
|
|
3083 | 3083 | cvalid = len(valid[0]) |
|
|
3084 | 3084 | if cvalid >= 1: |
|
|
3085 | 3085 | # CΓ‘lculo del ruido promedio de snr para el i-Γ©simo grupo de alturas |
|
|
3086 | 3086 | nbins = int(numpy.max(snr)-numpy.min(snr))+1 # bin size = 1, similar to IDL |
|
|
3087 | 3087 | h = numpy.histogram(snr,bins=nbins) |
|
|
3088 | 3088 | hist = h[0] |
|
|
3089 | 3089 | values = numpy.round_(h[1]) |
|
|
3090 | 3090 | moda = values[numpy.where(hist == numpy.max(hist))] |
|
|
3091 | 3091 | indNoise = numpy.where(numpy.abs(snr - numpy.min(moda)) < 3)[0] |
|
|
3092 | 3092 | |
|
|
3093 | 3093 | noise = snr[indNoise] |
|
|
3094 | 3094 | noise_mean = numpy.sum(noise)/len(noise) |
|
|
3095 | 3095 | # CΓ‘lculo de media de snr |
|
|
3096 | 3096 | med = numpy.median(snr) |
|
|
3097 | 3097 | # Establece el umbral de snr |
|
|
3098 | 3098 | if noise_mean > med + 3: |
|
|
3099 | 3099 | th = med |
|
|
3100 | 3100 | else: |
|
|
3101 | 3101 | th = noise_mean + 3 |
|
|
3102 | 3102 | # Establece mΓ‘scara |
|
|
3103 | 3103 | novalid = numpy.where(snr <= th)[0] |
|
|
3104 | 3104 | mask[novalid] = numpy.nan |
|
|
3105 | 3105 | # Elimina datos que no sobrepasen el umbral: PARAMETRO |
|
|
3106 | 3106 | novalid = numpy.where(snr <= snrth) |
|
|
3107 | 3107 | cnovalid = len(novalid[0]) |
|
|
3108 | 3108 | if cnovalid > 0: |
|
|
3109 | 3109 | mask[novalid] = numpy.nan |
|
|
3110 | 3110 | novalid = numpy.where(numpy.isnan(snr)) |
|
|
3111 | 3111 | cnovalid = len(novalid[0]) |
|
|
3112 | 3112 | if cnovalid > 0: |
|
|
3113 | 3113 | mask[novalid] = numpy.nan |
|
|
3114 | 3114 | |
|
|
3115 | 3115 | new_parm = numpy.zeros((Sz0[0],Sz0[1])) |
|
|
3116 | 3116 | for i in range(Sz0[1]): |
|
|
3117 | 3117 | new_parm[:,i] = parm[:,i] * mask |
|
|
3118 | 3118 | |
|
|
3119 | 3119 | return new_parm, th |
|
|
3120 | 3120 | |
|
|
3121 | 3121 | |
|
|
3122 | 3122 | def statistics150km(self, veloc , sigma , threshold , old_veloc=None, count=0, \ |
|
|
3123 |
currTime=None, oldTime=None, amountdata= |
|
|
|
3123 | currTime=None, oldTime=None, amountdata=3, clearAll = None, timeFactor=1800, debug = False): | |
|
|
3124 | 3124 | |
|
|
3125 | 3125 | if oldTime == None: |
|
|
3126 | 3126 | oldTime = currTime |
|
|
3127 | 3127 | |
|
|
3128 | 3128 | step = (threshold/2)*(numpy.abs(currTime - oldTime)//timeFactor + 1) |
|
|
3129 | 3129 | factor = 2 |
|
|
3130 | 3130 | avg_threshold = 100 |
|
|
3131 | 3131 | # Calcula la mediana en todas las alturas por tiempo |
|
|
3132 | 3132 | val1=numpy.nanmedian(veloc) |
|
|
3133 | 3133 | |
|
|
3134 | 3134 | # Calcula la media ponderada en todas las alturas por tiempo |
|
|
3135 | 3135 | val2 = self.newtotal(veloc/numpy.power(sigma,2))/self.newtotal(1/numpy.power(sigma,2)) |
|
|
3136 | 3136 | |
|
|
3137 | 3137 | # Verifica la cercanΓa de los valores calculados de mediana y media, si son cercanos escoge la media ponderada |
|
|
3138 | 3138 | op1=numpy.abs(val2-val1) |
|
|
3139 | 3139 | op2=threshold/factor |
|
|
3140 | 3140 | cond = op1 < op2 |
|
|
3141 | 3141 | |
|
|
3142 | 3142 | veloc_prof = val2 if cond else val1 |
|
|
3143 | 3143 | sigma_prof = numpy.nan |
|
|
3144 | 3144 | sets=numpy.array([-1]) |
|
|
3145 | 3145 | |
|
|
3146 | 3146 | if op1 > avg_threshold: #Si son muy lejanos no toma en cuenta estos datos |
|
|
3147 | 3147 | veloc_prof = numpy.nan |
|
|
3148 | 3148 | |
|
|
3149 | 3149 | # Se calcula nuevamente media ponderada, en base a estimado inicial de la media |
|
|
3150 | 3150 | # a fin de eliminar valores que estΓ‘n muy lejanos a dicho valor |
|
|
3151 | 3151 | |
|
|
3152 | 3152 | if debug: |
|
|
3153 | 3153 | print('veloc_prof:', veloc_prof) |
|
|
3154 | 3154 | print('veloc:',veloc) |
|
|
3155 | 3155 | print('threshold:',threshold) |
|
|
3156 | 3156 | print('factor:',factor) |
|
|
3157 | 3157 | print('threshold/factor:',threshold/factor) |
|
|
3158 | 3158 | print('numpy.abs(veloc-veloc_prof):', numpy.abs(veloc-veloc_prof)) |
|
|
3159 | 3159 | print('numpy.where(numpy.abs(veloc-veloc_prof) < threshold/factor)[0]:', numpy.where(numpy.abs(veloc-veloc_prof) < threshold/factor)[0]) |
|
|
3160 | 3160 | |
|
|
3161 | 3161 | junk = numpy.where(numpy.abs(veloc-veloc_prof) < threshold/factor)[0] |
|
|
3162 |
if junk.size > |
|
|
|
3162 | if junk.size >= amountdata: | |
|
|
3163 | 3163 | veloc_prof = self.newtotal(veloc[junk]/numpy.power(sigma[junk],2))/self.newtotal(1/numpy.power(sigma[junk],2)) |
|
|
3164 | 3164 | sigma_prof1 = numpy.sqrt(1/self.newtotal(1/numpy.power(sigma[junk],2))) |
|
|
3165 | 3165 | sigma_prof2 = numpy.sqrt(self.newtotal(numpy.power(veloc[junk]-veloc_prof,2)/numpy.power(sigma[junk],2)))*sigma_prof1 |
|
|
3166 | 3166 | sigma_prof = numpy.sqrt(numpy.power(sigma_prof1,2)+numpy.power(sigma_prof2,2)) |
|
|
3167 | 3167 | sets = junk |
|
|
3168 | 3168 | |
|
|
3169 | 3169 | # Compara con valor anterior para evitar presencia de "outliers" |
|
|
3170 | 3170 | if debug: |
|
|
3171 | 3171 | print('old_veloc:',old_veloc) |
|
|
3172 | 3172 | print('step:', step) |
|
|
3173 | 3173 | |
|
|
3174 | 3174 | if old_veloc == None: |
|
|
3175 | 3175 | valid=numpy.isfinite(veloc_prof) |
|
|
3176 | 3176 | else: |
|
|
3177 | 3177 | valid=numpy.abs(veloc_prof-old_veloc) < step |
|
|
3178 | 3178 | |
|
|
3179 | 3179 | if debug: |
|
|
3180 | 3180 | print('valid:', valid) |
|
|
3181 | 3181 | |
|
|
3182 | 3182 | if not valid: |
|
|
3183 | 3183 | aver_veloc=numpy.nan |
|
|
3184 | 3184 | aver_sigma=numpy.nan |
|
|
3185 | 3185 | sets=numpy.array([-1]) |
|
|
3186 | 3186 | else: |
|
|
3187 | 3187 | aver_veloc=veloc_prof |
|
|
3188 | 3188 | aver_sigma=sigma_prof |
|
|
3189 | 3189 | clearAll=0 |
|
|
3190 | 3190 | if old_veloc != None and count < 5: |
|
|
3191 | 3191 | if numpy.abs(veloc_prof-old_veloc) > step: |
|
|
3192 | 3192 | clearAll=1 |
|
|
3193 | 3193 | count=0 |
|
|
3194 | 3194 | old_veloc=None |
|
|
3195 | 3195 | if numpy.isfinite(aver_veloc): |
|
|
3196 | 3196 | |
|
|
3197 | 3197 | count+=1 |
|
|
3198 | 3198 | if old_veloc != None: |
|
|
3199 | 3199 | old_veloc = (old_veloc + aver_veloc) * 0.5 |
|
|
3200 | 3200 | else: |
|
|
3201 | 3201 | old_veloc=aver_veloc |
|
|
3202 | 3202 | oldTime=currTime |
|
|
3203 | 3203 | if debug: |
|
|
3204 | 3204 | print('count:',count) |
|
|
3205 | 3205 | print('sets:',sets) |
|
|
3206 | 3206 | return sets, old_veloc, count, oldTime, aver_veloc, aver_sigma, clearAll |
|
|
3207 | 3207 | |
|
|
3208 | 3208 | |
|
|
3209 | def run(self, dataOut, zenith, zenithCorrection=0.0, heights=[125, 185], nchan=2, chan=0, clean=False, driftstdv_th=100, zonalstdv_th=200): | |
|
|
3209 | def run(self, dataOut, zenith, zenithCorrection=0.0, heights=[125, 185], nchan=2, chan=0, clean=False, driftstdv_th=100, zonalstdv_th=200, amountdata=3): | |
|
|
3210 | 3210 | |
|
|
3211 | 3211 | dataOut.lat=-11.95 |
|
|
3212 | 3212 | dataOut.lon=-76.87 |
|
|
3213 | 3213 | |
|
|
3214 | 3214 | nCh=dataOut.spcpar.shape[0] |
|
|
3215 | 3215 | nHei=dataOut.spcpar.shape[1] |
|
|
3216 | 3216 | nParam=dataOut.spcpar.shape[2] |
|
|
3217 | 3217 | |
|
|
3218 | 3218 | # SelecciΓ³n de alturas |
|
|
3219 | 3219 | hei=dataOut.heightList |
|
|
3220 | 3220 | hvalid=numpy.where([hei >= heights[0]][0] & [hei <= heights[1]][0])[0] |
|
|
3221 | 3221 | nhvalid=len(hvalid) |
|
|
3222 | 3222 | dataOut.heightList = hei[hvalid] |
|
|
3223 | 3223 | parm=numpy.empty((nCh,nhvalid,nParam)); parm[:]=numpy.nan |
|
|
3224 | 3224 | parm[:] = dataOut.spcpar[:,hvalid,:] |
|
|
3225 | 3225 | # Primer filtrado: Umbral de SNR |
|
|
3226 | 3226 | for i in range(nCh): |
|
|
3227 | 3227 | parm[i,:,:] = self.data_filter(parm[i,:,:])[0] |
|
|
3228 | 3228 | |
|
|
3229 | 3229 | zenith = numpy.array(zenith) |
|
|
3230 | 3230 | zenith -= zenithCorrection |
|
|
3231 | 3231 | zenith *= numpy.pi/180 |
|
|
3232 | 3232 | alpha = zenith[0] |
|
|
3233 | 3233 | beta = zenith[1] |
|
|
3234 | 3234 | dopplerCH0 = parm[0,:,0] |
|
|
3235 | 3235 | dopplerCH1 = parm[1,:,0] |
|
|
3236 | 3236 | swCH0 = parm[0,:,1] |
|
|
3237 | 3237 | swCH1 = parm[1,:,1] |
|
|
3238 | 3238 | snrCH0 = 10*numpy.log10(parm[0,:,2]) |
|
|
3239 | 3239 | snrCH1 = 10*numpy.log10(parm[1,:,2]) |
|
|
3240 | 3240 | noiseCH0 = parm[0,:,3] |
|
|
3241 | 3241 | noiseCH1 = parm[1,:,3] |
|
|
3242 | 3242 | wErrCH0 = parm[0,:,5] |
|
|
3243 | 3243 | wErrCH1 = parm[1,:,5] |
|
|
3244 | 3244 | |
|
|
3245 | 3245 | # Vertical and zonal calculation: nchan=2 by default |
|
|
3246 | 3246 | # Only vertical calculation, for offline processing with only one channel with good signal |
|
|
3247 | 3247 | if nchan == 1: |
|
|
3248 | 3248 | if chan == 1: |
|
|
3249 | 3249 | drift = - dopplerCH1 |
|
|
3250 | 3250 | snr = snrCH1 |
|
|
3251 | 3251 | noise = noiseCH1 |
|
|
3252 | 3252 | sw = swCH1 |
|
|
3253 | 3253 | w_w_err = wErrCH1 |
|
|
3254 | 3254 | elif chan == 0: |
|
|
3255 | 3255 | drift = - dopplerCH0 |
|
|
3256 | 3256 | snr = snrCH0 |
|
|
3257 | 3257 | noise = noiseCH0 |
|
|
3258 | 3258 | sw = swCH0 |
|
|
3259 | 3259 | w_w_err = wErrCH0 |
|
|
3260 | 3260 | |
|
|
3261 | 3261 | elif nchan == 2: |
|
|
3262 | 3262 | sinB_A = numpy.sin(beta)*numpy.cos(alpha) - numpy.sin(alpha)* numpy.cos(beta) |
|
|
3263 | 3263 | drift = -(dopplerCH0 * numpy.sin(beta) - dopplerCH1 * numpy.sin(alpha))/ sinB_A |
|
|
3264 | 3264 | zonal = (dopplerCH0 * numpy.cos(beta) - dopplerCH1 * numpy.cos(alpha))/ sinB_A |
|
|
3265 | 3265 | snr = (snrCH0 + snrCH1)/2 |
|
|
3266 | 3266 | noise = (noiseCH0 + noiseCH1)/2 |
|
|
3267 | 3267 | sw = (swCH0 + swCH1)/2 |
|
|
3268 | 3268 | w_w_err= numpy.sqrt(numpy.power(wErrCH0 * numpy.sin(beta)/numpy.abs(sinB_A),2) + numpy.power(wErrCH1 * numpy.sin(alpha)/numpy.abs(sinB_A),2)) |
|
|
3269 | 3269 | w_e_err= numpy.sqrt(numpy.power(wErrCH0 * numpy.cos(beta)/numpy.abs(-1*sinB_A),2) + numpy.power(wErrCH1 * numpy.cos(alpha)/numpy.abs(-1*sinB_A),2)) |
|
|
3270 | 3270 | |
|
|
3271 | 3271 | # 150Km statistics to clean data |
|
|
3272 | 3272 | clean_drift = drift.copy() |
|
|
3273 | 3273 | clean_drift[:] = numpy.nan |
|
|
3274 | 3274 | if nchan == 2: |
|
|
3275 | 3275 | clean_zonal = zonal.copy() |
|
|
3276 | 3276 | clean_zonal[:] = numpy.nan |
|
|
3277 | 3277 | |
|
|
3278 | 3278 | # Vertical |
|
|
3279 | 3279 | sets1, self.old_drift, self.count_drift, self.oldTime_drift, aver_veloc, aver_sigma, clearAll = self.statistics150km(drift, w_w_err, driftstdv_th, \ |
|
|
3280 | 3280 | old_veloc=self.old_drift, count=self.count_drift, currTime=dataOut.utctime, \ |
|
|
3281 | oldTime=self.oldTime_drift, timeFactor=120) | |
|
|
3281 | oldTime=self.oldTime_drift, amountdata = amountdata, timeFactor=120, debug = False) | |
|
|
3282 | 3282 | if clearAll == 1: |
|
|
3283 | 3283 | mean_zonal = numpy.nan |
|
|
3284 | 3284 | sigma_zonal = numpy.nan |
|
|
3285 | 3285 | mean_drift = aver_veloc |
|
|
3286 | 3286 | sigma_drift = aver_sigma |
|
|
3287 | 3287 | |
|
|
3288 | 3288 | if sets1.size != 1: |
|
|
3289 | 3289 | clean_drift[sets1] = drift[sets1] |
|
|
3290 | 3290 | |
|
|
3291 | 3291 | novalid=numpy.where(numpy.isnan(clean_drift))[0]; cnovalid=novalid.size |
|
|
3292 | 3292 | if cnovalid > 0: drift[novalid] = numpy.nan |
|
|
3293 | 3293 | if cnovalid > 0: snr[novalid] = numpy.nan |
|
|
3294 | 3294 | |
|
|
3295 | 3295 | # Zonal |
|
|
3296 | 3296 | if nchan == 2: |
|
|
3297 | 3297 | sets2, self.old_zonal, self.count_zonal, self.oldTime_zonal, aver_veloc, aver_sigma, clearAll = self.statistics150km(zonal, w_e_err, zonalstdv_th, \ |
|
|
3298 | 3298 | old_veloc=self.old_zonal, count=self.count_zonal, currTime=dataOut.utctime, \ |
|
|
3299 | oldTime=self.oldTime_zonal, timeFactor=600) | |
|
|
3299 | oldTime=self.oldTime_zonal, amountdata = amountdata, timeFactor=600, debug = False) | |
|
|
3300 | 3300 | if clearAll == 1: |
|
|
3301 | 3301 | mean_zonal = numpy.nan |
|
|
3302 | 3302 | sigma_zonal = numpy.nan |
|
|
3303 | 3303 | mean_zonal = aver_veloc |
|
|
3304 | 3304 | sigma_zonal = aver_sigma |
|
|
3305 | 3305 | if sets2.size != 1: |
|
|
3306 | 3306 | clean_zonal[sets2] = zonal[sets2] |
|
|
3307 | 3307 | |
|
|
3308 | 3308 | novalid=numpy.where(numpy.isnan(clean_zonal))[0]; cnovalid=novalid.size |
|
|
3309 | 3309 | if cnovalid > 0: zonal[novalid] = numpy.nan |
|
|
3310 | 3310 | if cnovalid > 0: snr[novalid] = numpy.nan |
|
|
3311 | 3311 | |
|
|
3312 | 3312 | n_avg_par=4 |
|
|
3313 | 3313 | avg_par=numpy.empty((n_avg_par,)); avg_par[:] = numpy.nan |
|
|
3314 | 3314 | avg_par[0,]=mean_drift |
|
|
3315 | 3315 | avg_par[1,]=mean_zonal |
|
|
3316 | 3316 | avg_par[2,]=sigma_drift |
|
|
3317 | 3317 | avg_par[3,]=sigma_zonal |
|
|
3318 | 3318 | |
|
|
3319 | 3319 | set1 = 1.0 |
|
|
3320 | 3320 | navg = set1 |
|
|
3321 | 3321 | nci = dataOut.nCohInt |
|
|
3322 | 3322 | # ---------------------------------- |
|
|
3323 | 3323 | ipp = 252.0 |
|
|
3324 | 3324 | nincoh = dataOut.nIncohInt |
|
|
3325 | 3325 | nptsfft = dataOut.nProfiles |
|
|
3326 | 3326 | hardcoded=False # if True, similar to IDL processing |
|
|
3327 | 3327 | if hardcoded: |
|
|
3328 | 3328 | ipp=200.1 |
|
|
3329 | 3329 | nincoh=22 |
|
|
3330 | 3330 | nptsfft=128 |
|
|
3331 | 3331 | # ---------------------------------- |
|
|
3332 | 3332 | nipp = ipp * nci |
|
|
3333 | 3333 | height = dataOut.heightList |
|
|
3334 | 3334 | nHei = len(height) |
|
|
3335 | 3335 | kd = 213.6 |
|
|
3336 | 3336 | nint = nptsfft * nincoh |
|
|
3337 | 3337 | drift1D = drift.copy() |
|
|
3338 | 3338 | if nchan == 2: |
|
|
3339 | 3339 | zonal1D=zonal.copy() |
|
|
3340 | 3340 | snr1D = snr.copy() |
|
|
3341 | 3341 | snr1D = 10*numpy.power(10, 0.1*snr1D) |
|
|
3342 | 3342 | noise1D = noise.copy() |
|
|
3343 | 3343 | noise0 = numpy.nanmedian(noise1D) |
|
|
3344 | 3344 | noise = noise0 + noise0 |
|
|
3345 | 3345 | sw1D = sw.copy() |
|
|
3346 | 3346 | pow0 = snr1D * noise0 + noise0 |
|
|
3347 | 3347 | acf0 = snr1D * noise0 * numpy.exp((-drift1D*nipp*numpy.pi/(1.5e5*1.5))*1j) * (1-0.5*numpy.power(sw1D*nipp*numpy.pi/(1.5e5*1.5),2)) |
|
|
3348 | 3348 | acf0 /= pow0 |
|
|
3349 | 3349 | acf1 = acf0 |
|
|
3350 | 3350 | dt= nint * nipp /1.5e5 |
|
|
3351 | 3351 | |
|
|
3352 | 3352 | if nchan == 2: |
|
|
3353 | 3353 | dccf = pow0 * pow0 * numpy.exp((zonal1D*kd*dt/(height*1e3))*(1j)) |
|
|
3354 | 3354 | else: |
|
|
3355 | 3355 | dccf = numpy.empty(nHei); dccf[:]=numpy.nan # complex? |
|
|
3356 | 3356 | dccf /= pow0 * pow0 |
|
|
3357 | 3357 | sno=(pow0+pow0-noise)/noise |
|
|
3358 | 3358 | |
|
|
3359 | 3359 | # First parameter: Signal to noise ratio and its error |
|
|
3360 | 3360 | sno = numpy.log10(sno) |
|
|
3361 | 3361 | sno10 = 10 * sno |
|
|
3362 | 3362 | dsno = 1.0/numpy.sqrt(nint*navg)*(1+1/sno10) |
|
|
3363 | 3363 | |
|
|
3364 | 3364 | # Second parameter: Vertical Drifts |
|
|
3365 | 3365 | s=numpy.sqrt(numpy.abs(acf0)*numpy.abs(acf1)) |
|
|
3366 | 3366 | sp = s*(1.0 + 1.0/sno10) |
|
|
3367 | 3367 | vzo = -numpy.arctan2(numpy.imag(acf0+acf1),numpy.real(acf0+acf1))* \ |
|
|
3368 | 3368 | 1.5e5*1.5/(nipp*numpy.pi) |
|
|
3369 | 3369 | dvzo = numpy.sqrt(1-sp*sp)*0.338*1.5e5/(numpy.sqrt(nint*navg)*sp*nipp) |
|
|
3370 | 3370 | |
|
|
3371 | 3371 | # Third parameter: Zonal Drifts |
|
|
3372 | 3372 | dt = nint*nipp/1.5e5 |
|
|
3373 | 3373 | ss = numpy.sqrt(numpy.abs(dccf)) |
|
|
3374 | 3374 | vxo = numpy.arctan2(numpy.imag(dccf),numpy.real(dccf))*height*1e3/(kd*dt) |
|
|
3375 | 3375 | dvxo = numpy.sqrt(1.0-ss*ss)*height*1e3/(numpy.sqrt(nint*navg)*ss*kd*dt) |
|
|
3376 | 3376 | |
|
|
3377 | 3377 | npar = 5 |
|
|
3378 | 3378 | par = numpy.empty((npar, nHei)); par[:] = numpy.nan |
|
|
3379 | 3379 | |
|
|
3380 | 3380 | par[0,:] = sno |
|
|
3381 | 3381 | par[1,:] = vzo |
|
|
3382 | 3382 | par[2,:] = vxo |
|
|
3383 | 3383 | par[3,:] = dvzo |
|
|
3384 | 3384 | par[4,:] = dvxo |
|
|
3385 | 3385 | |
|
|
3386 | 3386 | # Segundo filtrado: |
|
|
3387 | 3387 | # RemociΓ³n por altura: Menos de dos datos finitos no son considerados como eco 150Km. |
|
|
3388 | 3388 | clean_par=numpy.empty((npar,nHei)); clean_par[:]=numpy.nan |
|
|
3389 | 3389 | if clean: |
|
|
3390 | 3390 | |
|
|
3391 | 3391 | for p in range(npar): |
|
|
3392 | 3392 | ih=0 |
|
|
3393 | 3393 | while ih < nHei-1: |
|
|
3394 | 3394 | j=ih |
|
|
3395 | 3395 | if numpy.isfinite(snr1D[ih]): |
|
|
3396 | 3396 | while numpy.isfinite(snr1D[j]): |
|
|
3397 | 3397 | j+=1 |
|
|
3398 | 3398 | if j >= nHei: |
|
|
3399 | 3399 | break |
|
|
3400 | 3400 | if j > ih + 1: |
|
|
3401 | 3401 | for k in range(ih,j): |
|
|
3402 | 3402 | clean_par[p][k] = par[p][k] |
|
|
3403 | 3403 | ih = j - 1 |
|
|
3404 | 3404 | ih+=1 |
|
|
3405 | 3405 | else: |
|
|
3406 | 3406 | clean_par[:] = par[:] |
|
|
3407 | 3407 | |
|
|
3408 | 3408 | mad_output = numpy.vstack((clean_par[0,:], clean_par[1,:], clean_par[2,:], clean_par[3,:], clean_par[4,:])) |
|
|
3409 | 3409 | graph = numpy.vstack((clean_par[0,:], clean_par[1,:], clean_par[2,:])) |
|
|
3410 | 3410 | dataOut.data_output = mad_output |
|
|
3411 | 3411 | dataOut.data_graph = graph |
|
|
3412 | 3412 | dataOut.avg_output = avg_par |
|
|
3413 | 3413 | dataOut.utctimeInit = dataOut.utctime |
|
|
3414 | 3414 | dataOut.outputInterval = dataOut.timeInterval |
|
|
3415 | 3415 | |
|
|
3416 | 3416 | dataOut.flagNoData = numpy.all(numpy.isnan(dataOut.data_output[0])) # NAN vectors are not written |
|
|
3417 | 3417 | |
|
|
3418 | 3418 | return dataOut |
|
|
3419 | 3419 | |
|
|
3420 | 3420 | |
|
|
3421 | 3421 | class JULIA_NightVelocities(Operation): |
|
|
3422 | 3422 | ''' |
|
|
3423 | 3423 | Function SpreadFVelocities() |
|
|
3424 | 3424 | |
|
|
3425 | 3425 | Calculates SNL and drifts |
|
|
3426 | 3426 | |
|
|
3427 | 3427 | Type of dataIn: Parameters |
|
|
3428 | 3428 | |
|
|
3429 | 3429 | Configuration Parameters: |
|
|
3430 | 3430 | |
|
|
3431 | 3431 | mymode : (0) Interferometry, |
|
|
3432 | 3432 | (1) Doppler beam swinging. |
|
|
3433 | 3433 | myproc : (0) JULIA_V, |
|
|
3434 | 3434 | (1) JULIA_EW. |
|
|
3435 | 3435 | myantenna : (0) 1/4 antenna, |
|
|
3436 | 3436 | (1) 1/2 antenna. |
|
|
3437 | 3437 | jset : Number of Incoherent integrations. |
|
|
3438 | 3438 | |
|
|
3439 | 3439 | |
|
|
3440 | 3440 | Input: |
|
|
3441 | 3441 | channelList : simple channel list to select e.g. [2,3,7] |
|
|
3442 | 3442 | self.dataOut.data_pre : Spectral data |
|
|
3443 | 3443 | self.dataOut.abscissaList : List of frequencies |
|
|
3444 | 3444 | self.dataOut.noise : Noise level per channel |
|
|
3445 | 3445 | |
|
|
3446 | 3446 | Affected: |
|
|
3447 | 3447 | self.dataOut.moments : Parameters per channel |
|
|
3448 | 3448 | self.dataOut.data_snr : SNR per channel |
|
|
3449 | 3449 | |
|
|
3450 | 3450 | ''' |
|
|
3451 | 3451 | def __init__(self): |
|
|
3452 | 3452 | Operation.__init__(self) |
|
|
3453 | 3453 | |
|
|
3454 | 3454 | def newtotal(self, data): |
|
|
3455 | 3455 | return numpy.nansum(data) |
|
|
3456 | 3456 | |
|
|
3457 | 3457 | def data_filter(self, parm, snrth=-17, swth=20, dopth=500.0, debug=False): |
|
|
3458 | 3458 | |
|
|
3459 | 3459 | Sz0 = parm.shape # Sz0: h,p |
|
|
3460 | 3460 | drift = parm[:,0] |
|
|
3461 | 3461 | sw = 2*parm[:,1] |
|
|
3462 | 3462 | snr = 10*numpy.log10(parm[:,2]) |
|
|
3463 | 3463 | Sz = drift.shape # Sz: h |
|
|
3464 | 3464 | mask = numpy.ones((Sz[0])) |
|
|
3465 | 3465 | th=0 |
|
|
3466 | 3466 | valid=numpy.where(numpy.isfinite(snr)) |
|
|
3467 | 3467 | cvalid = len(valid[0]) |
|
|
3468 | 3468 | if cvalid >= 1: |
|
|
3469 | 3469 | # CΓ‘lculo del ruido promedio de snr para el i-Γ©simo grupo de alturas |
|
|
3470 | 3470 | nbins = int(numpy.max(snr)-numpy.min(snr))+1 # bin size = 1, similar to IDL |
|
|
3471 | 3471 | h = numpy.histogram(snr,bins=nbins) |
|
|
3472 | 3472 | hist = h[0] |
|
|
3473 | 3473 | values = numpy.round_(h[1]) |
|
|
3474 | 3474 | moda = values[numpy.where(hist == numpy.max(hist))] |
|
|
3475 | 3475 | indNoise = numpy.where(numpy.abs(snr - numpy.min(moda)) < 3)[0] |
|
|
3476 | 3476 | |
|
|
3477 | 3477 | noise = snr[indNoise] |
|
|
3478 | 3478 | noise_mean = numpy.sum(noise)/len(noise) |
|
|
3479 | 3479 | # CΓ‘lculo de media de snr |
|
|
3480 | 3480 | med = numpy.median(snr) |
|
|
3481 | 3481 | # Establece el umbral de snr |
|
|
3482 | 3482 | if noise_mean > med + 3: |
|
|
3483 | 3483 | th = med |
|
|
3484 | 3484 | else: |
|
|
3485 | 3485 | th = noise_mean + 3 |
|
|
3486 | 3486 | # Establece mΓ‘scara |
|
|
3487 | 3487 | novalid = numpy.where(snr <= th)[0] |
|
|
3488 | 3488 | mask[novalid] = numpy.nan |
|
|
3489 | 3489 | # Elimina datos que no sobrepasen el umbral: PARAMETRO |
|
|
3490 | 3490 | novalid = numpy.where(snr <= snrth) |
|
|
3491 | 3491 | cnovalid = len(novalid[0]) |
|
|
3492 | 3492 | if cnovalid > 0: |
|
|
3493 | 3493 | mask[novalid] = numpy.nan |
|
|
3494 | 3494 | novalid = numpy.where(numpy.isnan(snr)) |
|
|
3495 | 3495 | cnovalid = len(novalid[0]) |
|
|
3496 | 3496 | if cnovalid > 0: |
|
|
3497 | 3497 | mask[novalid] = numpy.nan |
|
|
3498 | 3498 | # umbral de velocidad |
|
|
3499 | 3499 | if dopth != None: |
|
|
3500 | 3500 | novalid = numpy.where(numpy.logical_or(drift< dopth*(-1), drift > dopth)) |
|
|
3501 | 3501 | cnovalid = len(novalid[0]) |
|
|
3502 | 3502 | if cnovalid > 0: |
|
|
3503 | 3503 | mask[novalid] = numpy.nan |
|
|
3504 | 3504 | if debug: |
|
|
3505 | 3505 | print('Descartados:%i de %i:' %(cnovalid, len(drift))) |
|
|
3506 | 3506 | print('Porcentaje:%3.1f' %(100.0*cnovalid/len(drift))) |
|
|
3507 | 3507 | |
|
|
3508 | 3508 | new_parm = numpy.zeros((Sz0[0],Sz0[1])) |
|
|
3509 | 3509 | for i in range(Sz0[1]): |
|
|
3510 | 3510 | new_parm[:,i] = parm[:,i] * mask |
|
|
3511 | 3511 | |
|
|
3512 | 3512 | return new_parm, mask |
|
|
3513 | 3513 | |
|
|
3514 | 3514 | |
|
|
3515 | 3515 | def run(self, dataOut, zenith, zenithCorrection, mymode=1, dbs_sel=0, myproc=0, myantenna=0, jset=None, clean=False): |
|
|
3516 | 3516 | |
|
|
3517 | 3517 | |
|
|
3518 | 3518 | dataOut.lat=-11.95 |
|
|
3519 | 3519 | dataOut.lon=-76.87 |
|
|
3520 | 3520 | mode=mymode |
|
|
3521 | 3521 | proc=myproc |
|
|
3522 | 3522 | antenna=myantenna |
|
|
3523 | 3523 | nci=dataOut.nCohInt |
|
|
3524 | 3524 | nptsfft=dataOut.nProfiles |
|
|
3525 | 3525 | navg= 3 if jset is None else jset |
|
|
3526 | 3526 | nint=dataOut.nIncohInt//navg |
|
|
3527 | 3527 | navg1=dataOut.nProfiles * nint * navg |
|
|
3528 | 3528 | tau1=dataOut.ippSeconds |
|
|
3529 | 3529 | nipp=dataOut.radarControllerHeaderObj.ipp |
|
|
3530 | 3530 | jlambda=6 |
|
|
3531 | 3531 | kd=213.6 |
|
|
3532 | 3532 | hei=dataOut.heightList.copy() |
|
|
3533 | 3533 | |
|
|
3534 | 3534 | nCh=dataOut.spcpar.shape[0] |
|
|
3535 | 3535 | nHei=dataOut.spcpar.shape[1] |
|
|
3536 | 3536 | nParam=dataOut.spcpar.shape[2] |
|
|
3537 | 3537 | |
|
|
3538 | 3538 | parm = numpy.zeros((nCh,nHei,nParam)) |
|
|
3539 | 3539 | parm[:] = dataOut.spcpar[:] |
|
|
3540 | 3540 | mask=numpy.ones(nHei) |
|
|
3541 | 3541 | mask0=mask.copy() |
|
|
3542 | 3542 | # Primer filtrado: Umbral de SNR |
|
|
3543 | 3543 | for i in range(nCh): |
|
|
3544 | 3544 | parm[i,:,:], mask = self.data_filter(parm[i,:,:], snrth = 0.1) # umbral 0.1 filtra seΓ±al que no corresponde a ESF, para interferometrΓa usar -17dB |
|
|
3545 | 3545 | mask0 *= mask |
|
|
3546 | 3546 | |
|
|
3547 | 3547 | ccf_results=numpy.transpose(dataOut.ccfpar,(2,1,0)) |
|
|
3548 | 3548 | |
|
|
3549 | 3549 | for i in range(3): |
|
|
3550 | 3550 | ccf_results[i,:,0] *= mask0 |
|
|
3551 | 3551 | |
|
|
3552 | 3552 | zenith = numpy.array(zenith) |
|
|
3553 | 3553 | zenith -= zenithCorrection |
|
|
3554 | 3554 | zenith *= numpy.pi/180 |
|
|
3555 | 3555 | alpha = zenith[0] |
|
|
3556 | 3556 | beta = zenith[1] |
|
|
3557 | 3557 | |
|
|
3558 | 3558 | w_w = parm[0,:,0] |
|
|
3559 | 3559 | w_e = parm[1,:,0] |
|
|
3560 | 3560 | |
|
|
3561 | 3561 | if mode==1: |
|
|
3562 | 3562 | # Vertical and zonal calculation |
|
|
3563 | 3563 | sinB_A = numpy.sin(beta)*numpy.cos(alpha) - numpy.sin(alpha)* numpy.cos(beta) |
|
|
3564 | 3564 | w = -(w_w * numpy.sin(beta) - w_e * numpy.sin(alpha))/ sinB_A |
|
|
3565 | 3565 | u = (w_w * numpy.cos(beta) - w_e * numpy.cos(alpha))/ sinB_A |
|
|
3566 | 3566 | |
|
|
3567 | 3567 | #Noise |
|
|
3568 | 3568 | n0 = parm[0,:,3] |
|
|
3569 | 3569 | n1 = parm[1,:,3] |
|
|
3570 | 3570 | jn0_1 = numpy.nanmedian(n0) |
|
|
3571 | 3571 | jn0_2 = numpy.nanmean(n0) |
|
|
3572 | 3572 | jn1_1 = numpy.nanmedian(n1) |
|
|
3573 | 3573 | jn1_2 = numpy.nanmean(n1) |
|
|
3574 | 3574 | noise0 = jn0_2 if numpy.abs(jn0_1-jn0_2)/(jn0_1+jn0_2) <= 0.1 else jn0_1 |
|
|
3575 | 3575 | noise1 = jn1_2 if numpy.abs(jn1_1-jn1_2)/(jn1_1+jn1_2) <= 0.1 else jn1_1 |
|
|
3576 | 3576 | |
|
|
3577 | 3577 | noise = noise0 + noise0 if mode == 1 else noise0 + noise1 |
|
|
3578 | 3578 | |
|
|
3579 | 3579 | #Power |
|
|
3580 | 3580 | apow1 = (parm[0,:,2]/numpy.sqrt(nint))*noise0 + n0 |
|
|
3581 | 3581 | apow2 = (parm[1,:,2]/numpy.sqrt(nint))*noise1 + n1 |
|
|
3582 | 3582 | |
|
|
3583 | 3583 | #SNR SNR=Detectability/ SQRT(nint) or (Pow-Noise)/Noise |
|
|
3584 | 3584 | s_n0 = (apow1 - noise0)/noise0 |
|
|
3585 | 3585 | s_n1 = (apow2 - noise1)/noise1 |
|
|
3586 | 3586 | |
|
|
3587 | 3587 | swCH0 = parm[0,:,1] |
|
|
3588 | 3588 | swCH1 = parm[1,:,1] |
|
|
3589 | 3589 | |
|
|
3590 | 3590 | if mode == 1: |
|
|
3591 | 3591 | aacf1=(1-numpy.square(tau1)*numpy.square(4*numpy.pi/jlambda*swCH0)/2)* \ |
|
|
3592 | 3592 | numpy.exp(-4*numpy.pi/jlambda*w*tau1*1j)* \ |
|
|
3593 | 3593 | apow1 |
|
|
3594 | 3594 | aacf2=(1-numpy.square(tau1)*numpy.square(4*numpy.pi/jlambda*swCH1)/2)* \ |
|
|
3595 | 3595 | numpy.exp(-4*numpy.pi/jlambda*w*tau1*1j)* \ |
|
|
3596 | 3596 | apow2 |
|
|
3597 | 3597 | dccf_0=numpy.zeros(nHei, dtype=complex) |
|
|
3598 | 3598 | |
|
|
3599 | 3599 | else: |
|
|
3600 | 3600 | aacf1=(1-numpy.square(tau1)*numpy.square(4*numpy.pi/jlambda*swCH0)/2)* \ |
|
|
3601 | 3601 | numpy.exp(4*numpy.pi/jlambda*w_w*tau1*1j)* \ |
|
|
3602 | 3602 | apow1 |
|
|
3603 | 3603 | aacf2=(1-numpy.square(tau1)*numpy.square(4*numpy.pi/jlambda*swCH1)/2)* \ |
|
|
3604 | 3604 | numpy.exp(4*numpy.pi/jlambda*w_e*tau1*1j)* \ |
|
|
3605 | 3605 | apow2 |
|
|
3606 | 3606 | dccf_0=numpy.power(ccf_results[0,:,0],2)*apow1*apow2* \ |
|
|
3607 | 3607 | numpy.exp( \ |
|
|
3608 | 3608 | ( \ |
|
|
3609 | 3609 | (1+1*(antenna==1))* \ |
|
|
3610 | 3610 | (-1+2*(proc == 1))* \ |
|
|
3611 | 3611 | ccf_results[2,:,0] \ |
|
|
3612 | 3612 | )*1j) |
|
|
3613 | 3613 | |
|
|
3614 | 3614 | nsamp=len(hei) |
|
|
3615 | 3615 | pow0 = numpy.empty(nsamp); pow0[:] = numpy.nan |
|
|
3616 | 3616 | pow1 = numpy.empty(nsamp); pow1[:] = numpy.nan |
|
|
3617 | 3617 | acf0 = numpy.empty(nsamp, dtype=complex); acf0[:] = numpy.nan |
|
|
3618 | 3618 | acf1 = numpy.empty(nsamp, dtype=complex); acf1[:] = numpy.nan |
|
|
3619 | 3619 | dccf = numpy.empty(nsamp, dtype=complex); dccf[:] = numpy.nan |
|
|
3620 | 3620 | dop0 = numpy.empty(nsamp); dop0[:] = numpy.nan |
|
|
3621 | 3621 | dop1 = numpy.empty(nsamp); dop1[:] = numpy.nan |
|
|
3622 | 3622 | p_w = numpy.empty(nsamp); p_w[:] = numpy.nan |
|
|
3623 | 3623 | p_u = numpy.empty(nsamp); p_u[:] = numpy.nan |
|
|
3624 | 3624 | |
|
|
3625 | 3625 | if mode == 0 or (mode == 1 and dbs_sel == 0): |
|
|
3626 | 3626 | ih=0 |
|
|
3627 | 3627 | while ih < nsamp-10: |
|
|
3628 | 3628 | j=ih |
|
|
3629 | 3629 | if numpy.isfinite(s_n0[ih]) and numpy.isfinite(s_n1[ih]): |
|
|
3630 | 3630 | while numpy.isfinite(s_n0[j]) and numpy.isfinite(s_n1[j]): |
|
|
3631 | 3631 | j+=1 |
|
|
3632 | 3632 | if j > ih + 2: |
|
|
3633 | 3633 | for k in range(ih,j): |
|
|
3634 | 3634 | pow0[k] = apow1[k] |
|
|
3635 | 3635 | pow1[k] = apow2[k] |
|
|
3636 | 3636 | acf0[k] = aacf1[k] |
|
|
3637 | 3637 | acf1[k] = aacf2[k] |
|
|
3638 | 3638 | dccf[k] = dccf_0[k] |
|
|
3639 | 3639 | ih = j - 1 |
|
|
3640 | 3640 | ih+=1 |
|
|
3641 | 3641 | else: |
|
|
3642 | 3642 | ih=0 |
|
|
3643 | 3643 | while ih < nsamp-10: |
|
|
3644 | 3644 | j=ih |
|
|
3645 | 3645 | if numpy.isfinite(s_n0[ih]): |
|
|
3646 | 3646 | while numpy.isfinite(s_n0[j]) and j < nsamp-10: |
|
|
3647 | 3647 | j+=1 |
|
|
3648 | 3648 | #if j > ih + 6: |
|
|
3649 | 3649 | if j > ih + 2: |
|
|
3650 | 3650 | #if j > ih + 3: |
|
|
3651 | 3651 | for k in range(ih,j): |
|
|
3652 | 3652 | pow0[k] = apow1[k] |
|
|
3653 | 3653 | #acf0[k] = aacf1[k] |
|
|
3654 | 3654 | #dccf[k] = dccf_0[k] |
|
|
3655 | 3655 | p_w[k] = w[k] |
|
|
3656 | 3656 | dop0[k] = w_w[k] |
|
|
3657 | 3657 | ih = j - 1 |
|
|
3658 | 3658 | ih+=1 |
|
|
3659 | 3659 | ih=0 |
|
|
3660 | 3660 | while ih < nsamp-10: |
|
|
3661 | 3661 | j=ih |
|
|
3662 | 3662 | if numpy.isfinite(s_n1[ih]): |
|
|
3663 | 3663 | while numpy.isfinite(s_n1[j]) and j < nsamp-10: |
|
|
3664 | 3664 | j+=1 |
|
|
3665 | 3665 | #if j > ih + 6: |
|
|
3666 | 3666 | if j > ih + 2: |
|
|
3667 | 3667 | #if j > ih + 3: |
|
|
3668 | 3668 | for k in range(ih,j): |
|
|
3669 | 3669 | pow1[k] = apow2[k] |
|
|
3670 | 3670 | #acf1[k] = aacf2[k] |
|
|
3671 | 3671 | p_u[k] = u[k] |
|
|
3672 | 3672 | dop1[k] = w_e[k] |
|
|
3673 | 3673 | ih = j - 1 |
|
|
3674 | 3674 | ih+=1 |
|
|
3675 | 3675 | |
|
|
3676 | 3676 | acf0 = numpy.zeros(nsamp, dtype=complex) |
|
|
3677 | 3677 | acf1 = numpy.zeros(nsamp, dtype=complex) |
|
|
3678 | 3678 | dccf = numpy.zeros(nsamp, dtype=complex) |
|
|
3679 | 3679 | |
|
|
3680 | 3680 | acf0 /= pow0 |
|
|
3681 | 3681 | acf1 /= pow1 |
|
|
3682 | 3682 | dccf /= pow0 * pow1 |
|
|
3683 | 3683 | |
|
|
3684 | 3684 | if mode == 0 or (mode == 1 and dbs_sel == 0): |
|
|
3685 | 3685 | sno=(pow0+pow1-noise)/noise |
|
|
3686 | 3686 | # First parameter: Signal to noise ratio and its error |
|
|
3687 | 3687 | sno=numpy.log10(sno) |
|
|
3688 | 3688 | dsno=1.0/numpy.sqrt(nint*navg)*(1+1/sno) |
|
|
3689 | 3689 | # Second parameter: Vertical Drifts |
|
|
3690 | 3690 | s=numpy.sqrt(numpy.abs(acf0)*numpy.abs(acf1)) |
|
|
3691 | 3691 | ind=numpy.where(numpy.abs(s)>=1.0) |
|
|
3692 | 3692 | if numpy.size(ind)>0: |
|
|
3693 | 3693 | s[ind]=numpy.sqrt(0.9999) |
|
|
3694 | 3694 | sp=s*(1.0 + 1.0/sno) |
|
|
3695 | 3695 | vzo=-numpy.arctan2(numpy.imag(acf0+acf1),numpy.real(acf0+acf1))* \ |
|
|
3696 | 3696 | 1.5e5*1.5/(nipp*numpy.pi) |
|
|
3697 | 3697 | dvzo=numpy.sqrt(1-sp*sp)*0.338*1.5e5/(numpy.sqrt(nint*navg)*sp*nipp) |
|
|
3698 | 3698 | ind=numpy.where(dvzo<=0.1) |
|
|
3699 | 3699 | if numpy.size(ind)>0: |
|
|
3700 | 3700 | dvzo[ind]=0.1 |
|
|
3701 | 3701 | # Third parameter: Zonal Drifts |
|
|
3702 | 3702 | dt=nint*nipp/1.5e5 |
|
|
3703 | 3703 | ss=numpy.sqrt(numpy.abs(dccf)) |
|
|
3704 | 3704 | ind=numpy.where(ss>=1.0) |
|
|
3705 | 3705 | if numpy.size(ind)>0: |
|
|
3706 | 3706 | ss[ind]=numpy.sqrt(0.99999) |
|
|
3707 | 3707 | ind=numpy.where(ss<=0.1) |
|
|
3708 | 3708 | if numpy.size(ind)>0: |
|
|
3709 | 3709 | ss[ind]=numpy.sqrt(0.1) |
|
|
3710 | 3710 | vxo=numpy.arctan2(numpy.imag(dccf),numpy.real(dccf))*hei*1e3/(kd*dt) |
|
|
3711 | 3711 | dvxo=numpy.sqrt(1.0-ss*ss)*hei*1e3/(numpy.sqrt(nint*navg)*ss*kd*dt) |
|
|
3712 | 3712 | ind=numpy.where(dvxo<=0.1) |
|
|
3713 | 3713 | if numpy.size(ind)>0: |
|
|
3714 | 3714 | dvxo[ind]=0.1 |
|
|
3715 | 3715 | else: |
|
|
3716 | 3716 | sno0=(pow0-noise0)/noise0 |
|
|
3717 | 3717 | sno1=(pow1-noise1)/noise1 |
|
|
3718 | 3718 | |
|
|
3719 | 3719 | # First parameter: Signal to noise ratio and its error |
|
|
3720 | 3720 | sno0=numpy.log10(sno0) |
|
|
3721 | 3721 | dsno0=1.0/numpy.sqrt(nint*navg)*(1+1/sno0) |
|
|
3722 | 3722 | sno1=numpy.log10(sno1) |
|
|
3723 | 3723 | dsno1=1.0/numpy.sqrt(nint*navg)*(1+1/sno1) |
|
|
3724 | 3724 | |
|
|
3725 | 3725 | npar=6 |
|
|
3726 | 3726 | par = numpy.empty((npar, nHei)); par[:] = numpy.nan |
|
|
3727 | 3727 | |
|
|
3728 | 3728 | if mode == 0: |
|
|
3729 | 3729 | par[0,:] = sno |
|
|
3730 | 3730 | par[1,:] = vxo |
|
|
3731 | 3731 | par[2,:] = dvxo |
|
|
3732 | 3732 | par[3,:] = vzo |
|
|
3733 | 3733 | par[4,:] = dvzo |
|
|
3734 | 3734 | |
|
|
3735 | 3735 | elif mode == 1 and dbs_sel == 0: |
|
|
3736 | 3736 | par[0,:] = sno |
|
|
3737 | 3737 | par[1,:] = vzo |
|
|
3738 | 3738 | else: |
|
|
3739 | 3739 | par[0,:] = sno0 |
|
|
3740 | 3740 | par[1,:] = sno1 |
|
|
3741 | 3741 | par[2,:] = dop0 |
|
|
3742 | 3742 | par[3,:] = dop1 |
|
|
3743 | 3743 | #par[4,:] = p_w |
|
|
3744 | 3744 | #par[5,:] = p_u |
|
|
3745 | 3745 | |
|
|
3746 | 3746 | if mode == 0: |
|
|
3747 | 3747 | winds = numpy.vstack((par[0,:], par[1,:], par[2,:], par[3,:], par[4,:])) |
|
|
3748 | 3748 | elif mode == 1 and dbs_sel == 0: |
|
|
3749 | 3749 | winds = numpy.vstack((par[0,:], par[1,:])) |
|
|
3750 | 3750 | else: |
|
|
3751 | 3751 | winds = numpy.vstack((par[0,:], par[1,:], par[2,:], par[3,:])) |
|
|
3752 | 3752 | |
|
|
3753 | 3753 | dataOut.data_output = winds |
|
|
3754 | 3754 | dataOut.data_snr = par[0,:] |
|
|
3755 | 3755 | |
|
|
3756 | 3756 | dataOut.utctimeInit = dataOut.utctime |
|
|
3757 | 3757 | dataOut.outputInterval = dataOut.timeInterval |
|
|
3758 | 3758 | |
|
|
3759 | 3759 | aux1= numpy.all(numpy.isnan(dataOut.data_output[0])) # NAN vectors are not written |
|
|
3760 | 3760 | aux2= numpy.all(numpy.isnan(dataOut.data_output[1])) # NAN vectors are not written |
|
|
3761 | 3761 | dataOut.flagNoData = aux1 or aux2 |
|
|
3762 | 3762 | |
|
|
3763 | 3763 | return dataOut |
|
|
3764 | 3764 | |
|
|
3765 | 3765 | class SALags(Operation): |
|
|
3766 | 3766 | ''' |
|
|
3767 | 3767 | Function GetMoments() |
|
|
3768 | 3768 | |
|
|
3769 | 3769 | Input: |
|
|
3770 | 3770 | self.dataOut.data_pre |
|
|
3771 | 3771 | self.dataOut.abscissaList |
|
|
3772 | 3772 | self.dataOut.noise |
|
|
3773 | 3773 | self.dataOut.normFactor |
|
|
3774 | 3774 | self.dataOut.data_snr |
|
|
3775 | 3775 | self.dataOut.groupList |
|
|
3776 | 3776 | self.dataOut.nChannels |
|
|
3777 | 3777 | |
|
|
3778 | 3778 | Affected: |
|
|
3779 | 3779 | self.dataOut.data_param |
|
|
3780 | 3780 | |
|
|
3781 | 3781 | ''' |
|
|
3782 | 3782 | def run(self, dataOut): |
|
|
3783 | 3783 | data_acf = dataOut.data_pre[0] |
|
|
3784 | 3784 | data_ccf = dataOut.data_pre[1] |
|
|
3785 | 3785 | normFactor_acf = dataOut.normFactor[0] |
|
|
3786 | 3786 | normFactor_ccf = dataOut.normFactor[1] |
|
|
3787 | 3787 | pairs_acf = dataOut.groupList[0] |
|
|
3788 | 3788 | pairs_ccf = dataOut.groupList[1] |
|
|
3789 | 3789 | |
|
|
3790 | 3790 | nHeights = dataOut.nHeights |
|
|
3791 | 3791 | absc = dataOut.abscissaList |
|
|
3792 | 3792 | noise = dataOut.noise |
|
|
3793 | 3793 | SNR = dataOut.data_snr |
|
|
3794 | 3794 | nChannels = dataOut.nChannels |
|
|
3795 | 3795 | # pairsList = dataOut.groupList |
|
|
3796 | 3796 | # pairsAutoCorr, pairsCrossCorr = self.__getPairsAutoCorr(pairsList, nChannels) |
|
|
3797 | 3797 | |
|
|
3798 | 3798 | for l in range(len(pairs_acf)): |
|
|
3799 | 3799 | data_acf[l,:,:] = data_acf[l,:,:]/normFactor_acf[l,:] |
|
|
3800 | 3800 | |
|
|
3801 | 3801 | for l in range(len(pairs_ccf)): |
|
|
3802 | 3802 | data_ccf[l,:,:] = data_ccf[l,:,:]/normFactor_ccf[l,:] |
|
|
3803 | 3803 | |
|
|
3804 | 3804 | dataOut.data_param = numpy.zeros((len(pairs_ccf)*2 + 1, nHeights)) |
|
|
3805 | 3805 | dataOut.data_param[:-1,:] = self.__calculateTaus(data_acf, data_ccf, absc) |
|
|
3806 | 3806 | dataOut.data_param[-1,:] = self.__calculateLag1Phase(data_acf, absc) |
|
|
3807 | 3807 | return |
|
|
3808 | 3808 | |
|
|
3809 | 3809 | # def __getPairsAutoCorr(self, pairsList, nChannels): |
|
|
3810 | 3810 | # |
|
|
3811 | 3811 | # pairsAutoCorr = numpy.zeros(nChannels, dtype = 'int')*numpy.nan |
|
|
3812 | 3812 | # |
|
|
3813 | 3813 | # for l in range(len(pairsList)): |
|
|
3814 | 3814 | # firstChannel = pairsList[l][0] |
|
|
3815 | 3815 | # secondChannel = pairsList[l][1] |
|
|
3816 | 3816 | # |
|
|
3817 | 3817 | # #Obteniendo pares de Autocorrelacion |
|
|
3818 | 3818 | # if firstChannel == secondChannel: |
|
|
3819 | 3819 | # pairsAutoCorr[firstChannel] = int(l) |
|
|
3820 | 3820 | # |
|
|
3821 | 3821 | # pairsAutoCorr = pairsAutoCorr.astype(int) |
|
|
3822 | 3822 | # |
|
|
3823 | 3823 | # pairsCrossCorr = range(len(pairsList)) |
|
|
3824 | 3824 | # pairsCrossCorr = numpy.delete(pairsCrossCorr,pairsAutoCorr) |
|
|
3825 | 3825 | # |
|
|
3826 | 3826 | # return pairsAutoCorr, pairsCrossCorr |
|
|
3827 | 3827 | |
|
|
3828 | 3828 | def __calculateTaus(self, data_acf, data_ccf, lagRange): |
|
|
3829 | 3829 | |
|
|
3830 | 3830 | lag0 = data_acf.shape[1]/2 |
|
|
3831 | 3831 | #Funcion de Autocorrelacion |
|
|
3832 | 3832 | mean_acf = stats.nanmean(data_acf, axis = 0) |
|
|
3833 | 3833 | |
|
|
3834 | 3834 | #Obtencion Indice de TauCross |
|
|
3835 | 3835 | ind_ccf = data_ccf.argmax(axis = 1) |
|
|
3836 | 3836 | #Obtencion Indice de TauAuto |
|
|
3837 | 3837 | ind_acf = numpy.zeros(ind_ccf.shape,dtype = 'int') |
|
|
3838 | 3838 | ccf_lag0 = data_ccf[:,lag0,:] |
|
|
3839 | 3839 | |
|
|
3840 | 3840 | for i in range(ccf_lag0.shape[0]): |
|
|
3841 | 3841 | ind_acf[i,:] = numpy.abs(mean_acf - ccf_lag0[i,:]).argmin(axis = 0) |
|
|
3842 | 3842 | |
|
|
3843 | 3843 | #Obtencion de TauCross y TauAuto |
|
|
3844 | 3844 | tau_ccf = lagRange[ind_ccf] |
|
|
3845 | 3845 | tau_acf = lagRange[ind_acf] |
|
|
3846 | 3846 | |
|
|
3847 | 3847 | Nan1, Nan2 = numpy.where(tau_ccf == lagRange[0]) |
|
|
3848 | 3848 | |
|
|
3849 | 3849 | tau_ccf[Nan1,Nan2] = numpy.nan |
|
|
3850 | 3850 | tau_acf[Nan1,Nan2] = numpy.nan |
|
|
3851 | 3851 | tau = numpy.vstack((tau_ccf,tau_acf)) |
|
|
3852 | 3852 | |
|
|
3853 | 3853 | return tau |
|
|
3854 | 3854 | |
|
|
3855 | 3855 | def __calculateLag1Phase(self, data, lagTRange): |
|
|
3856 | 3856 | data1 = stats.nanmean(data, axis = 0) |
|
|
3857 | 3857 | lag1 = numpy.where(lagTRange == 0)[0][0] + 1 |
|
|
3858 | 3858 | |
|
|
3859 | 3859 | phase = numpy.angle(data1[lag1,:]) |
|
|
3860 | 3860 | |
|
|
3861 | 3861 | return phase |
|
|
3862 | 3862 | |
|
|
3863 | 3863 | def fit_func( x, a0, a1, a2): #, a3, a4, a5): |
|
|
3864 | 3864 | z = (x - a1) / a2 |
|
|
3865 | 3865 | y = a0 * numpy.exp(-z**2 / a2) #+ a3 + a4 * x + a5 * x**2 |
|
|
3866 | 3866 | return y |
|
|
3867 | 3867 | |
|
|
3868 | 3868 | |
|
|
3869 | 3869 | class SpectralFitting(Operation): |
|
|
3870 | 3870 | ''' |
|
|
3871 | 3871 | Function GetMoments() |
|
|
3872 | 3872 | |
|
|
3873 | 3873 | Input: |
|
|
3874 | 3874 | Output: |
|
|
3875 | 3875 | Variables modified: |
|
|
3876 | 3876 | ''' |
|
|
3877 | 3877 | def __calculateMoments(self, oldspec, oldfreq, n0, nicoh = None, graph = None, smooth = None, type1 = None, fwindow = None, snrth = None, dc = None, aliasing = None, oldfd = None, wwauto = None): |
|
|
3878 | 3878 | |
|
|
3879 | 3879 | if (nicoh is None): nicoh = 1 |
|
|
3880 | 3880 | if (graph is None): graph = 0 |
|
|
3881 | 3881 | if (smooth is None): smooth = 0 |
|
|
3882 | 3882 | elif (self.smooth < 3): smooth = 0 |
|
|
3883 | 3883 | |
|
|
3884 | 3884 | if (type1 is None): type1 = 0 |
|
|
3885 | 3885 | if (fwindow is None): fwindow = numpy.zeros(oldfreq.size) + 1 |
|
|
3886 | 3886 | if (snrth is None): snrth = -3 |
|
|
3887 | 3887 | if (dc is None): dc = 0 |
|
|
3888 | 3888 | if (aliasing is None): aliasing = 0 |
|
|
3889 | 3889 | if (oldfd is None): oldfd = 0 |
|
|
3890 | 3890 | if (wwauto is None): wwauto = 0 |
|
|
3891 | 3891 | |
|
|
3892 | 3892 | if (n0 < 1.e-20): n0 = 1.e-20 |
|
|
3893 | 3893 | |
|
|
3894 | 3894 | freq = oldfreq |
|
|
3895 | 3895 | vec_power = numpy.zeros(oldspec.shape[1]) |
|
|
3896 | 3896 | vec_fd = numpy.zeros(oldspec.shape[1]) |
|
|
3897 | 3897 | vec_w = numpy.zeros(oldspec.shape[1]) |
|
|
3898 | 3898 | vec_snr = numpy.zeros(oldspec.shape[1]) |
|
|
3899 | 3899 | |
|
|
3900 | 3900 | oldspec = numpy.ma.masked_invalid(oldspec) |
|
|
3901 | 3901 | |
|
|
3902 | 3902 | for ind in range(oldspec.shape[1]): |
|
|
3903 | 3903 | |
|
|
3904 | 3904 | spec = oldspec[:,ind] |
|
|
3905 | 3905 | aux = spec*fwindow |
|
|
3906 | 3906 | max_spec = aux.max() |
|
|
3907 | 3907 | m = list(aux).index(max_spec) |
|
|
3908 | 3908 | |
|
|
3909 | 3909 | #Smooth |
|
|
3910 | 3910 | if (smooth == 0): spec2 = spec |
|
|
3911 | 3911 | else: spec2 = scipy.ndimage.filters.uniform_filter1d(spec,size=smooth) |
|
|
3912 | 3912 | |
|
|
3913 | 3913 | # Calculo de Momentos |
|
|
3914 | 3914 | bb = spec2[list(range(m,spec2.size))] |
|
|
3915 | 3915 | bb = (bb<n0).nonzero() |
|
|
3916 | 3916 | bb = bb[0] |
|
|
3917 | 3917 | |
|
|
3918 | 3918 | ss = spec2[list(range(0,m + 1))] |
|
|
3919 | 3919 | ss = (ss<n0).nonzero() |
|
|
3920 | 3920 | ss = ss[0] |
|
|
3921 | 3921 | |
|
|
3922 | 3922 | if (bb.size == 0): |
|
|
3923 | 3923 | bb0 = spec.size - 1 - m |
|
|
3924 | 3924 | else: |
|
|
3925 | 3925 | bb0 = bb[0] - 1 |
|
|
3926 | 3926 | if (bb0 < 0): |
|
|
3927 | 3927 | bb0 = 0 |
|
|
3928 | 3928 | |
|
|
3929 | 3929 | if (ss.size == 0): ss1 = 1 |
|
|
3930 | 3930 | else: ss1 = max(ss) + 1 |
|
|
3931 | 3931 | |
|
|
3932 | 3932 | if (ss1 > m): ss1 = m |
|
|
3933 | 3933 | |
|
|
3934 | 3934 | valid = numpy.asarray(list(range(int(m + bb0 - ss1 + 1)))) + ss1 |
|
|
3935 | 3935 | power = ((spec2[valid] - n0)*fwindow[valid]).sum() |
|
|
3936 | 3936 | fd = ((spec2[valid]- n0)*freq[valid]*fwindow[valid]).sum()/power |
|
|
3937 | 3937 | w = math.sqrt(((spec2[valid] - n0)*fwindow[valid]*(freq[valid]- fd)**2).sum()/power) |
|
|
3938 | 3938 | snr = (spec2.mean()-n0)/n0 |
|
|
3939 | 3939 | |
|
|
3940 | 3940 | if (snr < 1.e-20) : |
|
|
3941 | 3941 | snr = 1.e-20 |
|
|
3942 | 3942 | |
|
|
3943 | 3943 | vec_power[ind] = power |
|
|
3944 | 3944 | vec_fd[ind] = fd |
|
|
3945 | 3945 | vec_w[ind] = w |
|
|
3946 | 3946 | vec_snr[ind] = snr |
|
|
3947 | 3947 | |
|
|
3948 | 3948 | moments = numpy.vstack((vec_snr, vec_power, vec_fd, vec_w)) |
|
|
3949 | 3949 | return moments |
|
|
3950 | 3950 | |
|
|
3951 | 3951 | #def __DiffCoherent(self,snrth, spectra, cspectra, nProf, heights,nChan, nHei, nPairs, channels, noise, crosspairs): |
|
|
3952 | 3952 | def __DiffCoherent(self, spectra, cspectra, dataOut, noise, snrth, coh_th, hei_th): |
|
|
3953 | 3953 | |
|
|
3954 | 3954 | import matplotlib.pyplot as plt |
|
|
3955 | 3955 | nProf = dataOut.nProfiles |
|
|
3956 | 3956 | heights = dataOut.heightList |
|
|
3957 | 3957 | nHei = len(heights) |
|
|
3958 | 3958 | channels = dataOut.channelList |
|
|
3959 | 3959 | nChan = len(channels) |
|
|
3960 | 3960 | crosspairs = dataOut.groupList |
|
|
3961 | 3961 | nPairs = len(crosspairs) |
|
|
3962 | 3962 | #Separar espectros incoherentes de coherentes snr > 20 dB' |
|
|
3963 | 3963 | snr_th = 10**(snrth/10.0) |
|
|
3964 | 3964 | my_incoh_spectra = numpy.zeros([nChan, nProf,nHei], dtype='float') |
|
|
3965 | 3965 | my_incoh_cspectra = numpy.zeros([nPairs,nProf, nHei], dtype='complex') |
|
|
3966 | 3966 | my_incoh_aver = numpy.zeros([nChan, nHei]) |
|
|
3967 | 3967 | my_coh_aver = numpy.zeros([nChan, nHei]) |
|
|
3968 | 3968 | |
|
|
3969 | 3969 | coh_spectra = numpy.zeros([nChan, nProf, nHei], dtype='float') |
|
|
3970 | 3970 | coh_cspectra = numpy.zeros([nPairs, nProf, nHei], dtype='complex') |
|
|
3971 | 3971 | coh_aver = numpy.zeros([nChan, nHei]) |
|
|
3972 | 3972 | |
|
|
3973 | 3973 | incoh_spectra = numpy.zeros([nChan, nProf, nHei], dtype='float') |
|
|
3974 | 3974 | incoh_cspectra = numpy.zeros([nPairs, nProf, nHei], dtype='complex') |
|
|
3975 | 3975 | incoh_aver = numpy.zeros([nChan, nHei]) |
|
|
3976 | 3976 | power = numpy.sum(spectra, axis=1) |
|
|
3977 | 3977 | |
|
|
3978 | 3978 | if coh_th == None : coh_th = numpy.array([0.75,0.65,0.15]) # 0.65 |
|
|
3979 | 3979 | if hei_th == None : hei_th = numpy.array([60,300,650]) |
|
|
3980 | 3980 | for ic in range(2): |
|
|
3981 | 3981 | pair = crosspairs[ic] |
|
|
3982 | 3982 | #si el SNR es mayor que el SNR threshold los datos se toman coherentes |
|
|
3983 | 3983 | s_n0 = power[pair[0],:]/noise[pair[0]] |
|
|
3984 | 3984 | s_n1 = power[pair[1],:]/noise[pair[1]] |
|
|
3985 | 3985 | |
|
|
3986 | 3986 | valid1 =(s_n0>=snr_th).nonzero() |
|
|
3987 | 3987 | valid2 = (s_n1>=snr_th).nonzero() |
|
|
3988 | 3988 | #valid = valid2 + valid1 #numpy.concatenate((valid1,valid2), axis=None) |
|
|
3989 | 3989 | valid1 = numpy.array(valid1[0]) |
|
|
3990 | 3990 | valid2 = numpy.array(valid2[0]) |
|
|
3991 | 3991 | valid = valid1 |
|
|
3992 | 3992 | for iv in range(len(valid2)): |
|
|
3993 | 3993 | #for ivv in range(len(valid1)) : |
|
|
3994 | 3994 | indv = numpy.array((valid1 == valid2[iv]).nonzero()) |
|
|
3995 | 3995 | if len(indv[0]) == 0 : |
|
|
3996 | 3996 | valid = numpy.concatenate((valid,valid2[iv]), axis=None) |
|
|
3997 | 3997 | if len(valid)>0: |
|
|
3998 | 3998 | my_coh_aver[pair[0],valid]=1 |
|
|
3999 | 3999 | my_coh_aver[pair[1],valid]=1 |
|
|
4000 | 4000 | # si la coherencia es mayor a la coherencia threshold los datos se toman |
|
|
4001 | 4001 | #print my_coh_aver[0,:] |
|
|
4002 | 4002 | coh = numpy.squeeze(numpy.nansum(cspectra[ic,:,:], axis=0)/numpy.sqrt(numpy.nansum(spectra[pair[0],:,:], axis=0)*numpy.nansum(spectra[pair[1],:,:], axis=0))) |
|
|
4003 | 4003 | #print('coh',numpy.absolute(coh)) |
|
|
4004 | 4004 | for ih in range(len(hei_th)): |
|
|
4005 | 4005 | hvalid = (heights>hei_th[ih]).nonzero() |
|
|
4006 | 4006 | hvalid = hvalid[0] |
|
|
4007 | 4007 | if len(hvalid)>0: |
|
|
4008 | 4008 | valid = (numpy.absolute(coh[hvalid])>coh_th[ih]).nonzero() |
|
|
4009 | 4009 | valid = valid[0] |
|
|
4010 | 4010 | #print('hvalid:',hvalid) |
|
|
4011 | 4011 | #print('valid', valid) |
|
|
4012 | 4012 | if len(valid)>0: |
|
|
4013 | 4013 | my_coh_aver[pair[0],hvalid[valid]] =1 |
|
|
4014 | 4014 | my_coh_aver[pair[1],hvalid[valid]] =1 |
|
|
4015 | 4015 | |
|
|
4016 | 4016 | coh_echoes = (my_coh_aver[pair[0],:] == 1).nonzero() |
|
|
4017 | 4017 | incoh_echoes = (my_coh_aver[pair[0],:] != 1).nonzero() |
|
|
4018 | 4018 | incoh_echoes = incoh_echoes[0] |
|
|
4019 | 4019 | if len(incoh_echoes) > 0: |
|
|
4020 | 4020 | my_incoh_spectra[pair[0],:,incoh_echoes] = spectra[pair[0],:,incoh_echoes] |
|
|
4021 | 4021 | my_incoh_spectra[pair[1],:,incoh_echoes] = spectra[pair[1],:,incoh_echoes] |
|
|
4022 | 4022 | my_incoh_cspectra[ic,:,incoh_echoes] = cspectra[ic,:,incoh_echoes] |
|
|
4023 | 4023 | my_incoh_aver[pair[0],incoh_echoes] = 1 |
|
|
4024 | 4024 | my_incoh_aver[pair[1],incoh_echoes] = 1 |
|
|
4025 | 4025 | |
|
|
4026 | 4026 | |
|
|
4027 | 4027 | for ic in range(2): |
|
|
4028 | 4028 | pair = crosspairs[ic] |
|
|
4029 | 4029 | |
|
|
4030 | 4030 | valid1 =(my_coh_aver[pair[0],:]==1 ).nonzero() |
|
|
4031 | 4031 | valid2 = (my_coh_aver[pair[1],:]==1).nonzero() |
|
|
4032 | 4032 | valid1 = numpy.array(valid1[0]) |
|
|
4033 | 4033 | valid2 = numpy.array(valid2[0]) |
|
|
4034 | 4034 | valid = valid1 |
|
|
4035 | 4035 | #print valid1 , valid2 |
|
|
4036 | 4036 | for iv in range(len(valid2)): |
|
|
4037 | 4037 | #for ivv in range(len(valid1)) : |
|
|
4038 | 4038 | indv = numpy.array((valid1 == valid2[iv]).nonzero()) |
|
|
4039 | 4039 | if len(indv[0]) == 0 : |
|
|
4040 | 4040 | valid = numpy.concatenate((valid,valid2[iv]), axis=None) |
|
|
4041 | 4041 | #print valid |
|
|
4042 | 4042 | #valid = numpy.concatenate((valid1,valid2), axis=None) |
|
|
4043 | 4043 | valid1 =(my_coh_aver[pair[0],:] !=1 ).nonzero() |
|
|
4044 | 4044 | valid2 = (my_coh_aver[pair[1],:] !=1).nonzero() |
|
|
4045 | 4045 | valid1 = numpy.array(valid1[0]) |
|
|
4046 | 4046 | valid2 = numpy.array(valid2[0]) |
|
|
4047 | 4047 | incoh_echoes = valid1 |
|
|
4048 | 4048 | #print valid1, valid2 |
|
|
4049 | 4049 | #incoh_echoes= numpy.concatenate((valid1,valid2), axis=None) |
|
|
4050 | 4050 | for iv in range(len(valid2)): |
|
|
4051 | 4051 | #for ivv in range(len(valid1)) : |
|
|
4052 | 4052 | indv = numpy.array((valid1 == valid2[iv]).nonzero()) |
|
|
4053 | 4053 | if len(indv[0]) == 0 : |
|
|
4054 | 4054 | incoh_echoes = numpy.concatenate(( incoh_echoes,valid2[iv]), axis=None) |
|
|
4055 | 4055 | #print incoh_echoes |
|
|
4056 | 4056 | if len(valid)>0: |
|
|
4057 | 4057 | #print pair |
|
|
4058 | 4058 | coh_spectra[pair[0],:,valid] = spectra[pair[0],:,valid] |
|
|
4059 | 4059 | coh_spectra[pair[1],:,valid] = spectra[pair[1],:,valid] |
|
|
4060 | 4060 | coh_cspectra[ic,:,valid] = cspectra[ic,:,valid] |
|
|
4061 | 4061 | coh_aver[pair[0],valid]=1 |
|
|
4062 | 4062 | coh_aver[pair[1],valid]=1 |
|
|
4063 | 4063 | if len(incoh_echoes)>0: |
|
|
4064 | 4064 | incoh_spectra[pair[0],:,incoh_echoes] = spectra[pair[0],:,incoh_echoes] |
|
|
4065 | 4065 | incoh_spectra[pair[1],:,incoh_echoes] = spectra[pair[1],:,incoh_echoes] |
|
|
4066 | 4066 | incoh_cspectra[ic,:,incoh_echoes] = cspectra[ic,:,incoh_echoes] |
|
|
4067 | 4067 | incoh_aver[pair[0],incoh_echoes]=1 |
|
|
4068 | 4068 | incoh_aver[pair[1],incoh_echoes]=1 |
|
|
4069 | 4069 | #plt.imshow(spectra[0,:,:],vmin=20000000) |
|
|
4070 | 4070 | #plt.show() |
|
|
4071 | 4071 | #my_incoh_aver = my_incoh_aver+1 |
|
|
4072 | 4072 | |
|
|
4073 | 4073 | #spec = my_incoh_spectra.copy() |
|
|
4074 | 4074 | #cspec = my_incoh_cspectra.copy() |
|
|
4075 | 4075 | #print('######################', spec) |
|
|
4076 | 4076 | #print(self.numpy) |
|
|
4077 | 4077 | #return spec, cspec,coh_aver |
|
|
4078 | 4078 | return my_incoh_spectra ,my_incoh_cspectra,my_incoh_aver,my_coh_aver, incoh_spectra, coh_spectra, incoh_cspectra, coh_cspectra, incoh_aver, coh_aver |
|
|
4079 | 4079 | |
|
|
4080 | 4080 | def __CleanCoherent(self,snrth, spectra, cspectra, coh_aver,dataOut, noise,clean_coh_echoes,index): |
|
|
4081 | 4081 | |
|
|
4082 | 4082 | import matplotlib.pyplot as plt |
|
|
4083 | 4083 | nProf = dataOut.nProfiles |
|
|
4084 | 4084 | heights = dataOut.heightList |
|
|
4085 | 4085 | nHei = len(heights) |
|
|
4086 | 4086 | channels = dataOut.channelList |
|
|
4087 | 4087 | nChan = len(channels) |
|
|
4088 | 4088 | crosspairs = dataOut.groupList |
|
|
4089 | 4089 | nPairs = len(crosspairs) |
|
|
4090 | 4090 | |
|
|
4091 | 4091 | #data = dataOut.data_pre[0] |
|
|
4092 | 4092 | absc = dataOut.abscissaList[:-1] |
|
|
4093 | 4093 | #noise = dataOut.noise |
|
|
4094 | 4094 | #nChannel = data.shape[0] |
|
|
4095 | 4095 | data_param = numpy.zeros((nChan, 4, spectra.shape[2])) |
|
|
4096 | 4096 | |
|
|
4097 | 4097 | |
|
|
4098 | 4098 | #plt.plot(absc) |
|
|
4099 | 4099 | #plt.show() |
|
|
4100 | 4100 | clean_coh_spectra = spectra.copy() |
|
|
4101 | 4101 | clean_coh_cspectra = cspectra.copy() |
|
|
4102 | 4102 | clean_coh_aver = coh_aver.copy() |
|
|
4103 | 4103 | |
|
|
4104 | 4104 | spwd_th=[10,6] #spwd_th[0] --> For satellites ; spwd_th[1] --> For special events like SUN. |
|
|
4105 | 4105 | coh_th = 0.75 |
|
|
4106 | 4106 | |
|
|
4107 | 4107 | rtime0 = [6,18] # periodo sin ESF |
|
|
4108 | 4108 | rtime1 = [10.5,13.5] # periodo con alta coherencia y alto ancho espectral (esperado): SOL. |
|
|
4109 | 4109 | |
|
|
4110 | 4110 | time = index*5./60 |
|
|
4111 | 4111 | if clean_coh_echoes == 1 : |
|
|
4112 | 4112 | for ind in range(nChan): |
|
|
4113 | 4113 | data_param[ind,:,:] = self.__calculateMoments( spectra[ind,:,:] , absc , noise[ind] ) |
|
|
4114 | 4114 | #print data_param[:,3] |
|
|
4115 | 4115 | spwd = data_param[:,3] |
|
|
4116 | 4116 | #print spwd.shape |
|
|
4117 | 4117 | # SPECB_JULIA,header=anal_header,jspectra=spectra,vel=velocities,hei=heights, num_aver=1, mode_fit=0,smoothing=smoothing,jvelr=velr,jspwd=spwd,jsnr=snr,jnoise=noise,jstdvnoise=stdvnoise |
|
|
4118 | 4118 | #spwd1=[ 1.65607, 1.43416, 0.500373, 0.208361, 0.000000, 26.7767, 22.5936, 26.7530, 20.6962, 29.1098, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 28.0300, 27.0511, 27.8810, 26.3126, 27.8445, 24.6181, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000] |
|
|
4119 | 4119 | #spwd=numpy.array([spwd1,spwd1,spwd1,spwd1]) |
|
|
4120 | 4120 | #print spwd.shape, heights.shape,coh_aver.shape |
|
|
4121 | 4121 | # para obtener spwd |
|
|
4122 | 4122 | for ic in range(nPairs): |
|
|
4123 | 4123 | pair = crosspairs[ic] |
|
|
4124 | 4124 | coh = numpy.squeeze(numpy.sum(cspectra[ic,:,:], axis=1)/numpy.sqrt(numpy.sum(spectra[pair[0],:,:], axis=1)*numpy.sum(spectra[pair[1],:,:], axis=1))) |
|
|
4125 | 4125 | for ih in range(nHei) : |
|
|
4126 | 4126 | # Considering heights higher than 200km in order to avoid removing phenomena like EEJ. |
|
|
4127 | 4127 | if heights[ih] >= 200 and coh_aver[pair[0],ih] == 1 and coh_aver[pair[1],ih] == 1 : |
|
|
4128 | 4128 | # Checking coherence |
|
|
4129 | 4129 | if (numpy.abs(coh[ih]) <= coh_th) or (time >= rtime0[0] and time <= rtime0[1]) : |
|
|
4130 | 4130 | # Checking spectral widths |
|
|
4131 | 4131 | if (spwd[pair[0],ih] > spwd_th[0]) or (spwd[pair[1],ih] > spwd_th[0]) : |
|
|
4132 | 4132 | # satelite |
|
|
4133 | 4133 | clean_coh_spectra[pair,ih,:] = 0.0 |
|
|
4134 | 4134 | clean_coh_cspectra[ic,ih,:] = 0.0 |
|
|
4135 | 4135 | clean_coh_aver[pair,ih] = 0 |
|
|
4136 | 4136 | else : |
|
|
4137 | 4137 | if ((spwd[pair[0],ih] < spwd_th[1]) or (spwd[pair[1],ih] < spwd_th[1])) : |
|
|
4138 | 4138 | # Especial event like sun. |
|
|
4139 | 4139 | clean_coh_spectra[pair,ih,:] = 0.0 |
|
|
4140 | 4140 | clean_coh_cspectra[ic,ih,:] = 0.0 |
|
|
4141 | 4141 | clean_coh_aver[pair,ih] = 0 |
|
|
4142 | 4142 | |
|
|
4143 | 4143 | return clean_coh_spectra, clean_coh_cspectra, clean_coh_aver |
|
|
4144 | 4144 | |
|
|
4145 | 4145 | isConfig = False |
|
|
4146 | 4146 | __dataReady = False |
|
|
4147 | 4147 | bloques = None |
|
|
4148 | 4148 | bloque0 = None |
|
|
4149 | 4149 | |
|
|
4150 | 4150 | def __init__(self): |
|
|
4151 | 4151 | Operation.__init__(self) |
|
|
4152 | 4152 | self.i=0 |
|
|
4153 | 4153 | self.isConfig = False |
|
|
4154 | 4154 | |
|
|
4155 | 4155 | |
|
|
4156 | 4156 | def setup(self,nChan,nProf,nHei,nBlocks): |
|
|
4157 | 4157 | self.__dataReady = False |
|
|
4158 | 4158 | self.bloques = numpy.zeros([2, nProf, nHei,nBlocks], dtype= complex) |
|
|
4159 | 4159 | self.bloque0 = numpy.zeros([nChan, nProf, nHei, nBlocks]) |
|
|
4160 | 4160 | |
|
|
4161 | 4161 | #def CleanRayleigh(self,dataOut,spectra,cspectra,out_spectra,out_cspectra,sat_spectra,sat_cspectra,crosspairs,heights, channels, nProf,nHei,nChan,nPairs,nIncohInt,nBlocks): |
|
|
4162 | 4162 | def CleanRayleigh(self,dataOut,spectra,cspectra,save_drifts): |
|
|
4163 | 4163 | #import matplotlib.pyplot as plt |
|
|
4164 | 4164 | #for k in range(149): |
|
|
4165 | 4165 | |
|
|
4166 | 4166 | # self.bloque0[:,:,:,k] = spectra[:,:,0:nHei] |
|
|
4167 | 4167 | # self.bloques[:,:,:,k] = cspectra[:,:,0:nHei] |
|
|
4168 | 4168 | #if self.i==nBlocks: |
|
|
4169 | 4169 | # self.i==0 |
|
|
4170 | 4170 | rfunc = cspectra.copy() #self.bloques |
|
|
4171 | 4171 | n_funct = len(rfunc[0,:,0,0]) |
|
|
4172 | 4172 | val_spc = spectra*0.0 #self.bloque0*0.0 |
|
|
4173 | 4173 | val_cspc = cspectra*0.0 #self.bloques*0.0 |
|
|
4174 | 4174 | in_sat_spectra = spectra.copy() #self.bloque0 |
|
|
4175 | 4175 | in_sat_cspectra = cspectra.copy() #self.bloques |
|
|
4176 | 4176 | |
|
|
4177 | 4177 | #print( rfunc.shape) |
|
|
4178 | 4178 | min_hei = 200 |
|
|
4179 | 4179 | nProf = dataOut.nProfiles |
|
|
4180 | 4180 | heights = dataOut.heightList |
|
|
4181 | 4181 | nHei = len(heights) |
|
|
4182 | 4182 | channels = dataOut.channelList |
|
|
4183 | 4183 | nChan = len(channels) |
|
|
4184 | 4184 | crosspairs = dataOut.groupList |
|
|
4185 | 4185 | nPairs = len(crosspairs) |
|
|
4186 | 4186 | hval=(heights >= min_hei).nonzero() |
|
|
4187 | 4187 | ih=hval[0] |
|
|
4188 | 4188 | #print numpy.absolute(rfunc[:,0,0,14]) |
|
|
4189 | 4189 | for ih in range(hval[0][0],nHei): |
|
|
4190 | 4190 | for ifreq in range(nProf): |
|
|
4191 | 4191 | for ii in range(n_funct): |
|
|
4192 | 4192 | |
|
|
4193 | 4193 | func2clean = 10*numpy.log10(numpy.absolute(rfunc[:,ii,ifreq,ih])) |
|
|
4194 | 4194 | #print numpy.amin(func2clean) |
|
|
4195 | 4195 | val = (numpy.isfinite(func2clean)==True).nonzero() |
|
|
4196 | 4196 | if len(val)>0: |
|
|
4197 | 4197 | min_val = numpy.around(numpy.amin(func2clean)-2) #> (-40) |
|
|
4198 | 4198 | if min_val <= -40 : min_val = -40 |
|
|
4199 | 4199 | max_val = numpy.around(numpy.amax(func2clean)+2) #< 200 |
|
|
4200 | 4200 | if max_val >= 200 : max_val = 200 |
|
|
4201 | 4201 | #print min_val, max_val |
|
|
4202 | 4202 | step = 1 |
|
|
4203 | 4203 | #Getting bins and the histogram |
|
|
4204 | 4204 | x_dist = min_val + numpy.arange(1 + ((max_val-(min_val))/step))*step |
|
|
4205 | 4205 | y_dist,binstep = numpy.histogram(func2clean,bins=range(int(min_val),int(max_val+2),step)) |
|
|
4206 | 4206 | mean = numpy.sum(x_dist * y_dist) / numpy.sum(y_dist) |
|
|
4207 | 4207 | sigma = numpy.sqrt(numpy.sum(y_dist * (x_dist - mean)**2) / numpy.sum(y_dist)) |
|
|
4208 | 4208 | parg = [numpy.amax(y_dist),mean,sigma] |
|
|
4209 | 4209 | try : |
|
|
4210 | 4210 | gauss_fit, covariance = curve_fit(fit_func, x_dist, y_dist,p0=parg) |
|
|
4211 | 4211 | mode = gauss_fit[1] |
|
|
4212 | 4212 | stdv = gauss_fit[2] |
|
|
4213 | 4213 | except: |
|
|
4214 | 4214 | mode = mean |
|
|
4215 | 4215 | stdv = sigma |
|
|
4216 | 4216 | # if ih == 14 and ii == 0 and ifreq ==0 : |
|
|
4217 | 4217 | # print x_dist.shape, y_dist.shape |
|
|
4218 | 4218 | # print x_dist, y_dist |
|
|
4219 | 4219 | # print min_val, max_val, binstep |
|
|
4220 | 4220 | # print func2clean |
|
|
4221 | 4221 | # print mean,sigma |
|
|
4222 | 4222 | # mean1,std = norm.fit(y_dist) |
|
|
4223 | 4223 | # print mean1, std, gauss_fit |
|
|
4224 | 4224 | # print fit_func(x_dist,gauss_fit[0],gauss_fit[1],gauss_fit[2]) |
|
|
4225 | 4225 | # 7.84616 53.9307 3.61863 |
|
|
4226 | 4226 | #stdv = 3.61863 # 2.99089 |
|
|
4227 | 4227 | #mode = 53.9307 #7.79008 |
|
|
4228 | 4228 | |
|
|
4229 | 4229 | #Removing echoes greater than mode + 3*stdv |
|
|
4230 | 4230 | factor_stdv = 2.5 |
|
|
4231 | 4231 | noval = (abs(func2clean - mode)>=(factor_stdv*stdv)).nonzero() |
|
|
4232 | 4232 | |
|
|
4233 | 4233 | if len(noval[0]) > 0: |
|
|
4234 | 4234 | novall = ((func2clean - mode) >= (factor_stdv*stdv)).nonzero() |
|
|
4235 | 4235 | cross_pairs = crosspairs[ii] |
|
|
4236 | 4236 | #Getting coherent echoes which are removed. |
|
|
4237 | 4237 | if len(novall[0]) > 0: |
|
|
4238 | 4238 | #val_spc[(0,1),novall[a],ih] = 1 |
|
|
4239 | 4239 | #val_spc[,(2,3),novall[a],ih] = 1 |
|
|
4240 | 4240 | val_spc[novall[0],cross_pairs[0],ifreq,ih] = 1 |
|
|
4241 | 4241 | val_spc[novall[0],cross_pairs[1],ifreq,ih] = 1 |
|
|
4242 | 4242 | val_cspc[novall[0],ii,ifreq,ih] = 1 |
|
|
4243 | 4243 | #print("OUT NOVALL 1") |
|
|
4244 | 4244 | #Removing coherent from ISR data |
|
|
4245 | 4245 | # if ih == 17 and ii == 0 and ifreq ==0 : |
|
|
4246 | 4246 | # print spectra[:,cross_pairs[0],ifreq,ih] |
|
|
4247 | 4247 | spectra[noval,cross_pairs[0],ifreq,ih] = numpy.nan |
|
|
4248 | 4248 | spectra[noval,cross_pairs[1],ifreq,ih] = numpy.nan |
|
|
4249 | 4249 | cspectra[noval,ii,ifreq,ih] = numpy.nan |
|
|
4250 | 4250 | # if ih == 17 and ii == 0 and ifreq ==0 : |
|
|
4251 | 4251 | # print spectra[:,cross_pairs[0],ifreq,ih] |
|
|
4252 | 4252 | # print noval, len(noval[0]) |
|
|
4253 | 4253 | # print novall, len(novall[0]) |
|
|
4254 | 4254 | # print factor_stdv*stdv |
|
|
4255 | 4255 | # print func2clean-mode |
|
|
4256 | 4256 | # print val_spc[:,cross_pairs[0],ifreq,ih] |
|
|
4257 | 4257 | # print spectra[:,cross_pairs[0],ifreq,ih] |
|
|
4258 | 4258 | #no sale es para savedrifts >2 |
|
|
4259 | 4259 | ''' channels = channels |
|
|
4260 | 4260 | cross_pairs = cross_pairs |
|
|
4261 | 4261 | #print("OUT NOVALL 2") |
|
|
4262 | 4262 | |
|
|
4263 | 4263 | vcross0 = (cross_pairs[0] == channels[ii]).nonzero() |
|
|
4264 | 4264 | vcross1 = (cross_pairs[1] == channels[ii]).nonzero() |
|
|
4265 | 4265 | vcross = numpy.concatenate((vcross0,vcross1),axis=None) |
|
|
4266 | 4266 | #print('vcros =', vcross) |
|
|
4267 | 4267 | |
|
|
4268 | 4268 | #Getting coherent echoes which are removed. |
|
|
4269 | 4269 | if len(novall) > 0: |
|
|
4270 | 4270 | #val_spc[novall,ii,ifreq,ih] = 1 |
|
|
4271 | 4271 | val_spc[ii,ifreq,ih,novall] = 1 |
|
|
4272 | 4272 | if len(vcross) > 0: |
|
|
4273 | 4273 | val_cspc[vcross,ifreq,ih,novall] = 1 |
|
|
4274 | 4274 | |
|
|
4275 | 4275 | #Removing coherent from ISR data. |
|
|
4276 | 4276 | self.bloque0[ii,ifreq,ih,noval] = numpy.nan |
|
|
4277 | 4277 | if len(vcross) > 0: |
|
|
4278 | 4278 | self.bloques[vcross,ifreq,ih,noval] = numpy.nan |
|
|
4279 | 4279 | ''' |
|
|
4280 | 4280 | #Getting average of the spectra and cross-spectra from incoherent echoes. |
|
|
4281 | 4281 | out_spectra = numpy.zeros([nChan,nProf,nHei], dtype=float) #+numpy.nan |
|
|
4282 | 4282 | out_cspectra = numpy.zeros([nPairs,nProf,nHei], dtype=complex) #+numpy.nan |
|
|
4283 | 4283 | for ih in range(nHei): |
|
|
4284 | 4284 | for ifreq in range(nProf): |
|
|
4285 | 4285 | for ich in range(nChan): |
|
|
4286 | 4286 | tmp = spectra[:,ich,ifreq,ih] |
|
|
4287 | 4287 | valid = (numpy.isfinite(tmp[:])==True).nonzero() |
|
|
4288 | 4288 | # if ich == 0 and ifreq == 0 and ih == 17 : |
|
|
4289 | 4289 | # print tmp |
|
|
4290 | 4290 | # print valid |
|
|
4291 | 4291 | # print len(valid[0]) |
|
|
4292 | 4292 | #print('TMP',tmp) |
|
|
4293 | 4293 | if len(valid[0]) >0 : |
|
|
4294 | 4294 | out_spectra[ich,ifreq,ih] = numpy.nansum(tmp)/len(valid[0]) |
|
|
4295 | 4295 | #for icr in range(nPairs): |
|
|
4296 | 4296 | for icr in range(nPairs): |
|
|
4297 | 4297 | tmp = numpy.squeeze(cspectra[:,icr,ifreq,ih]) |
|
|
4298 | 4298 | valid = (numpy.isfinite(tmp)==True).nonzero() |
|
|
4299 | 4299 | if len(valid[0]) > 0: |
|
|
4300 | 4300 | out_cspectra[icr,ifreq,ih] = numpy.nansum(tmp)/len(valid[0]) |
|
|
4301 | 4301 | # print('##########################################################') |
|
|
4302 | 4302 | #Removing fake coherent echoes (at least 4 points around the point) |
|
|
4303 | 4303 | |
|
|
4304 | 4304 | val_spectra = numpy.sum(val_spc,0) |
|
|
4305 | 4305 | val_cspectra = numpy.sum(val_cspc,0) |
|
|
4306 | 4306 | |
|
|
4307 | 4307 | val_spectra = self.REM_ISOLATED_POINTS(val_spectra,4) |
|
|
4308 | 4308 | val_cspectra = self.REM_ISOLATED_POINTS(val_cspectra,4) |
|
|
4309 | 4309 | |
|
|
4310 | 4310 | for i in range(nChan): |
|
|
4311 | 4311 | for j in range(nProf): |
|
|
4312 | 4312 | for k in range(nHei): |
|
|
4313 | 4313 | if numpy.isfinite(val_spectra[i,j,k]) and val_spectra[i,j,k] < 1 : |
|
|
4314 | 4314 | val_spc[:,i,j,k] = 0.0 |
|
|
4315 | 4315 | for i in range(nPairs): |
|
|
4316 | 4316 | for j in range(nProf): |
|
|
4317 | 4317 | for k in range(nHei): |
|
|
4318 | 4318 | if numpy.isfinite(val_cspectra[i,j,k]) and val_cspectra[i,j,k] < 1 : |
|
|
4319 | 4319 | val_cspc[:,i,j,k] = 0.0 |
|
|
4320 | 4320 | # val_spc = numpy.reshape(val_spc, (len(spectra[:,0,0,0]),nProf*nHei*nChan)) |
|
|
4321 | 4321 | # if numpy.isfinite(val_spectra)==str(True): |
|
|
4322 | 4322 | # noval = (val_spectra<1).nonzero() |
|
|
4323 | 4323 | # if len(noval) > 0: |
|
|
4324 | 4324 | # val_spc[:,noval] = 0.0 |
|
|
4325 | 4325 | # val_spc = numpy.reshape(val_spc, (149,nChan,nProf,nHei)) |
|
|
4326 | 4326 | |
|
|
4327 | 4327 | #val_cspc = numpy.reshape(val_spc, (149,nChan*nHei*nProf)) |
|
|
4328 | 4328 | #if numpy.isfinite(val_cspectra)==str(True): |
|
|
4329 | 4329 | # noval = (val_cspectra<1).nonzero() |
|
|
4330 | 4330 | # if len(noval) > 0: |
|
|
4331 | 4331 | # val_cspc[:,noval] = 0.0 |
|
|
4332 | 4332 | # val_cspc = numpy.reshape(val_cspc, (149,nChan,nProf,nHei)) |
|
|
4333 | 4333 | |
|
|
4334 | 4334 | tmp_sat_spectra = spectra.copy() |
|
|
4335 | 4335 | tmp_sat_spectra = tmp_sat_spectra*numpy.nan |
|
|
4336 | 4336 | tmp_sat_cspectra = cspectra.copy() |
|
|
4337 | 4337 | tmp_sat_cspectra = tmp_sat_cspectra*numpy.nan |
|
|
4338 | 4338 | |
|
|
4339 | 4339 | # fig = plt.figure(figsize=(6,5)) |
|
|
4340 | 4340 | # left, bottom, width, height = 0.1, 0.1, 0.8, 0.8 |
|
|
4341 | 4341 | # ax = fig.add_axes([left, bottom, width, height]) |
|
|
4342 | 4342 | # cp = ax.contour(10*numpy.log10(numpy.absolute(spectra[0,0,:,:]))) |
|
|
4343 | 4343 | # ax.clabel(cp, inline=True,fontsize=10) |
|
|
4344 | 4344 | # plt.show() |
|
|
4345 | 4345 | |
|
|
4346 | 4346 | val = (val_spc > 0).nonzero() |
|
|
4347 | 4347 | if len(val[0]) > 0: |
|
|
4348 | 4348 | tmp_sat_spectra[val] = in_sat_spectra[val] |
|
|
4349 | 4349 | |
|
|
4350 | 4350 | val = (val_cspc > 0).nonzero() |
|
|
4351 | 4351 | if len(val[0]) > 0: |
|
|
4352 | 4352 | tmp_sat_cspectra[val] = in_sat_cspectra[val] |
|
|
4353 | 4353 | |
|
|
4354 | 4354 | #Getting average of the spectra and cross-spectra from incoherent echoes. |
|
|
4355 | 4355 | sat_spectra = numpy.zeros((nChan,nProf,nHei), dtype=float) |
|
|
4356 | 4356 | sat_cspectra = numpy.zeros((nPairs,nProf,nHei), dtype=complex) |
|
|
4357 | 4357 | for ih in range(nHei): |
|
|
4358 | 4358 | for ifreq in range(nProf): |
|
|
4359 | 4359 | for ich in range(nChan): |
|
|
4360 | 4360 | tmp = numpy.squeeze(tmp_sat_spectra[:,ich,ifreq,ih]) |
|
|
4361 | 4361 | valid = (numpy.isfinite(tmp)).nonzero() |
|
|
4362 | 4362 | if len(valid[0]) > 0: |
|
|
4363 | 4363 | sat_spectra[ich,ifreq,ih] = numpy.nansum(tmp)/len(valid[0]) |
|
|
4364 | 4364 | |
|
|
4365 | 4365 | for icr in range(nPairs): |
|
|
4366 | 4366 | tmp = numpy.squeeze(tmp_sat_cspectra[:,icr,ifreq,ih]) |
|
|
4367 | 4367 | valid = (numpy.isfinite(tmp)).nonzero() |
|
|
4368 | 4368 | if len(valid[0]) > 0: |
|
|
4369 | 4369 | sat_cspectra[icr,ifreq,ih] = numpy.nansum(tmp)/len(valid[0]) |
|
|
4370 | 4370 | #self.__dataReady= True |
|
|
4371 | 4371 | #sat_spectra, sat_cspectra= sat_spectra, sat_cspectra |
|
|
4372 | 4372 | #if not self.__dataReady: |
|
|
4373 | 4373 | #return None, None |
|
|
4374 | 4374 | return out_spectra, out_cspectra,sat_spectra,sat_cspectra |
|
|
4375 | 4375 | def REM_ISOLATED_POINTS(self,array,rth): |
|
|
4376 | 4376 | # import matplotlib.pyplot as plt |
|
|
4377 | 4377 | if rth == None : rth = 4 |
|
|
4378 | 4378 | |
|
|
4379 | 4379 | num_prof = len(array[0,:,0]) |
|
|
4380 | 4380 | num_hei = len(array[0,0,:]) |
|
|
4381 | 4381 | n2d = len(array[:,0,0]) |
|
|
4382 | 4382 | |
|
|
4383 | 4383 | for ii in range(n2d) : |
|
|
4384 | 4384 | #print ii,n2d |
|
|
4385 | 4385 | tmp = array[ii,:,:] |
|
|
4386 | 4386 | #print tmp.shape, array[ii,101,:],array[ii,102,:] |
|
|
4387 | 4387 | |
|
|
4388 | 4388 | # fig = plt.figure(figsize=(6,5)) |
|
|
4389 | 4389 | # left, bottom, width, height = 0.1, 0.1, 0.8, 0.8 |
|
|
4390 | 4390 | # ax = fig.add_axes([left, bottom, width, height]) |
|
|
4391 | 4391 | # x = range(num_prof) |
|
|
4392 | 4392 | # y = range(num_hei) |
|
|
4393 | 4393 | # cp = ax.contour(y,x,tmp) |
|
|
4394 | 4394 | # ax.clabel(cp, inline=True,fontsize=10) |
|
|
4395 | 4395 | # plt.show() |
|
|
4396 | 4396 | |
|
|
4397 | 4397 | #indxs = WHERE(FINITE(tmp) AND tmp GT 0,cindxs) |
|
|
4398 | 4398 | tmp = numpy.reshape(tmp,num_prof*num_hei) |
|
|
4399 | 4399 | indxs1 = (numpy.isfinite(tmp)==True).nonzero() |
|
|
4400 | 4400 | indxs2 = (tmp > 0).nonzero() |
|
|
4401 | 4401 | |
|
|
4402 | 4402 | indxs1 = (indxs1[0]) |
|
|
4403 | 4403 | indxs2 = indxs2[0] |
|
|
4404 | 4404 | #indxs1 = numpy.array(indxs1[0]) |
|
|
4405 | 4405 | #indxs2 = numpy.array(indxs2[0]) |
|
|
4406 | 4406 | indxs = None |
|
|
4407 | 4407 | #print indxs1 , indxs2 |
|
|
4408 | 4408 | for iv in range(len(indxs2)): |
|
|
4409 | 4409 | indv = numpy.array((indxs1 == indxs2[iv]).nonzero()) |
|
|
4410 | 4410 | #print len(indxs2), indv |
|
|
4411 | 4411 | if len(indv[0]) > 0 : |
|
|
4412 | 4412 | indxs = numpy.concatenate((indxs,indxs2[iv]), axis=None) |
|
|
4413 | 4413 | # print indxs |
|
|
4414 | 4414 | indxs = indxs[1:] |
|
|
4415 | 4415 | #print indxs, len(indxs) |
|
|
4416 | 4416 | if len(indxs) < 4 : |
|
|
4417 | 4417 | array[ii,:,:] = 0. |
|
|
4418 | 4418 | return |
|
|
4419 | 4419 | |
|
|
4420 | 4420 | xpos = numpy.mod(indxs ,num_hei) |
|
|
4421 | 4421 | ypos = (indxs / num_hei) |
|
|
4422 | 4422 | sx = numpy.argsort(xpos) # Ordering respect to "x" (time) |
|
|
4423 | 4423 | #print sx |
|
|
4424 | 4424 | xpos = xpos[sx] |
|
|
4425 | 4425 | ypos = ypos[sx] |
|
|
4426 | 4426 | |
|
|
4427 | 4427 | # *********************************** Cleaning isolated points ********************************** |
|
|
4428 | 4428 | ic = 0 |
|
|
4429 | 4429 | while True : |
|
|
4430 | 4430 | r = numpy.sqrt(list(numpy.power((xpos[ic]-xpos),2)+ numpy.power((ypos[ic]-ypos),2))) |
|
|
4431 | 4431 | #no_coh = WHERE(FINITE(r) AND (r LE rth),cno_coh) |
|
|
4432 | 4432 | #plt.plot(r) |
|
|
4433 | 4433 | #plt.show() |
|
|
4434 | 4434 | no_coh1 = (numpy.isfinite(r)==True).nonzero() |
|
|
4435 | 4435 | no_coh2 = (r <= rth).nonzero() |
|
|
4436 | 4436 | #print r, no_coh1, no_coh2 |
|
|
4437 | 4437 | no_coh1 = numpy.array(no_coh1[0]) |
|
|
4438 | 4438 | no_coh2 = numpy.array(no_coh2[0]) |
|
|
4439 | 4439 | no_coh = None |
|
|
4440 | 4440 | #print valid1 , valid2 |
|
|
4441 | 4441 | for iv in range(len(no_coh2)): |
|
|
4442 | 4442 | indv = numpy.array((no_coh1 == no_coh2[iv]).nonzero()) |
|
|
4443 | 4443 | if len(indv[0]) > 0 : |
|
|
4444 | 4444 | no_coh = numpy.concatenate((no_coh,no_coh2[iv]), axis=None) |
|
|
4445 | 4445 | no_coh = no_coh[1:] |
|
|
4446 | 4446 | #print len(no_coh), no_coh |
|
|
4447 | 4447 | if len(no_coh) < 4 : |
|
|
4448 | 4448 | #print xpos[ic], ypos[ic], ic |
|
|
4449 | 4449 | # plt.plot(r) |
|
|
4450 | 4450 | # plt.show() |
|
|
4451 | 4451 | xpos[ic] = numpy.nan |
|
|
4452 | 4452 | ypos[ic] = numpy.nan |
|
|
4453 | 4453 | |
|
|
4454 | 4454 | ic = ic + 1 |
|
|
4455 | 4455 | if (ic == len(indxs)) : |
|
|
4456 | 4456 | break |
|
|
4457 | 4457 | #print( xpos, ypos) |
|
|
4458 | 4458 | |
|
|
4459 | 4459 | indxs = (numpy.isfinite(list(xpos))==True).nonzero() |
|
|
4460 | 4460 | #print indxs[0] |
|
|
4461 | 4461 | if len(indxs[0]) < 4 : |
|
|
4462 | 4462 | array[ii,:,:] = 0. |
|
|
4463 | 4463 | return |
|
|
4464 | 4464 | |
|
|
4465 | 4465 | xpos = xpos[indxs[0]] |
|
|
4466 | 4466 | ypos = ypos[indxs[0]] |
|
|
4467 | 4467 | for i in range(0,len(ypos)): |
|
|
4468 | 4468 | ypos[i]=int(ypos[i]) |
|
|
4469 | 4469 | junk = tmp |
|
|
4470 | 4470 | tmp = junk*0.0 |
|
|
4471 | 4471 | |
|
|
4472 | 4472 | tmp[list(xpos + (ypos*num_hei))] = junk[list(xpos + (ypos*num_hei))] |
|
|
4473 | 4473 | array[ii,:,:] = numpy.reshape(tmp,(num_prof,num_hei)) |
|
|
4474 | 4474 | |
|
|
4475 | 4475 | #print array.shape |
|
|
4476 | 4476 | #tmp = numpy.reshape(tmp,(num_prof,num_hei)) |
|
|
4477 | 4477 | #print tmp.shape |
|
|
4478 | 4478 | |
|
|
4479 | 4479 | # fig = plt.figure(figsize=(6,5)) |
|
|
4480 | 4480 | # left, bottom, width, height = 0.1, 0.1, 0.8, 0.8 |
|
|
4481 | 4481 | # ax = fig.add_axes([left, bottom, width, height]) |
|
|
4482 | 4482 | # x = range(num_prof) |
|
|
4483 | 4483 | # y = range(num_hei) |
|
|
4484 | 4484 | # cp = ax.contour(y,x,array[ii,:,:]) |
|
|
4485 | 4485 | # ax.clabel(cp, inline=True,fontsize=10) |
|
|
4486 | 4486 | # plt.show() |
|
|
4487 | 4487 | return array |
|
|
4488 | 4488 | def moments(self,doppler,yarray,npoints): |
|
|
4489 | 4489 | ytemp = yarray |
|
|
4490 | 4490 | #val = WHERE(ytemp GT 0,cval) |
|
|
4491 | 4491 | #if cval == 0 : val = range(npoints-1) |
|
|
4492 | 4492 | val = (ytemp > 0).nonzero() |
|
|
4493 | 4493 | val = val[0] |
|
|
4494 | 4494 | #print('hvalid:',hvalid) |
|
|
4495 | 4495 | #print('valid', valid) |
|
|
4496 | 4496 | if len(val) == 0 : val = range(npoints-1) |
|
|
4497 | 4497 | |
|
|
4498 | 4498 | ynew = 0.5*(ytemp[val[0]]+ytemp[val[len(val)-1]]) |
|
|
4499 | 4499 | ytemp[len(ytemp):] = [ynew] |
|
|
4500 | 4500 | |
|
|
4501 | 4501 | index = 0 |
|
|
4502 | 4502 | index = numpy.argmax(ytemp) |
|
|
4503 | 4503 | ytemp = numpy.roll(ytemp,int(npoints/2)-1-index) |
|
|
4504 | 4504 | ytemp = ytemp[0:npoints-1] |
|
|
4505 | 4505 | |
|
|
4506 | 4506 | fmom = numpy.sum(doppler*ytemp)/numpy.sum(ytemp)+(index-(npoints/2-1))*numpy.abs(doppler[1]-doppler[0]) |
|
|
4507 | 4507 | smom = numpy.sum(doppler*doppler*ytemp)/numpy.sum(ytemp) |
|
|
4508 | 4508 | return [fmom,numpy.sqrt(smom)] |
|
|
4509 | 4509 | # ********************************************************************************************** |
|
|
4510 | 4510 | index = 0 |
|
|
4511 | 4511 | fint = 0 |
|
|
4512 | 4512 | buffer = 0 |
|
|
4513 | 4513 | buffer2 = 0 |
|
|
4514 | 4514 | buffer3 = 0 |
|
|
4515 | 4515 | def run(self, dataOut, getSNR = True, path=None, file=None, groupList=None): |
|
|
4516 | 4516 | nChannels = dataOut.nChannels |
|
|
4517 | 4517 | nHeights= dataOut.heightList.size |
|
|
4518 | 4518 | nProf = dataOut.nProfiles |
|
|
4519 | 4519 | tini=time.localtime(dataOut.utctime) |
|
|
4520 | 4520 | if (tini.tm_min % 5) == 0 and (tini.tm_sec < 5 and self.fint==0): |
|
|
4521 | 4521 | # print tini.tm_min |
|
|
4522 | 4522 | self.index = 0 |
|
|
4523 | 4523 | jspc = self.buffer |
|
|
4524 | 4524 | jcspc = self.buffer2 |
|
|
4525 | 4525 | jnoise = self.buffer3 |
|
|
4526 | 4526 | self.buffer = dataOut.data_spc |
|
|
4527 | 4527 | self.buffer2 = dataOut.data_cspc |
|
|
4528 | 4528 | self.buffer3 = dataOut.noise |
|
|
4529 | 4529 | self.fint = 1 |
|
|
4530 | 4530 | if numpy.any(jspc) : |
|
|
4531 | 4531 | jspc= numpy.reshape(jspc,(int(len(jspc)/4),nChannels,nProf,nHeights)) |
|
|
4532 | 4532 | jcspc= numpy.reshape(jcspc,(int(len(jcspc)/2),2,nProf,nHeights)) |
|
|
4533 | 4533 | jnoise= numpy.reshape(jnoise,(int(len(jnoise)/4),nChannels)) |
|
|
4534 | 4534 | else: |
|
|
4535 | 4535 | dataOut.flagNoData = True |
|
|
4536 | 4536 | return dataOut |
|
|
4537 | 4537 | else : |
|
|
4538 | 4538 | if (tini.tm_min % 5) == 0 : self.fint = 1 |
|
|
4539 | 4539 | else : self.fint = 0 |
|
|
4540 | 4540 | self.index += 1 |
|
|
4541 | 4541 | if numpy.any(self.buffer): |
|
|
4542 | 4542 | self.buffer = numpy.concatenate((self.buffer,dataOut.data_spc), axis=0) |
|
|
4543 | 4543 | self.buffer2 = numpy.concatenate((self.buffer2,dataOut.data_cspc), axis=0) |
|
|
4544 | 4544 | self.buffer3 = numpy.concatenate((self.buffer3,dataOut.noise), axis=0) |
|
|
4545 | 4545 | else: |
|
|
4546 | 4546 | self.buffer = dataOut.data_spc |
|
|
4547 | 4547 | self.buffer2 = dataOut.data_cspc |
|
|
4548 | 4548 | self.buffer3 = dataOut.noise |
|
|
4549 | 4549 | dataOut.flagNoData = True |
|
|
4550 | 4550 | return dataOut |
|
|
4551 | 4551 | if path != None: |
|
|
4552 | 4552 | sys.path.append(path) |
|
|
4553 | 4553 | self.library = importlib.import_module(file) |
|
|
4554 | 4554 | |
|
|
4555 | 4555 | #To be inserted as a parameter |
|
|
4556 | 4556 | groupArray = numpy.array(groupList) |
|
|
4557 | 4557 | #groupArray = numpy.array([[0,1],[2,3]]) |
|
|
4558 | 4558 | dataOut.groupList = groupArray |
|
|
4559 | 4559 | |
|
|
4560 | 4560 | nGroups = groupArray.shape[0] |
|
|
4561 | 4561 | nChannels = dataOut.nChannels |
|
|
4562 | 4562 | nHeights = dataOut.heightList.size |
|
|
4563 | 4563 | |
|
|
4564 | 4564 | #Parameters Array |
|
|
4565 | 4565 | dataOut.data_param = None |
|
|
4566 | 4566 | dataOut.data_paramC = None |
|
|
4567 | 4567 | |
|
|
4568 | 4568 | #Set constants |
|
|
4569 | 4569 | constants = self.library.setConstants(dataOut) |
|
|
4570 | 4570 | dataOut.constants = constants |
|
|
4571 | 4571 | M = dataOut.normFactor |
|
|
4572 | 4572 | N = dataOut.nFFTPoints |
|
|
4573 | 4573 | ippSeconds = dataOut.ippSeconds |
|
|
4574 | 4574 | K = dataOut.nIncohInt |
|
|
4575 | 4575 | pairsArray = numpy.array(dataOut.pairsList) |
|
|
4576 | 4576 | |
|
|
4577 | 4577 | snrth= 20 |
|
|
4578 | 4578 | spectra = dataOut.data_spc |
|
|
4579 | 4579 | cspectra = dataOut.data_cspc |
|
|
4580 | 4580 | nProf = dataOut.nProfiles |
|
|
4581 | 4581 | heights = dataOut.heightList |
|
|
4582 | 4582 | nHei = len(heights) |
|
|
4583 | 4583 | channels = dataOut.channelList |
|
|
4584 | 4584 | nChan = len(channels) |
|
|
4585 | 4585 | nIncohInt = dataOut.nIncohInt |
|
|
4586 | 4586 | crosspairs = dataOut.groupList |
|
|
4587 | 4587 | noise = dataOut.noise |
|
|
4588 | 4588 | jnoise = jnoise/N |
|
|
4589 | 4589 | noise = numpy.nansum(jnoise,axis=0)#/len(jnoise) |
|
|
4590 | 4590 | power = numpy.sum(spectra, axis=1) |
|
|
4591 | 4591 | nPairs = len(crosspairs) |
|
|
4592 | 4592 | absc = dataOut.abscissaList[:-1] |
|
|
4593 | 4593 | |
|
|
4594 | 4594 | if not self.isConfig: |
|
|
4595 | 4595 | self.isConfig = True |
|
|
4596 | 4596 | |
|
|
4597 | 4597 | index = tini.tm_hour*12+tini.tm_min/5 |
|
|
4598 | 4598 | jspc = jspc/N/N |
|
|
4599 | 4599 | jcspc = jcspc/N/N |
|
|
4600 | 4600 | tmp_spectra,tmp_cspectra,sat_spectra,sat_cspectra = self.CleanRayleigh(dataOut,jspc,jcspc,2) |
|
|
4601 | 4601 | jspectra = tmp_spectra*len(jspc[:,0,0,0]) |
|
|
4602 | 4602 | jcspectra = tmp_cspectra*len(jspc[:,0,0,0]) |
|
|
4603 | 4603 | my_incoh_spectra ,my_incoh_cspectra,my_incoh_aver,my_coh_aver, incoh_spectra, coh_spectra, incoh_cspectra, coh_cspectra, incoh_aver, coh_aver = self.__DiffCoherent(jspectra, jcspectra, dataOut, noise, snrth, None, None) |
|
|
4604 | 4604 | clean_coh_spectra, clean_coh_cspectra, clean_coh_aver = self.__CleanCoherent(snrth, coh_spectra, coh_cspectra, coh_aver, dataOut, noise,1,index) |
|
|
4605 | 4605 | dataOut.data_spc = incoh_spectra |
|
|
4606 | 4606 | dataOut.data_cspc = incoh_cspectra |
|
|
4607 | 4607 | |
|
|
4608 | 4608 | clean_num_aver = incoh_aver*len(jspc[:,0,0,0]) |
|
|
4609 | 4609 | coh_num_aver = clean_coh_aver*len(jspc[:,0,0,0]) |
|
|
4610 | 4610 | #List of possible combinations |
|
|
4611 | 4611 | listComb = itertools.combinations(numpy.arange(groupArray.shape[1]),2) |
|
|
4612 | 4612 | indCross = numpy.zeros(len(list(listComb)), dtype = 'int') |
|
|
4613 | 4613 | |
|
|
4614 | 4614 | if getSNR: |
|
|
4615 | 4615 | listChannels = groupArray.reshape((groupArray.size)) |
|
|
4616 | 4616 | listChannels.sort() |
|
|
4617 | 4617 | dataOut.data_SNR = self.__getSNR(dataOut.data_spc[listChannels,:,:], noise[listChannels]) |
|
|
4618 | 4618 | if dataOut.data_paramC is None: |
|
|
4619 | 4619 | dataOut.data_paramC = numpy.zeros((nGroups*4, nHeights,2))*numpy.nan |
|
|
4620 | 4620 | for i in range(nGroups): |
|
|
4621 | 4621 | coord = groupArray[i,:] |
|
|
4622 | 4622 | #Input data array |
|
|
4623 | 4623 | data = dataOut.data_spc[coord,:,:]/(M*N) |
|
|
4624 | 4624 | data = data.reshape((data.shape[0]*data.shape[1],data.shape[2])) |
|
|
4625 | 4625 | |
|
|
4626 | 4626 | #Cross Spectra data array for Covariance Matrixes |
|
|
4627 | 4627 | ind = 0 |
|
|
4628 | 4628 | for pairs in listComb: |
|
|
4629 | 4629 | pairsSel = numpy.array([coord[x],coord[y]]) |
|
|
4630 | 4630 | indCross[ind] = int(numpy.where(numpy.all(pairsArray == pairsSel, axis = 1))[0][0]) |
|
|
4631 | 4631 | ind += 1 |
|
|
4632 | 4632 | dataCross = dataOut.data_cspc[indCross,:,:]/(M*N) |
|
|
4633 | 4633 | dataCross = dataCross**2 |
|
|
4634 | 4634 | nhei = nHeights |
|
|
4635 | 4635 | poweri = numpy.sum(dataOut.data_spc[:,1:nProf-0,:],axis=1)/clean_num_aver[:,:] |
|
|
4636 | 4636 | if i == 0 : my_noises = numpy.zeros(4,dtype=float) #FLTARR(4) |
|
|
4637 | 4637 | n0i = numpy.nanmin(poweri[0+i*2,0:nhei-0])/(nProf-1) |
|
|
4638 | 4638 | n1i = numpy.nanmin(poweri[1+i*2,0:nhei-0])/(nProf-1) |
|
|
4639 | 4639 | n0 = n0i |
|
|
4640 | 4640 | n1= n1i |
|
|
4641 | 4641 | my_noises[2*i+0] = n0 |
|
|
4642 | 4642 | my_noises[2*i+1] = n1 |
|
|
4643 | 4643 | snrth = -16.0 |
|
|
4644 | 4644 | snrth = 10**(snrth/10.0) |
|
|
4645 | 4645 | |
|
|
4646 | 4646 | for h in range(nHeights): |
|
|
4647 | 4647 | d = data[:,h] |
|
|
4648 | 4648 | smooth = clean_num_aver[i+1,h] #dataOut.data_spc[:,1:nProf-0,:] |
|
|
4649 | 4649 | signalpn0 = (dataOut.data_spc[i*2,1:(nProf-0),h])/smooth |
|
|
4650 | 4650 | signalpn1 = (dataOut.data_spc[i*2+1,1:(nProf-0),h])/smooth |
|
|
4651 | 4651 | signal0 = signalpn0-n0 |
|
|
4652 | 4652 | signal1 = signalpn1-n1 |
|
|
4653 | 4653 | snr0 = numpy.sum(signal0/n0)/(nProf-1) |
|
|
4654 | 4654 | snr1 = numpy.sum(signal1/n1)/(nProf-1) |
|
|
4655 | 4655 | if snr0 > snrth and snr1 > snrth and clean_num_aver[i+1,h] > 0 : |
|
|
4656 | 4656 | #Covariance Matrix |
|
|
4657 | 4657 | D = numpy.diag(d**2) |
|
|
4658 | 4658 | ind = 0 |
|
|
4659 | 4659 | for pairs in listComb: |
|
|
4660 | 4660 | #Coordinates in Covariance Matrix |
|
|
4661 | 4661 | x = pairs[0] |
|
|
4662 | 4662 | y = pairs[1] |
|
|
4663 | 4663 | #Channel Index |
|
|
4664 | 4664 | S12 = dataCross[ind,:,h] |
|
|
4665 | 4665 | D12 = numpy.diag(S12) |
|
|
4666 | 4666 | #Completing Covariance Matrix with Cross Spectras |
|
|
4667 | 4667 | D[x*N:(x+1)*N,y*N:(y+1)*N] = D12 |
|
|
4668 | 4668 | D[y*N:(y+1)*N,x*N:(x+1)*N] = D12 |
|
|
4669 | 4669 | ind += 1 |
|
|
4670 | 4670 | diagD = numpy.zeros(256) |
|
|
4671 | 4671 | if h == 17 : |
|
|
4672 | 4672 | for ii in range(256): diagD[ii] = D[ii,ii] |
|
|
4673 | 4673 | #Dinv=numpy.linalg.inv(D) |
|
|
4674 | 4674 | #L=numpy.linalg.cholesky(Dinv) |
|
|
4675 | 4675 | try: |
|
|
4676 | 4676 | Dinv=numpy.linalg.inv(D) |
|
|
4677 | 4677 | L=numpy.linalg.cholesky(Dinv) |
|
|
4678 | 4678 | except: |
|
|
4679 | 4679 | Dinv = D*numpy.nan |
|
|
4680 | 4680 | L= D*numpy.nan |
|
|
4681 | 4681 | LT=L.T |
|
|
4682 | 4682 | |
|
|
4683 | 4683 | dp = numpy.dot(LT,d) |
|
|
4684 | 4684 | |
|
|
4685 | 4685 | #Initial values |
|
|
4686 | 4686 | data_spc = dataOut.data_spc[coord,:,h] |
|
|
4687 | 4687 | |
|
|
4688 | 4688 | if (h>0)and(error1[3]<5): |
|
|
4689 | 4689 | p0 = dataOut.data_param[i,:,h-1] |
|
|
4690 | 4690 | else: |
|
|
4691 | 4691 | p0 = numpy.array(self.library.initialValuesFunction(data_spc, constants))# sin el i(data_spc, constants, i) |
|
|
4692 | 4692 | try: |
|
|
4693 | 4693 | #Least Squares |
|
|
4694 | 4694 | #print (dp,LT,constants) |
|
|
4695 | 4695 | #value =self.__residFunction(p0,dp,LT,constants) |
|
|
4696 | 4696 | #print ("valueREADY",value.shape, type(value)) |
|
|
4697 | 4697 | #optimize.leastsq(value) |
|
|
4698 | 4698 | minp,covp,infodict,mesg,ier = optimize.leastsq(self.__residFunction,p0,args=(dp,LT,constants),full_output=True) |
|
|
4699 | 4699 | #minp,covp = optimize.leastsq(self.__residFunction,p0,args=(dp,LT,constants)) |
|
|
4700 | 4700 | #Chi square error |
|
|
4701 | 4701 | #print(minp,covp.infodict,mesg,ier) |
|
|
4702 | 4702 | #print("REALIZA OPTIMIZ") |
|
|
4703 | 4703 | error0 = numpy.sum(infodict['fvec']**2)/(2*N) |
|
|
4704 | 4704 | #Error with Jacobian |
|
|
4705 | 4705 | error1 = self.library.errorFunction(minp,constants,LT) |
|
|
4706 | 4706 | # print self.__residFunction(p0,dp,LT, constants) |
|
|
4707 | 4707 | # print infodict['fvec'] |
|
|
4708 | 4708 | # print self.__residFunction(minp,dp,LT,constants) |
|
|
4709 | 4709 | |
|
|
4710 | 4710 | except: |
|
|
4711 | 4711 | minp = p0*numpy.nan |
|
|
4712 | 4712 | error0 = numpy.nan |
|
|
4713 | 4713 | error1 = p0*numpy.nan |
|
|
4714 | 4714 | #print ("EXCEPT 0000000000") |
|
|
4715 | 4715 | # s_sq = (self.__residFunction(minp,dp,LT,constants)).sum()/(len(dp)-len(p0)) |
|
|
4716 | 4716 | # covp = covp*s_sq |
|
|
4717 | 4717 | # #print("TRY___________________________________________1") |
|
|
4718 | 4718 | # error = [] |
|
|
4719 | 4719 | # for ip in range(len(minp)): |
|
|
4720 | 4720 | # try: |
|
|
4721 | 4721 | # error.append(numpy.absolute(covp[ip][ip])**0.5) |
|
|
4722 | 4722 | # except: |
|
|
4723 | 4723 | # error.append( 0.00 ) |
|
|
4724 | 4724 | else : |
|
|
4725 | 4725 | data_spc = dataOut.data_spc[coord,:,h] |
|
|
4726 | 4726 | p0 = numpy.array(self.library.initialValuesFunction(data_spc, constants)) |
|
|
4727 | 4727 | minp = p0*numpy.nan |
|
|
4728 | 4728 | error0 = numpy.nan |
|
|
4729 | 4729 | error1 = p0*numpy.nan |
|
|
4730 | 4730 | #Save |
|
|
4731 | 4731 | if dataOut.data_param is None: |
|
|
4732 | 4732 | dataOut.data_param = numpy.zeros((nGroups, p0.size, nHeights))*numpy.nan |
|
|
4733 | 4733 | dataOut.data_error = numpy.zeros((nGroups, p0.size + 1, nHeights))*numpy.nan |
|
|
4734 | 4734 | |
|
|
4735 | 4735 | dataOut.data_error[i,:,h] = numpy.hstack((error0,error1)) |
|
|
4736 | 4736 | dataOut.data_param[i,:,h] = minp |
|
|
4737 | 4737 | |
|
|
4738 | 4738 | for ht in range(nHeights-1) : |
|
|
4739 | 4739 | smooth = coh_num_aver[i+1,ht] #datc[0,ht,0,beam] |
|
|
4740 | 4740 | dataOut.data_paramC[4*i,ht,1] = smooth |
|
|
4741 | 4741 | signalpn0 = (coh_spectra[i*2 ,1:(nProf-0),ht])/smooth #coh_spectra |
|
|
4742 | 4742 | signalpn1 = (coh_spectra[i*2+1,1:(nProf-0),ht])/smooth |
|
|
4743 | 4743 | |
|
|
4744 | 4744 | #val0 = WHERE(signalpn0 > 0,cval0) |
|
|
4745 | 4745 | val0 = (signalpn0 > 0).nonzero() |
|
|
4746 | 4746 | val0 = val0[0] |
|
|
4747 | 4747 | #print('hvalid:',hvalid) |
|
|
4748 | 4748 | #print('valid', valid) |
|
|
4749 | 4749 | if len(val0) == 0 : val0_npoints = nProf |
|
|
4750 | 4750 | else : val0_npoints = len(val0) |
|
|
4751 | 4751 | |
|
|
4752 | 4752 | #val1 = WHERE(signalpn1 > 0,cval1) |
|
|
4753 | 4753 | val1 = (signalpn1 > 0).nonzero() |
|
|
4754 | 4754 | val1 = val1[0] |
|
|
4755 | 4755 | if len(val1) == 0 : val1_npoints = nProf |
|
|
4756 | 4756 | else : val1_npoints = len(val1) |
|
|
4757 | 4757 | |
|
|
4758 | 4758 | dataOut.data_paramC[0+4*i,ht,0] = numpy.sum((signalpn0/val0_npoints))/n0 |
|
|
4759 | 4759 | dataOut.data_paramC[1+4*i,ht,0] = numpy.sum((signalpn1/val1_npoints))/n1 |
|
|
4760 | 4760 | |
|
|
4761 | 4761 | signal0 = (signalpn0-n0) # > 0 |
|
|
4762 | 4762 | vali = (signal0 < 0).nonzero() |
|
|
4763 | 4763 | vali = vali[0] |
|
|
4764 | 4764 | if len(vali) > 0 : signal0[vali] = 0 |
|
|
4765 | 4765 | signal1 = (signalpn1-n1) #> 0 |
|
|
4766 | 4766 | vali = (signal1 < 0).nonzero() |
|
|
4767 | 4767 | vali = vali[0] |
|
|
4768 | 4768 | if len(vali) > 0 : signal1[vali] = 0 |
|
|
4769 | 4769 | snr0 = numpy.sum(signal0/n0)/(nProf-1) |
|
|
4770 | 4770 | snr1 = numpy.sum(signal1/n1)/(nProf-1) |
|
|
4771 | 4771 | doppler = absc[1:] |
|
|
4772 | 4772 | if snr0 >= snrth and snr1 >= snrth and smooth : |
|
|
4773 | 4773 | signalpn0_n0 = signalpn0 |
|
|
4774 | 4774 | signalpn0_n0[val0] = signalpn0[val0] - n0 |
|
|
4775 | 4775 | mom0 = self.moments(doppler,signalpn0-n0,nProf) |
|
|
4776 | 4776 | # sigtmp= numpy.transpose(numpy.tile(signalpn0, [4,1])) |
|
|
4777 | 4777 | # momt= self.__calculateMoments( sigtmp, doppler , n0 ) |
|
|
4778 | 4778 | signalpn1_n1 = signalpn1 |
|
|
4779 | 4779 | signalpn1_n1[val1] = signalpn1[val1] - n1 |
|
|
4780 | 4780 | mom1 = self.moments(doppler,signalpn1_n1,nProf) |
|
|
4781 | 4781 | dataOut.data_paramC[2+4*i,ht,0] = (mom0[0]+mom1[0])/2. |
|
|
4782 | 4782 | dataOut.data_paramC[3+4*i,ht,0] = (mom0[1]+mom1[1])/2. |
|
|
4783 | 4783 | # if graph == 1 : |
|
|
4784 | 4784 | # window, 13 |
|
|
4785 | 4785 | # plot,doppler,signalpn0 |
|
|
4786 | 4786 | # oplot,doppler,signalpn1,linest=1 |
|
|
4787 | 4787 | # oplot,mom0(0)*doppler/doppler,signalpn0 |
|
|
4788 | 4788 | # oplot,mom1(0)*doppler/doppler,signalpn1 |
|
|
4789 | 4789 | # print,interval/12.,beam,45+ht*15,snr0,snr1,mom0(0),mom1(0),mom0(1),mom1(1) |
|
|
4790 | 4790 | #ENDIF |
|
|
4791 | 4791 | #ENDIF |
|
|
4792 | 4792 | #ENDFOR End height |
|
|
4793 | 4793 | |
|
|
4794 | 4794 | dataOut.data_spc = jspectra |
|
|
4795 | 4795 | if getSNR: |
|
|
4796 | 4796 | listChannels = groupArray.reshape((groupArray.size)) |
|
|
4797 | 4797 | listChannels.sort() |
|
|
4798 | 4798 | |
|
|
4799 | 4799 | dataOut.data_snr = self.__getSNR(dataOut.data_spc[listChannels,:,:], my_noises[listChannels]) |
|
|
4800 | 4800 | return dataOut |
|
|
4801 | 4801 | |
|
|
4802 | 4802 | def __residFunction(self, p, dp, LT, constants): |
|
|
4803 | 4803 | |
|
|
4804 | 4804 | fm = self.library.modelFunction(p, constants) |
|
|
4805 | 4805 | fmp=numpy.dot(LT,fm) |
|
|
4806 | 4806 | return dp-fmp |
|
|
4807 | 4807 | |
|
|
4808 | 4808 | def __getSNR(self, z, noise): |
|
|
4809 | 4809 | |
|
|
4810 | 4810 | avg = numpy.average(z, axis=1) |
|
|
4811 | 4811 | SNR = (avg.T-noise)/noise |
|
|
4812 | 4812 | SNR = SNR.T |
|
|
4813 | 4813 | return SNR |
|
|
4814 | 4814 | |
|
|
4815 | 4815 | def __chisq(self, p, chindex, hindex): |
|
|
4816 | 4816 | #similar to Resid but calculates CHI**2 |
|
|
4817 | 4817 | [LT,d,fm]=setupLTdfm(p,chindex,hindex) |
|
|
4818 | 4818 | dp=numpy.dot(LT,d) |
|
|
4819 | 4819 | fmp=numpy.dot(LT,fm) |
|
|
4820 | 4820 | chisq=numpy.dot((dp-fmp).T,(dp-fmp)) |
|
|
4821 | 4821 | return chisq |
|
|
4822 | 4822 | |
|
|
4823 | 4823 | class WindProfiler(Operation): |
|
|
4824 | 4824 | |
|
|
4825 | 4825 | __isConfig = False |
|
|
4826 | 4826 | |
|
|
4827 | 4827 | __initime = None |
|
|
4828 | 4828 | __lastdatatime = None |
|
|
4829 | 4829 | __integrationtime = None |
|
|
4830 | 4830 | |
|
|
4831 | 4831 | __buffer = None |
|
|
4832 | 4832 | |
|
|
4833 | 4833 | __dataReady = False |
|
|
4834 | 4834 | |
|
|
4835 | 4835 | __firstdata = None |
|
|
4836 | 4836 | |
|
|
4837 | 4837 | n = None |
|
|
4838 | 4838 | |
|
|
4839 | 4839 | def __init__(self): |
|
|
4840 | 4840 | Operation.__init__(self) |
|
|
4841 | 4841 | |
|
|
4842 | 4842 | def __calculateCosDir(self, elev, azim): |
|
|
4843 | 4843 | zen = (90 - elev)*numpy.pi/180 |
|
|
4844 | 4844 | azim = azim*numpy.pi/180 |
|
|
4845 | 4845 | cosDirX = numpy.sqrt((1-numpy.cos(zen)**2)/((1+numpy.tan(azim)**2))) |
|
|
4846 | 4846 | cosDirY = numpy.sqrt(1-numpy.cos(zen)**2-cosDirX**2) |
|
|
4847 | 4847 | |
|
|
4848 | 4848 | signX = numpy.sign(numpy.cos(azim)) |
|
|
4849 | 4849 | signY = numpy.sign(numpy.sin(azim)) |
|
|
4850 | 4850 | |
|
|
4851 | 4851 | cosDirX = numpy.copysign(cosDirX, signX) |
|
|
4852 | 4852 | cosDirY = numpy.copysign(cosDirY, signY) |
|
|
4853 | 4853 | return cosDirX, cosDirY |
|
|
4854 | 4854 | |
|
|
4855 | 4855 | def __calculateAngles(self, theta_x, theta_y, azimuth): |
|
|
4856 | 4856 | |
|
|
4857 | 4857 | dir_cosw = numpy.sqrt(1-theta_x**2-theta_y**2) |
|
|
4858 | 4858 | zenith_arr = numpy.arccos(dir_cosw) |
|
|
4859 | 4859 | azimuth_arr = numpy.arctan2(theta_x,theta_y) + azimuth*math.pi/180 |
|
|
4860 | 4860 | |
|
|
4861 | 4861 | dir_cosu = numpy.sin(azimuth_arr)*numpy.sin(zenith_arr) |
|
|
4862 | 4862 | dir_cosv = numpy.cos(azimuth_arr)*numpy.sin(zenith_arr) |
|
|
4863 | 4863 | |
|
|
4864 | 4864 | return azimuth_arr, zenith_arr, dir_cosu, dir_cosv, dir_cosw |
|
|
4865 | 4865 | |
|
|
4866 | 4866 | def __calculateMatA(self, dir_cosu, dir_cosv, dir_cosw, horOnly): |
|
|
4867 | 4867 | |
|
|
4868 | 4868 | if horOnly: |
|
|
4869 | 4869 | A = numpy.c_[dir_cosu,dir_cosv] |
|
|
4870 | 4870 | else: |
|
|
4871 | 4871 | A = numpy.c_[dir_cosu,dir_cosv,dir_cosw] |
|
|
4872 | 4872 | A = numpy.asmatrix(A) |
|
|
4873 | 4873 | A1 = numpy.linalg.inv(A.transpose()*A)*A.transpose() |
|
|
4874 | 4874 | |
|
|
4875 | 4875 | return A1 |
|
|
4876 | 4876 | |
|
|
4877 | 4877 | def __correctValues(self, heiRang, phi, velRadial, SNR): |
|
|
4878 | 4878 | listPhi = phi.tolist() |
|
|
4879 | 4879 | maxid = listPhi.index(max(listPhi)) |
|
|
4880 | 4880 | minid = listPhi.index(min(listPhi)) |
|
|
4881 | 4881 | |
|
|
4882 | 4882 | rango = list(range(len(phi))) |
|
|
4883 | 4883 | # rango = numpy.delete(rango,maxid) |
|
|
4884 | 4884 | |
|
|
4885 | 4885 | heiRang1 = heiRang*math.cos(phi[maxid]) |
|
|
4886 | 4886 | heiRangAux = heiRang*math.cos(phi[minid]) |
|
|
4887 | 4887 | indOut = (heiRang1 < heiRangAux[0]).nonzero() |
|
|
4888 | 4888 | heiRang1 = numpy.delete(heiRang1,indOut) |
|
|
4889 | 4889 | |
|
|
4890 | 4890 | velRadial1 = numpy.zeros([len(phi),len(heiRang1)]) |
|
|
4891 | 4891 | SNR1 = numpy.zeros([len(phi),len(heiRang1)]) |
|
|
4892 | 4892 | |
|
|
4893 | 4893 | for i in rango: |
|
|
4894 | 4894 | x = heiRang*math.cos(phi[i]) |
|
|
4895 | 4895 | y1 = velRadial[i,:] |
|
|
4896 | 4896 | f1 = interpolate.interp1d(x,y1,kind = 'cubic') |
|
|
4897 | 4897 | |
|
|
4898 | 4898 | x1 = heiRang1 |
|
|
4899 | 4899 | y11 = f1(x1) |
|
|
4900 | 4900 | |
|
|
4901 | 4901 | y2 = SNR[i,:] |
|
|
4902 | 4902 | f2 = interpolate.interp1d(x,y2,kind = 'cubic') |
|
|
4903 | 4903 | y21 = f2(x1) |
|
|
4904 | 4904 | |
|
|
4905 | 4905 | velRadial1[i,:] = y11 |
|
|
4906 | 4906 | SNR1[i,:] = y21 |
|
|
4907 | 4907 | |
|
|
4908 | 4908 | return heiRang1, velRadial1, SNR1 |
|
|
4909 | 4909 | |
|
|
4910 | 4910 | def __calculateVelUVW(self, A, velRadial): |
|
|
4911 | 4911 | |
|
|
4912 | 4912 | #Operacion Matricial |
|
|
4913 | 4913 | # velUVW = numpy.zeros((velRadial.shape[1],3)) |
|
|
4914 | 4914 | # for ind in range(velRadial.shape[1]): |
|
|
4915 | 4915 | # velUVW[ind,:] = numpy.dot(A,velRadial[:,ind]) |
|
|
4916 | 4916 | # velUVW = velUVW.transpose() |
|
|
4917 | 4917 | velUVW = numpy.zeros((A.shape[0],velRadial.shape[1])) |
|
|
4918 | 4918 | velUVW[:,:] = numpy.dot(A,velRadial) |
|
|
4919 | 4919 | |
|
|
4920 | 4920 | |
|
|
4921 | 4921 | return velUVW |
|
|
4922 | 4922 | |
|
|
4923 | 4923 | # def techniqueDBS(self, velRadial0, dirCosx, disrCosy, azimuth, correct, horizontalOnly, heiRang, SNR0): |
|
|
4924 | 4924 | |
|
|
4925 | 4925 | def techniqueDBS(self, kwargs): |
|
|
4926 | 4926 | """ |
|
|
4927 | 4927 | Function that implements Doppler Beam Swinging (DBS) technique. |
|
|
4928 | 4928 | |
|
|
4929 | 4929 | Input: Radial velocities, Direction cosines (x and y) of the Beam, Antenna azimuth, |
|
|
4930 | 4930 | Direction correction (if necessary), Ranges and SNR |
|
|
4931 | 4931 | |
|
|
4932 | 4932 | Output: Winds estimation (Zonal, Meridional and Vertical) |
|
|
4933 | 4933 | |
|
|
4934 | 4934 | Parameters affected: Winds, height range, SNR |
|
|
4935 | 4935 | """ |
|
|
4936 | 4936 | velRadial0 = kwargs['velRadial'] |
|
|
4937 | 4937 | heiRang = kwargs['heightList'] |
|
|
4938 | 4938 | SNR0 = kwargs['SNR'] |
|
|
4939 | 4939 | |
|
|
4940 | 4940 | if 'dirCosx' in kwargs and 'dirCosy' in kwargs: |
|
|
4941 | 4941 | theta_x = numpy.array(kwargs['dirCosx']) |
|
|
4942 | 4942 | theta_y = numpy.array(kwargs['dirCosy']) |
|
|
4943 | 4943 | else: |
|
|
4944 | 4944 | elev = numpy.array(kwargs['elevation']) |
|
|
4945 | 4945 | azim = numpy.array(kwargs['azimuth']) |
|
|
4946 | 4946 | theta_x, theta_y = self.__calculateCosDir(elev, azim) |
|
|
4947 | 4947 | azimuth = kwargs['correctAzimuth'] |
|
|
4948 | 4948 | if 'horizontalOnly' in kwargs: |
|
|
4949 | 4949 | horizontalOnly = kwargs['horizontalOnly'] |
|
|
4950 | 4950 | else: horizontalOnly = False |
|
|
4951 | 4951 | if 'correctFactor' in kwargs: |
|
|
4952 | 4952 | correctFactor = kwargs['correctFactor'] |
|
|
4953 | 4953 | else: correctFactor = 1 |
|
|
4954 | 4954 | if 'channelList' in kwargs: |
|
|
4955 | 4955 | channelList = kwargs['channelList'] |
|
|
4956 | 4956 | if len(channelList) == 2: |
|
|
4957 | 4957 | horizontalOnly = True |
|
|
4958 | 4958 | arrayChannel = numpy.array(channelList) |
|
|
4959 | 4959 | param = param[arrayChannel,:,:] |
|
|
4960 | 4960 | theta_x = theta_x[arrayChannel] |
|
|
4961 | 4961 | theta_y = theta_y[arrayChannel] |
|
|
4962 | 4962 | |
|
|
4963 | 4963 | azimuth_arr, zenith_arr, dir_cosu, dir_cosv, dir_cosw = self.__calculateAngles(theta_x, theta_y, azimuth) |
|
|
4964 | 4964 | heiRang1, velRadial1, SNR1 = self.__correctValues(heiRang, zenith_arr, correctFactor*velRadial0, SNR0) |
|
|
4965 | 4965 | A = self.__calculateMatA(dir_cosu, dir_cosv, dir_cosw, horizontalOnly) |
|
|
4966 | 4966 | |
|
|
4967 | 4967 | #Calculo de Componentes de la velocidad con DBS |
|
|
4968 | 4968 | winds = self.__calculateVelUVW(A,velRadial1) |
|
|
4969 | 4969 | |
|
|
4970 | 4970 | return winds, heiRang1, SNR1 |
|
|
4971 | 4971 | |
|
|
4972 | 4972 | def __calculateDistance(self, posx, posy, pairs_ccf, azimuth = None): |
|
|
4973 | 4973 | |
|
|
4974 | 4974 | nPairs = len(pairs_ccf) |
|
|
4975 | 4975 | posx = numpy.asarray(posx) |
|
|
4976 | 4976 | posy = numpy.asarray(posy) |
|
|
4977 | 4977 | |
|
|
4978 | 4978 | #Rotacion Inversa para alinear con el azimuth |
|
|
4979 | 4979 | if azimuth!= None: |
|
|
4980 | 4980 | azimuth = azimuth*math.pi/180 |
|
|
4981 | 4981 | posx1 = posx*math.cos(azimuth) + posy*math.sin(azimuth) |
|
|
4982 | 4982 | posy1 = -posx*math.sin(azimuth) + posy*math.cos(azimuth) |
|
|
4983 | 4983 | else: |
|
|
4984 | 4984 | posx1 = posx |
|
|
4985 | 4985 | posy1 = posy |
|
|
4986 | 4986 | |
|
|
4987 | 4987 | #Calculo de Distancias |
|
|
4988 | 4988 | distx = numpy.zeros(nPairs) |
|
|
4989 | 4989 | disty = numpy.zeros(nPairs) |
|
|
4990 | 4990 | dist = numpy.zeros(nPairs) |
|
|
4991 | 4991 | ang = numpy.zeros(nPairs) |
|
|
4992 | 4992 | |
|
|
4993 | 4993 | for i in range(nPairs): |
|
|
4994 | 4994 | distx[i] = posx1[pairs_ccf[i][1]] - posx1[pairs_ccf[i][0]] |
|
|
4995 | 4995 | disty[i] = posy1[pairs_ccf[i][1]] - posy1[pairs_ccf[i][0]] |
|
|
4996 | 4996 | dist[i] = numpy.sqrt(distx[i]**2 + disty[i]**2) |
|
|
4997 | 4997 | ang[i] = numpy.arctan2(disty[i],distx[i]) |
|
|
4998 | 4998 | |
|
|
4999 | 4999 | return distx, disty, dist, ang |
|
|
5000 | 5000 | #Calculo de Matrices |
|
|
5001 | 5001 | # nPairs = len(pairs) |
|
|
5002 | 5002 | # ang1 = numpy.zeros((nPairs, 2, 1)) |
|
|
5003 | 5003 | # dist1 = numpy.zeros((nPairs, 2, 1)) |
|
|
5004 | 5004 | # |
|
|
5005 | 5005 | # for j in range(nPairs): |
|
|
5006 | 5006 | # dist1[j,0,0] = dist[pairs[j][0]] |
|
|
5007 | 5007 | # dist1[j,1,0] = dist[pairs[j][1]] |
|
|
5008 | 5008 | # ang1[j,0,0] = ang[pairs[j][0]] |
|
|
5009 | 5009 | # ang1[j,1,0] = ang[pairs[j][1]] |
|
|
5010 | 5010 | # |
|
|
5011 | 5011 | # return distx,disty, dist1,ang1 |
|
|
5012 | 5012 | |
|
|
5013 | 5013 | |
|
|
5014 | 5014 | def __calculateVelVer(self, phase, lagTRange, _lambda): |
|
|
5015 | 5015 | |
|
|
5016 | 5016 | Ts = lagTRange[1] - lagTRange[0] |
|
|
5017 | 5017 | velW = -_lambda*phase/(4*math.pi*Ts) |
|
|
5018 | 5018 | |
|
|
5019 | 5019 | return velW |
|
|
5020 | 5020 | |
|
|
5021 | 5021 | def __calculateVelHorDir(self, dist, tau1, tau2, ang): |
|
|
5022 | 5022 | nPairs = tau1.shape[0] |
|
|
5023 | 5023 | nHeights = tau1.shape[1] |
|
|
5024 | 5024 | vel = numpy.zeros((nPairs,3,nHeights)) |
|
|
5025 | 5025 | dist1 = numpy.reshape(dist, (dist.size,1)) |
|
|
5026 | 5026 | |
|
|
5027 | 5027 | angCos = numpy.cos(ang) |
|
|
5028 | 5028 | angSin = numpy.sin(ang) |
|
|
5029 | 5029 | |
|
|
5030 | 5030 | vel0 = dist1*tau1/(2*tau2**2) |
|
|
5031 | 5031 | vel[:,0,:] = (vel0*angCos).sum(axis = 1) |
|
|
5032 | 5032 | vel[:,1,:] = (vel0*angSin).sum(axis = 1) |
|
|
5033 | 5033 | |
|
|
5034 | 5034 | ind = numpy.where(numpy.isinf(vel)) |
|
|
5035 | 5035 | vel[ind] = numpy.nan |
|
|
5036 | 5036 | |
|
|
5037 | 5037 | return vel |
|
|
5038 | 5038 | |
|
|
5039 | 5039 | # def __getPairsAutoCorr(self, pairsList, nChannels): |
|
|
5040 | 5040 | # |
|
|
5041 | 5041 | # pairsAutoCorr = numpy.zeros(nChannels, dtype = 'int')*numpy.nan |
|
|
5042 | 5042 | # |
|
|
5043 | 5043 | # for l in range(len(pairsList)): |
|
|
5044 | 5044 | # firstChannel = pairsList[l][0] |
|
|
5045 | 5045 | # secondChannel = pairsList[l][1] |
|
|
5046 | 5046 | # |
|
|
5047 | 5047 | # #Obteniendo pares de Autocorrelacion |
|
|
5048 | 5048 | # if firstChannel == secondChannel: |
|
|
5049 | 5049 | # pairsAutoCorr[firstChannel] = int(l) |
|
|
5050 | 5050 | # |
|
|
5051 | 5051 | # pairsAutoCorr = pairsAutoCorr.astype(int) |
|
|
5052 | 5052 | # |
|
|
5053 | 5053 | # pairsCrossCorr = range(len(pairsList)) |
|
|
5054 | 5054 | # pairsCrossCorr = numpy.delete(pairsCrossCorr,pairsAutoCorr) |
|
|
5055 | 5055 | # |
|
|
5056 | 5056 | # return pairsAutoCorr, pairsCrossCorr |
|
|
5057 | 5057 | |
|
|
5058 | 5058 | # def techniqueSA(self, pairsSelected, pairsList, nChannels, tau, azimuth, _lambda, position_x, position_y, lagTRange, correctFactor): |
|
|
5059 | 5059 | def techniqueSA(self, kwargs): |
|
|
5060 | 5060 | |
|
|
5061 | 5061 | """ |
|
|
5062 | 5062 | Function that implements Spaced Antenna (SA) technique. |
|
|
5063 | 5063 | |
|
|
5064 | 5064 | Input: Radial velocities, Direction cosines (x and y) of the Beam, Antenna azimuth, |
|
|
5065 | 5065 | Direction correction (if necessary), Ranges and SNR |
|
|
5066 | 5066 | |
|
|
5067 | 5067 | Output: Winds estimation (Zonal, Meridional and Vertical) |
|
|
5068 | 5068 | |
|
|
5069 | 5069 | Parameters affected: Winds |
|
|
5070 | 5070 | """ |
|
|
5071 | 5071 | position_x = kwargs['positionX'] |
|
|
5072 | 5072 | position_y = kwargs['positionY'] |
|
|
5073 | 5073 | azimuth = kwargs['azimuth'] |
|
|
5074 | 5074 | |
|
|
5075 | 5075 | if 'correctFactor' in kwargs: |
|
|
5076 | 5076 | correctFactor = kwargs['correctFactor'] |
|
|
5077 | 5077 | else: |
|
|
5078 | 5078 | correctFactor = 1 |
|
|
5079 | 5079 | |
|
|
5080 | 5080 | groupList = kwargs['groupList'] |
|
|
5081 | 5081 | pairs_ccf = groupList[1] |
|
|
5082 | 5082 | tau = kwargs['tau'] |
|
|
5083 | 5083 | _lambda = kwargs['_lambda'] |
|
|
5084 | 5084 | |
|
|
5085 | 5085 | #Cross Correlation pairs obtained |
|
|
5086 | 5086 | # pairsAutoCorr, pairsCrossCorr = self.__getPairsAutoCorr(pairssList, nChannels) |
|
|
5087 | 5087 | # pairsArray = numpy.array(pairsList)[pairsCrossCorr] |
|
|
5088 | 5088 | # pairsSelArray = numpy.array(pairsSelected) |
|
|
5089 | 5089 | # pairs = [] |
|
|
5090 | 5090 | # |
|
|
5091 | 5091 | # #Wind estimation pairs obtained |
|
|
5092 | 5092 | # for i in range(pairsSelArray.shape[0]/2): |
|
|
5093 | 5093 | # ind1 = numpy.where(numpy.all(pairsArray == pairsSelArray[2*i], axis = 1))[0][0] |
|
|
5094 | 5094 | # ind2 = numpy.where(numpy.all(pairsArray == pairsSelArray[2*i + 1], axis = 1))[0][0] |
|
|
5095 | 5095 | # pairs.append((ind1,ind2)) |
|
|
5096 | 5096 | |
|
|
5097 | 5097 | indtau = tau.shape[0]/2 |
|
|
5098 | 5098 | tau1 = tau[:indtau,:] |
|
|
5099 | 5099 | tau2 = tau[indtau:-1,:] |
|
|
5100 | 5100 | # tau1 = tau1[pairs,:] |
|
|
5101 | 5101 | # tau2 = tau2[pairs,:] |
|
|
5102 | 5102 | phase1 = tau[-1,:] |
|
|
5103 | 5103 | |
|
|
5104 | 5104 | #--------------------------------------------------------------------- |
|
|
5105 | 5105 | #Metodo Directo |
|
|
5106 | 5106 | distx, disty, dist, ang = self.__calculateDistance(position_x, position_y, pairs_ccf,azimuth) |
|
|
5107 | 5107 | winds = self.__calculateVelHorDir(dist, tau1, tau2, ang) |
|
|
5108 | 5108 | winds = stats.nanmean(winds, axis=0) |
|
|
5109 | 5109 | #--------------------------------------------------------------------- |
|
|
5110 | 5110 | #Metodo General |
|
|
5111 | 5111 | # distx, disty, dist = self.calculateDistance(position_x,position_y,pairsCrossCorr, pairsList, azimuth) |
|
|
5112 | 5112 | # #Calculo Coeficientes de Funcion de Correlacion |
|
|
5113 | 5113 | # F,G,A,B,H = self.calculateCoef(tau1,tau2,distx,disty,n) |
|
|
5114 | 5114 | # #Calculo de Velocidades |
|
|
5115 | 5115 | # winds = self.calculateVelUV(F,G,A,B,H) |
|
|
5116 | 5116 | |
|
|
5117 | 5117 | #--------------------------------------------------------------------- |
|
|
5118 | 5118 | winds[2,:] = self.__calculateVelVer(phase1, lagTRange, _lambda) |
|
|
5119 | 5119 | winds = correctFactor*winds |
|
|
5120 | 5120 | return winds |
|
|
5121 | 5121 | |
|
|
5122 | 5122 | def __checkTime(self, currentTime, paramInterval, outputInterval): |
|
|
5123 | 5123 | |
|
|
5124 | 5124 | dataTime = currentTime + paramInterval |
|
|
5125 | 5125 | deltaTime = dataTime - self.__initime |
|
|
5126 | 5126 | |
|
|
5127 | 5127 | if deltaTime >= outputInterval or deltaTime < 0: |
|
|
5128 | 5128 | self.__dataReady = True |
|
|
5129 | 5129 | return |
|
|
5130 | 5130 | |
|
|
5131 | 5131 | def techniqueMeteors(self, arrayMeteor, meteorThresh, heightMin, heightMax): |
|
|
5132 | 5132 | ''' |
|
|
5133 | 5133 | Function that implements winds estimation technique with detected meteors. |
|
|
5134 | 5134 | |
|
|
5135 | 5135 | Input: Detected meteors, Minimum meteor quantity to wind estimation |
|
|
5136 | 5136 | |
|
|
5137 | 5137 | Output: Winds estimation (Zonal and Meridional) |
|
|
5138 | 5138 | |
|
|
5139 | 5139 | Parameters affected: Winds |
|
|
5140 | 5140 | ''' |
|
|
5141 | 5141 | #Settings |
|
|
5142 | 5142 | nInt = (heightMax - heightMin)/2 |
|
|
5143 | 5143 | nInt = int(nInt) |
|
|
5144 | 5144 | winds = numpy.zeros((2,nInt))*numpy.nan |
|
|
5145 | 5145 | |
|
|
5146 | 5146 | #Filter errors |
|
|
5147 | 5147 | error = numpy.where(arrayMeteor[:,-1] == 0)[0] |
|
|
5148 | 5148 | finalMeteor = arrayMeteor[error,:] |
|
|
5149 | 5149 | |
|
|
5150 | 5150 | #Meteor Histogram |
|
|
5151 | 5151 | finalHeights = finalMeteor[:,2] |
|
|
5152 | 5152 | hist = numpy.histogram(finalHeights, bins = nInt, range = (heightMin,heightMax)) |
|
|
5153 | 5153 | nMeteorsPerI = hist[0] |
|
|
5154 | 5154 | heightPerI = hist[1] |
|
|
5155 | 5155 | |
|
|
5156 | 5156 | #Sort of meteors |
|
|
5157 | 5157 | indSort = finalHeights.argsort() |
|
|
5158 | 5158 | finalMeteor2 = finalMeteor[indSort,:] |
|
|
5159 | 5159 | |
|
|
5160 | 5160 | # Calculating winds |
|
|
5161 | 5161 | ind1 = 0 |
|
|
5162 | 5162 | ind2 = 0 |
|
|
5163 | 5163 | |
|
|
5164 | 5164 | for i in range(nInt): |
|
|
5165 | 5165 | nMet = nMeteorsPerI[i] |
|
|
5166 | 5166 | ind1 = ind2 |
|
|
5167 | 5167 | ind2 = ind1 + nMet |
|
|
5168 | 5168 | |
|
|
5169 | 5169 | meteorAux = finalMeteor2[ind1:ind2,:] |
|
|
5170 | 5170 | |
|
|
5171 | 5171 | if meteorAux.shape[0] >= meteorThresh: |
|
|
5172 | 5172 | vel = meteorAux[:, 6] |
|
|
5173 | 5173 | zen = meteorAux[:, 4]*numpy.pi/180 |
|
|
5174 | 5174 | azim = meteorAux[:, 3]*numpy.pi/180 |
|
|
5175 | 5175 | |
|
|
5176 | 5176 | n = numpy.cos(zen) |
|
|
5177 | 5177 | # m = (1 - n**2)/(1 - numpy.tan(azim)**2) |
|
|
5178 | 5178 | # l = m*numpy.tan(azim) |
|
|
5179 | 5179 | l = numpy.sin(zen)*numpy.sin(azim) |
|
|
5180 | 5180 | m = numpy.sin(zen)*numpy.cos(azim) |
|
|
5181 | 5181 | |
|
|
5182 | 5182 | A = numpy.vstack((l, m)).transpose() |
|
|
5183 | 5183 | A1 = numpy.dot(numpy.linalg.inv( numpy.dot(A.transpose(),A) ),A.transpose()) |
|
|
5184 | 5184 | windsAux = numpy.dot(A1, vel) |
|
|
5185 | 5185 | |
|
|
5186 | 5186 | winds[0,i] = windsAux[0] |
|
|
5187 | 5187 | winds[1,i] = windsAux[1] |
|
|
5188 | 5188 | |
|
|
5189 | 5189 | return winds, heightPerI[:-1] |
|
|
5190 | 5190 | |
|
|
5191 | 5191 | def techniqueNSM_SA(self, **kwargs): |
|
|
5192 | 5192 | metArray = kwargs['metArray'] |
|
|
5193 | 5193 | heightList = kwargs['heightList'] |
|
|
5194 | 5194 | timeList = kwargs['timeList'] |
|
|
5195 | 5195 | |
|
|
5196 | 5196 | rx_location = kwargs['rx_location'] |
|
|
5197 | 5197 | groupList = kwargs['groupList'] |
|
|
5198 | 5198 | azimuth = kwargs['azimuth'] |
|
|
5199 | 5199 | dfactor = kwargs['dfactor'] |
|
|
5200 | 5200 | k = kwargs['k'] |
|
|
5201 | 5201 | |
|
|
5202 | 5202 | azimuth1, dist = self.__calculateAzimuth1(rx_location, groupList, azimuth) |
|
|
5203 | 5203 | d = dist*dfactor |
|
|
5204 | 5204 | #Phase calculation |
|
|
5205 | 5205 | metArray1 = self.__getPhaseSlope(metArray, heightList, timeList) |
|
|
5206 | 5206 | |
|
|
5207 | 5207 | metArray1[:,-2] = metArray1[:,-2]*metArray1[:,2]*1000/(k*d[metArray1[:,1].astype(int)]) #angles into velocities |
|
|
5208 | 5208 | |
|
|
5209 | 5209 | velEst = numpy.zeros((heightList.size,2))*numpy.nan |
|
|
5210 | 5210 | azimuth1 = azimuth1*numpy.pi/180 |
|
|
5211 | 5211 | |
|
|
5212 | 5212 | for i in range(heightList.size): |
|
|
5213 | 5213 | h = heightList[i] |
|
|
5214 | 5214 | indH = numpy.where((metArray1[:,2] == h)&(numpy.abs(metArray1[:,-2]) < 100))[0] |
|
|
5215 | 5215 | metHeight = metArray1[indH,:] |
|
|
5216 | 5216 | if metHeight.shape[0] >= 2: |
|
|
5217 | 5217 | velAux = numpy.asmatrix(metHeight[:,-2]).T #Radial Velocities |
|
|
5218 | 5218 | iazim = metHeight[:,1].astype(int) |
|
|
5219 | 5219 | azimAux = numpy.asmatrix(azimuth1[iazim]).T #Azimuths |
|
|
5220 | 5220 | A = numpy.hstack((numpy.cos(azimAux),numpy.sin(azimAux))) |
|
|
5221 | 5221 | A = numpy.asmatrix(A) |
|
|
5222 | 5222 | A1 = numpy.linalg.pinv(A.transpose()*A)*A.transpose() |
|
|
5223 | 5223 | velHor = numpy.dot(A1,velAux) |
|
|
5224 | 5224 | |
|
|
5225 | 5225 | velEst[i,:] = numpy.squeeze(velHor) |
|
|
5226 | 5226 | return velEst |
|
|
5227 | 5227 | |
|
|
5228 | 5228 | def __getPhaseSlope(self, metArray, heightList, timeList): |
|
|
5229 | 5229 | meteorList = [] |
|
|
5230 | 5230 | #utctime sec1 height SNR velRad ph0 ph1 ph2 coh0 coh1 coh2 |
|
|
5231 | 5231 | #Putting back together the meteor matrix |
|
|
5232 | 5232 | utctime = metArray[:,0] |
|
|
5233 | 5233 | uniqueTime = numpy.unique(utctime) |
|
|
5234 | 5234 | |
|
|
5235 | 5235 | phaseDerThresh = 0.5 |
|
|
5236 | 5236 | ippSeconds = timeList[1] - timeList[0] |
|
|
5237 | 5237 | sec = numpy.where(timeList>1)[0][0] |
|
|
5238 | 5238 | nPairs = metArray.shape[1] - 6 |
|
|
5239 | 5239 | nHeights = len(heightList) |
|
|
5240 | 5240 | |
|
|
5241 | 5241 | for t in uniqueTime: |
|
|
5242 | 5242 | metArray1 = metArray[utctime==t,:] |
|
|
5243 | 5243 | # phaseDerThresh = numpy.pi/4 #reducir Phase thresh |
|
|
5244 | 5244 | tmet = metArray1[:,1].astype(int) |
|
|
5245 | 5245 | hmet = metArray1[:,2].astype(int) |
|
|
5246 | 5246 | |
|
|
5247 | 5247 | metPhase = numpy.zeros((nPairs, heightList.size, timeList.size - 1)) |
|
|
5248 | 5248 | metPhase[:,:] = numpy.nan |
|
|
5249 | 5249 | metPhase[:,hmet,tmet] = metArray1[:,6:].T |
|
|
5250 | 5250 | |
|
|
5251 | 5251 | #Delete short trails |
|
|
5252 | 5252 | metBool = ~numpy.isnan(metPhase[0,:,:]) |
|
|
5253 | 5253 | heightVect = numpy.sum(metBool, axis = 1) |
|
|
5254 | 5254 | metBool[heightVect<sec,:] = False |
|
|
5255 | 5255 | metPhase[:,heightVect<sec,:] = numpy.nan |
|
|
5256 | 5256 | |
|
|
5257 | 5257 | #Derivative |
|
|
5258 | 5258 | metDer = numpy.abs(metPhase[:,:,1:] - metPhase[:,:,:-1]) |
|
|
5259 | 5259 | phDerAux = numpy.dstack((numpy.full((nPairs,nHeights,1), False, dtype=bool),metDer > phaseDerThresh)) |
|
|
5260 | 5260 | metPhase[phDerAux] = numpy.nan |
|
|
5261 | 5261 | |
|
|
5262 | 5262 | #--------------------------METEOR DETECTION ----------------------------------------- |
|
|
5263 | 5263 | indMet = numpy.where(numpy.any(metBool,axis=1))[0] |
|
|
5264 | 5264 | |
|
|
5265 | 5265 | for p in numpy.arange(nPairs): |
|
|
5266 | 5266 | phase = metPhase[p,:,:] |
|
|
5267 | 5267 | phDer = metDer[p,:,:] |
|
|
5268 | 5268 | |
|
|
5269 | 5269 | for h in indMet: |
|
|
5270 | 5270 | height = heightList[h] |
|
|
5271 | 5271 | phase1 = phase[h,:] #82 |
|
|
5272 | 5272 | phDer1 = phDer[h,:] |
|
|
5273 | 5273 | |
|
|
5274 | 5274 | phase1[~numpy.isnan(phase1)] = numpy.unwrap(phase1[~numpy.isnan(phase1)]) #Unwrap |
|
|
5275 | 5275 | |
|
|
5276 | 5276 | indValid = numpy.where(~numpy.isnan(phase1))[0] |
|
|
5277 | 5277 | initMet = indValid[0] |
|
|
5278 | 5278 | endMet = 0 |
|
|
5279 | 5279 | |
|
|
5280 | 5280 | for i in range(len(indValid)-1): |
|
|
5281 | 5281 | |
|
|
5282 | 5282 | #Time difference |
|
|
5283 | 5283 | inow = indValid[i] |
|
|
5284 | 5284 | inext = indValid[i+1] |
|
|
5285 | 5285 | idiff = inext - inow |
|
|
5286 | 5286 | #Phase difference |
|
|
5287 | 5287 | phDiff = numpy.abs(phase1[inext] - phase1[inow]) |
|
|
5288 | 5288 | |
|
|
5289 | 5289 | if idiff>sec or phDiff>numpy.pi/4 or inext==indValid[-1]: #End of Meteor |
|
|
5290 | 5290 | sizeTrail = inow - initMet + 1 |
|
|
5291 | 5291 | if sizeTrail>3*sec: #Too short meteors |
|
|
5292 | 5292 | x = numpy.arange(initMet,inow+1)*ippSeconds |
|
|
5293 | 5293 | y = phase1[initMet:inow+1] |
|
|
5294 | 5294 | ynnan = ~numpy.isnan(y) |
|
|
5295 | 5295 | x = x[ynnan] |
|
|
5296 | 5296 | y = y[ynnan] |
|
|
5297 | 5297 | slope, intercept, r_value, p_value, std_err = stats.linregress(x,y) |
|
|
5298 | 5298 | ylin = x*slope + intercept |
|
|
5299 | 5299 | rsq = r_value**2 |
|
|
5300 | 5300 | if rsq > 0.5: |
|
|
5301 | 5301 | vel = slope#*height*1000/(k*d) |
|
|
5302 | 5302 | estAux = numpy.array([utctime,p,height, vel, rsq]) |
|
|
5303 | 5303 | meteorList.append(estAux) |
|
|
5304 | 5304 | initMet = inext |
|
|
5305 | 5305 | metArray2 = numpy.array(meteorList) |
|
|
5306 | 5306 | |
|
|
5307 | 5307 | return metArray2 |
|
|
5308 | 5308 | |
|
|
5309 | 5309 | def __calculateAzimuth1(self, rx_location, pairslist, azimuth0): |
|
|
5310 | 5310 | |
|
|
5311 | 5311 | azimuth1 = numpy.zeros(len(pairslist)) |
|
|
5312 | 5312 | dist = numpy.zeros(len(pairslist)) |
|
|
5313 | 5313 | |
|
|
5314 | 5314 | for i in range(len(rx_location)): |
|
|
5315 | 5315 | ch0 = pairslist[i][0] |
|
|
5316 | 5316 | ch1 = pairslist[i][1] |
|
|
5317 | 5317 | |
|
|
5318 | 5318 | diffX = rx_location[ch0][0] - rx_location[ch1][0] |
|
|
5319 | 5319 | diffY = rx_location[ch0][1] - rx_location[ch1][1] |
|
|
5320 | 5320 | azimuth1[i] = numpy.arctan2(diffY,diffX)*180/numpy.pi |
|
|
5321 | 5321 | dist[i] = numpy.sqrt(diffX**2 + diffY**2) |
|
|
5322 | 5322 | |
|
|
5323 | 5323 | azimuth1 -= azimuth0 |
|
|
5324 | 5324 | return azimuth1, dist |
|
|
5325 | 5325 | |
|
|
5326 | 5326 | def techniqueNSM_DBS(self, **kwargs): |
|
|
5327 | 5327 | metArray = kwargs['metArray'] |
|
|
5328 | 5328 | heightList = kwargs['heightList'] |
|
|
5329 | 5329 | timeList = kwargs['timeList'] |
|
|
5330 | 5330 | azimuth = kwargs['azimuth'] |
|
|
5331 | 5331 | theta_x = numpy.array(kwargs['theta_x']) |
|
|
5332 | 5332 | theta_y = numpy.array(kwargs['theta_y']) |
|
|
5333 | 5333 | |
|
|
5334 | 5334 | utctime = metArray[:,0] |
|
|
5335 | 5335 | cmet = metArray[:,1].astype(int) |
|
|
5336 | 5336 | hmet = metArray[:,3].astype(int) |
|
|
5337 | 5337 | SNRmet = metArray[:,4] |
|
|
5338 | 5338 | vmet = metArray[:,5] |
|
|
5339 | 5339 | spcmet = metArray[:,6] |
|
|
5340 | 5340 | |
|
|
5341 | 5341 | nChan = numpy.max(cmet) + 1 |
|
|
5342 | 5342 | nHeights = len(heightList) |
|
|
5343 | 5343 | |
|
|
5344 | 5344 | azimuth_arr, zenith_arr, dir_cosu, dir_cosv, dir_cosw = self.__calculateAngles(theta_x, theta_y, azimuth) |
|
|
5345 | 5345 | hmet = heightList[hmet] |
|
|
5346 | 5346 | h1met = hmet*numpy.cos(zenith_arr[cmet]) #Corrected heights |
|
|
5347 | 5347 | |
|
|
5348 | 5348 | velEst = numpy.zeros((heightList.size,2))*numpy.nan |
|
|
5349 | 5349 | |
|
|
5350 | 5350 | for i in range(nHeights - 1): |
|
|
5351 | 5351 | hmin = heightList[i] |
|
|
5352 | 5352 | hmax = heightList[i + 1] |
|
|
5353 | 5353 | |
|
|
5354 | 5354 | thisH = (h1met>=hmin) & (h1met<hmax) & (cmet!=2) & (SNRmet>8) & (vmet<50) & (spcmet<10) |
|
|
5355 | 5355 | indthisH = numpy.where(thisH) |
|
|
5356 | 5356 | |
|
|
5357 | 5357 | if numpy.size(indthisH) > 3: |
|
|
5358 | 5358 | |
|
|
5359 | 5359 | vel_aux = vmet[thisH] |
|
|
5360 | 5360 | chan_aux = cmet[thisH] |
|
|
5361 | 5361 | cosu_aux = dir_cosu[chan_aux] |
|
|
5362 | 5362 | cosv_aux = dir_cosv[chan_aux] |
|
|
5363 | 5363 | cosw_aux = dir_cosw[chan_aux] |
|
|
5364 | 5364 | |
|
|
5365 | 5365 | nch = numpy.size(numpy.unique(chan_aux)) |
|
|
5366 | 5366 | if nch > 1: |
|
|
5367 | 5367 | A = self.__calculateMatA(cosu_aux, cosv_aux, cosw_aux, True) |
|
|
5368 | 5368 | velEst[i,:] = numpy.dot(A,vel_aux) |
|
|
5369 | 5369 | |
|
|
5370 | 5370 | return velEst |
|
|
5371 | 5371 | |
|
|
5372 | 5372 | def run(self, dataOut, technique, nHours=1, hmin=70, hmax=110, **kwargs): |
|
|
5373 | 5373 | |
|
|
5374 | 5374 | param = dataOut.data_param |
|
|
5375 | 5375 | if dataOut.abscissaList != None: |
|
|
5376 | 5376 | absc = dataOut.abscissaList[:-1] |
|
|
5377 | 5377 | # noise = dataOut.noise |
|
|
5378 | 5378 | heightList = dataOut.heightList |
|
|
5379 | 5379 | SNR = dataOut.data_snr |
|
|
5380 | 5380 | |
|
|
5381 | 5381 | if technique == 'DBS': |
|
|
5382 | 5382 | |
|
|
5383 | 5383 | kwargs['velRadial'] = param[:,1,:] #Radial velocity |
|
|
5384 | 5384 | kwargs['heightList'] = heightList |
|
|
5385 | 5385 | kwargs['SNR'] = SNR |
|
|
5386 | 5386 | |
|
|
5387 | 5387 | dataOut.data_output, dataOut.heightList, dataOut.data_snr = self.techniqueDBS(kwargs) #DBS Function |
|
|
5388 | 5388 | dataOut.utctimeInit = dataOut.utctime |
|
|
5389 | 5389 | dataOut.outputInterval = dataOut.paramInterval |
|
|
5390 | 5390 | |
|
|
5391 | 5391 | elif technique == 'SA': |
|
|
5392 | 5392 | |
|
|
5393 | 5393 | #Parameters |
|
|
5394 | 5394 | # position_x = kwargs['positionX'] |
|
|
5395 | 5395 | # position_y = kwargs['positionY'] |
|
|
5396 | 5396 | # azimuth = kwargs['azimuth'] |
|
|
5397 | 5397 | # |
|
|
5398 | 5398 | # if kwargs.has_key('crosspairsList'): |
|
|
5399 | 5399 | # pairs = kwargs['crosspairsList'] |
|
|
5400 | 5400 | # else: |
|
|
5401 | 5401 | # pairs = None |
|
|
5402 | 5402 | # |
|
|
5403 | 5403 | # if kwargs.has_key('correctFactor'): |
|
|
5404 | 5404 | # correctFactor = kwargs['correctFactor'] |
|
|
5405 | 5405 | # else: |
|
|
5406 | 5406 | # correctFactor = 1 |
|
|
5407 | 5407 | |
|
|
5408 | 5408 | # tau = dataOut.data_param |
|
|
5409 | 5409 | # _lambda = dataOut.C/dataOut.frequency |
|
|
5410 | 5410 | # pairsList = dataOut.groupList |
|
|
5411 | 5411 | # nChannels = dataOut.nChannels |
|
|
5412 | 5412 | |
|
|
5413 | 5413 | kwargs['groupList'] = dataOut.groupList |
|
|
5414 | 5414 | kwargs['tau'] = dataOut.data_param |
|
|
5415 | 5415 | kwargs['_lambda'] = dataOut.C/dataOut.frequency |
|
|
5416 | 5416 | # dataOut.data_output = self.techniqueSA(pairs, pairsList, nChannels, tau, azimuth, _lambda, position_x, position_y, absc, correctFactor) |
|
|
5417 | 5417 | dataOut.data_output = self.techniqueSA(kwargs) |
|
|
5418 | 5418 | dataOut.utctimeInit = dataOut.utctime |
|
|
5419 | 5419 | dataOut.outputInterval = dataOut.timeInterval |
|
|
5420 | 5420 | |
|
|
5421 | 5421 | elif technique == 'Meteors': |
|
|
5422 | 5422 | dataOut.flagNoData = True |
|
|
5423 | 5423 | self.__dataReady = False |
|
|
5424 | 5424 | |
|
|
5425 | 5425 | if 'nHours' in kwargs: |
|
|
5426 | 5426 | nHours = kwargs['nHours'] |
|
|
5427 | 5427 | else: |
|
|
5428 | 5428 | nHours = 1 |
|
|
5429 | 5429 | |
|
|
5430 | 5430 | if 'meteorsPerBin' in kwargs: |
|
|
5431 | 5431 | meteorThresh = kwargs['meteorsPerBin'] |
|
|
5432 | 5432 | else: |
|
|
5433 | 5433 | meteorThresh = 6 |
|
|
5434 | 5434 | |
|
|
5435 | 5435 | if 'hmin' in kwargs: |
|
|
5436 | 5436 | hmin = kwargs['hmin'] |
|
|
5437 | 5437 | else: hmin = 70 |
|
|
5438 | 5438 | if 'hmax' in kwargs: |
|
|
5439 | 5439 | hmax = kwargs['hmax'] |
|
|
5440 | 5440 | else: hmax = 110 |
|
|
5441 | 5441 | |
|
|
5442 | 5442 | dataOut.outputInterval = nHours*3600 |
|
|
5443 | 5443 | |
|
|
5444 | 5444 | if self.__isConfig == False: |
|
|
5445 | 5445 | # self.__initime = dataOut.datatime.replace(minute = 0, second = 0, microsecond = 03) |
|
|
5446 | 5446 | #Get Initial LTC time |
|
|
5447 | 5447 | self.__initime = datetime.datetime.utcfromtimestamp(dataOut.utctime) |
|
|
5448 | 5448 | self.__initime = (self.__initime.replace(minute = 0, second = 0, microsecond = 0) - datetime.datetime(1970, 1, 1)).total_seconds() |
|
|
5449 | 5449 | |
|
|
5450 | 5450 | self.__isConfig = True |
|
|
5451 | 5451 | |
|
|
5452 | 5452 | if self.__buffer is None: |
|
|
5453 | 5453 | self.__buffer = dataOut.data_param |
|
|
5454 | 5454 | self.__firstdata = copy.copy(dataOut) |
|
|
5455 | 5455 | |
|
|
5456 | 5456 | else: |
|
|
5457 | 5457 | self.__buffer = numpy.vstack((self.__buffer, dataOut.data_param)) |
|
|
5458 | 5458 | |
|
|
5459 | 5459 | self.__checkTime(dataOut.utctime, dataOut.paramInterval, dataOut.outputInterval) #Check if the buffer is ready |
|
|
5460 | 5460 | |
|
|
5461 | 5461 | if self.__dataReady: |
|
|
5462 | 5462 | dataOut.utctimeInit = self.__initime |
|
|
5463 | 5463 | |
|
|
5464 | 5464 | self.__initime += dataOut.outputInterval #to erase time offset |
|
|
5465 | 5465 | |
|
|
5466 | 5466 | dataOut.data_output, dataOut.heightList = self.techniqueMeteors(self.__buffer, meteorThresh, hmin, hmax) |
|
|
5467 | 5467 | dataOut.flagNoData = False |
|
|
5468 | 5468 | self.__buffer = None |
|
|
5469 | 5469 | |
|
|
5470 | 5470 | elif technique == 'Meteors1': |
|
|
5471 | 5471 | dataOut.flagNoData = True |
|
|
5472 | 5472 | self.__dataReady = False |
|
|
5473 | 5473 | |
|
|
5474 | 5474 | if 'nMins' in kwargs: |
|
|
5475 | 5475 | nMins = kwargs['nMins'] |
|
|
5476 | 5476 | else: nMins = 20 |
|
|
5477 | 5477 | if 'rx_location' in kwargs: |
|
|
5478 | 5478 | rx_location = kwargs['rx_location'] |
|
|
5479 | 5479 | else: rx_location = [(0,1),(1,1),(1,0)] |
|
|
5480 | 5480 | if 'azimuth' in kwargs: |
|
|
5481 | 5481 | azimuth = kwargs['azimuth'] |
|
|
5482 | 5482 | else: azimuth = 51.06 |
|
|
5483 | 5483 | if 'dfactor' in kwargs: |
|
|
5484 | 5484 | dfactor = kwargs['dfactor'] |
|
|
5485 | 5485 | if 'mode' in kwargs: |
|
|
5486 | 5486 | mode = kwargs['mode'] |
|
|
5487 | 5487 | if 'theta_x' in kwargs: |
|
|
5488 | 5488 | theta_x = kwargs['theta_x'] |
|
|
5489 | 5489 | if 'theta_y' in kwargs: |
|
|
5490 | 5490 | theta_y = kwargs['theta_y'] |
|
|
5491 | 5491 | else: mode = 'SA' |
|
|
5492 | 5492 | |
|
|
5493 | 5493 | #Borrar luego esto |
|
|
5494 | 5494 | if dataOut.groupList is None: |
|
|
5495 | 5495 | dataOut.groupList = [(0,1),(0,2),(1,2)] |
|
|
5496 | 5496 | groupList = dataOut.groupList |
|
|
5497 | 5497 | C = 3e8 |
|
|
5498 | 5498 | freq = 50e6 |
|
|
5499 | 5499 | lamb = C/freq |
|
|
5500 | 5500 | k = 2*numpy.pi/lamb |
|
|
5501 | 5501 | |
|
|
5502 | 5502 | timeList = dataOut.abscissaList |
|
|
5503 | 5503 | heightList = dataOut.heightList |
|
|
5504 | 5504 | |
|
|
5505 | 5505 | if self.__isConfig == False: |
|
|
5506 | 5506 | dataOut.outputInterval = nMins*60 |
|
|
5507 | 5507 | # self.__initime = dataOut.datatime.replace(minute = 0, second = 0, microsecond = 03) |
|
|
5508 | 5508 | #Get Initial LTC time |
|
|
5509 | 5509 | initime = datetime.datetime.utcfromtimestamp(dataOut.utctime) |
|
|
5510 | 5510 | minuteAux = initime.minute |
|
|
5511 | 5511 | minuteNew = int(numpy.floor(minuteAux/nMins)*nMins) |
|
|
5512 | 5512 | self.__initime = (initime.replace(minute = minuteNew, second = 0, microsecond = 0) - datetime.datetime(1970, 1, 1)).total_seconds() |
|
|
5513 | 5513 | |
|
|
5514 | 5514 | self.__isConfig = True |
|
|
5515 | 5515 | |
|
|
5516 | 5516 | if self.__buffer is None: |
|
|
5517 | 5517 | self.__buffer = dataOut.data_param |
|
|
5518 | 5518 | self.__firstdata = copy.copy(dataOut) |
|
|
5519 | 5519 | |
|
|
5520 | 5520 | else: |
|
|
5521 | 5521 | self.__buffer = numpy.vstack((self.__buffer, dataOut.data_param)) |
|
|
5522 | 5522 | |
|
|
5523 | 5523 | self.__checkTime(dataOut.utctime, dataOut.paramInterval, dataOut.outputInterval) #Check if the buffer is ready |
|
|
5524 | 5524 | |
|
|
5525 | 5525 | if self.__dataReady: |
|
|
5526 | 5526 | dataOut.utctimeInit = self.__initime |
|
|
5527 | 5527 | self.__initime += dataOut.outputInterval #to erase time offset |
|
|
5528 | 5528 | |
|
|
5529 | 5529 | metArray = self.__buffer |
|
|
5530 | 5530 | if mode == 'SA': |
|
|
5531 | 5531 | dataOut.data_output = self.techniqueNSM_SA(rx_location=rx_location, groupList=groupList, azimuth=azimuth, dfactor=dfactor, k=k,metArray=metArray, heightList=heightList,timeList=timeList) |
|
|
5532 | 5532 | elif mode == 'DBS': |
|
|
5533 | 5533 | dataOut.data_output = self.techniqueNSM_DBS(metArray=metArray,heightList=heightList,timeList=timeList, azimuth=azimuth, theta_x=theta_x, theta_y=theta_y) |
|
|
5534 | 5534 | dataOut.data_output = dataOut.data_output.T |
|
|
5535 | 5535 | dataOut.flagNoData = False |
|
|
5536 | 5536 | self.__buffer = None |
|
|
5537 | 5537 | |
|
|
5538 | 5538 | return |
|
|
5539 | 5539 | |
|
|
5540 | 5540 | class EWDriftsEstimation(Operation): |
|
|
5541 | 5541 | |
|
|
5542 | 5542 | def __init__(self): |
|
|
5543 | 5543 | Operation.__init__(self) |
|
|
5544 | 5544 | |
|
|
5545 | 5545 | def __correctValues(self, heiRang, phi, velRadial, SNR): |
|
|
5546 | 5546 | listPhi = phi.tolist() |
|
|
5547 | 5547 | maxid = listPhi.index(max(listPhi)) |
|
|
5548 | 5548 | minid = listPhi.index(min(listPhi)) |
|
|
5549 | 5549 | |
|
|
5550 | 5550 | rango = list(range(len(phi))) |
|
|
5551 | 5551 | # rango = numpy.delete(rango,maxid) |
|
|
5552 | 5552 | |
|
|
5553 | 5553 | heiRang1 = heiRang*math.cos(phi[maxid]) |
|
|
5554 | 5554 | heiRangAux = heiRang*math.cos(phi[minid]) |
|
|
5555 | 5555 | indOut = (heiRang1 < heiRangAux[0]).nonzero() |
|
|
5556 | 5556 | heiRang1 = numpy.delete(heiRang1,indOut) |
|
|
5557 | 5557 | |
|
|
5558 | 5558 | velRadial1 = numpy.zeros([len(phi),len(heiRang1)]) |
|
|
5559 | 5559 | SNR1 = numpy.zeros([len(phi),len(heiRang1)]) |
|
|
5560 | 5560 | |
|
|
5561 | 5561 | for i in rango: |
|
|
5562 | 5562 | x = heiRang*math.cos(phi[i]) |
|
|
5563 | 5563 | y1 = velRadial[i,:] |
|
|
5564 | 5564 | vali= (numpy.isfinite(y1)==True).nonzero() |
|
|
5565 | 5565 | y1=y1[vali] |
|
|
5566 | 5566 | x = x[vali] |
|
|
5567 | 5567 | f1 = interpolate.interp1d(x,y1,kind = 'cubic',bounds_error=False) |
|
|
5568 | 5568 | |
|
|
5569 | 5569 | #heiRang1 = x*math.cos(phi[maxid]) |
|
|
5570 | 5570 | x1 = heiRang1 |
|
|
5571 | 5571 | y11 = f1(x1) |
|
|
5572 | 5572 | |
|
|
5573 | 5573 | y2 = SNR[i,:] |
|
|
5574 | 5574 | #print 'snr ', y2 |
|
|
5575 | 5575 | x = heiRang*math.cos(phi[i]) |
|
|
5576 | 5576 | vali= (y2 != -1).nonzero() |
|
|
5577 | 5577 | y2 = y2[vali] |
|
|
5578 | 5578 | x = x[vali] |
|
|
5579 | 5579 | #print 'snr ',y2 |
|
|
5580 | 5580 | f2 = interpolate.interp1d(x,y2,kind = 'cubic',bounds_error=False) |
|
|
5581 | 5581 | y21 = f2(x1) |
|
|
5582 | 5582 | |
|
|
5583 | 5583 | velRadial1[i,:] = y11 |
|
|
5584 | 5584 | SNR1[i,:] = y21 |
|
|
5585 | 5585 | |
|
|
5586 | 5586 | return heiRang1, velRadial1, SNR1 |
|
|
5587 | 5587 | |
|
|
5588 | 5588 | |
|
|
5589 | 5589 | |
|
|
5590 | 5590 | def run(self, dataOut, zenith, zenithCorrection): |
|
|
5591 | 5591 | |
|
|
5592 | 5592 | heiRang = dataOut.heightList |
|
|
5593 | 5593 | velRadial = dataOut.data_param[:,3,:] |
|
|
5594 | 5594 | velRadialm = dataOut.data_param[:,2:4,:]*-1 |
|
|
5595 | 5595 | |
|
|
5596 | 5596 | rbufc=dataOut.data_paramC[:,:,0] |
|
|
5597 | 5597 | ebufc=dataOut.data_paramC[:,:,1] |
|
|
5598 | 5598 | SNR = dataOut.data_snr |
|
|
5599 | 5599 | velRerr = dataOut.data_error[:,4,:] |
|
|
5600 | 5600 | moments=numpy.vstack(([velRadialm[0,:]],[velRadialm[0,:]],[velRadialm[1,:]],[velRadialm[1,:]])) |
|
|
5601 | 5601 | dataOut.moments=moments |
|
|
5602 | 5602 | # Coherent |
|
|
5603 | 5603 | smooth_wC = ebufc[0,:] |
|
|
5604 | 5604 | p_w0C = rbufc[0,:] |
|
|
5605 | 5605 | p_w1C = rbufc[1,:] |
|
|
5606 | 5606 | w_wC = rbufc[2,:]*-1 #*radial_sign(radial EQ 1) |
|
|
5607 | 5607 | t_wC = rbufc[3,:] |
|
|
5608 | 5608 | my_nbeams = 2 |
|
|
5609 | 5609 | |
|
|
5610 | 5610 | zenith = numpy.array(zenith) |
|
|
5611 | 5611 | zenith -= zenithCorrection |
|
|
5612 | 5612 | zenith *= numpy.pi/180 |
|
|
5613 | 5613 | if zenithCorrection != 0 : |
|
|
5614 | 5614 | heiRang1, velRadial1, SNR1 = self.__correctValues(heiRang, numpy.abs(zenith), velRadial, SNR) |
|
|
5615 | 5615 | else : |
|
|
5616 | 5616 | heiRang1 = heiRang |
|
|
5617 | 5617 | velRadial1 = velRadial |
|
|
5618 | 5618 | SNR1 = SNR |
|
|
5619 | 5619 | |
|
|
5620 | 5620 | alp = zenith[0] |
|
|
5621 | 5621 | bet = zenith[1] |
|
|
5622 | 5622 | |
|
|
5623 | 5623 | w_w = velRadial1[0,:] |
|
|
5624 | 5624 | w_e = velRadial1[1,:] |
|
|
5625 | 5625 | w_w_err = velRerr[0,:] |
|
|
5626 | 5626 | w_e_err = velRerr[1,:] |
|
|
5627 | 5627 | |
|
|
5628 | 5628 | val = (numpy.isfinite(w_w)==False).nonzero() |
|
|
5629 | 5629 | val = val[0] |
|
|
5630 | 5630 | bad = val |
|
|
5631 | 5631 | if len(bad) > 0 : |
|
|
5632 | 5632 | w_w[bad] = w_wC[bad] |
|
|
5633 | 5633 | w_w_err[bad]= numpy.nan |
|
|
5634 | 5634 | if my_nbeams == 2: |
|
|
5635 | 5635 | smooth_eC=ebufc[4,:] |
|
|
5636 | 5636 | p_e0C = rbufc[4,:] |
|
|
5637 | 5637 | p_e1C = rbufc[5,:] |
|
|
5638 | 5638 | w_eC = rbufc[6,:]*-1 |
|
|
5639 | 5639 | t_eC = rbufc[7,:] |
|
|
5640 | 5640 | val = (numpy.isfinite(w_e)==False).nonzero() |
|
|
5641 | 5641 | val = val[0] |
|
|
5642 | 5642 | bad = val |
|
|
5643 | 5643 | if len(bad) > 0 : |
|
|
5644 | 5644 | w_e[bad] = w_eC[bad] |
|
|
5645 | 5645 | w_e_err[bad]= numpy.nan |
|
|
5646 | 5646 | |
|
|
5647 | 5647 | w = (w_w*numpy.sin(bet) - w_e*numpy.sin(alp))/(numpy.cos(alp)*numpy.sin(bet) - numpy.cos(bet)*numpy.sin(alp)) |
|
|
5648 | 5648 | u = (w_w*numpy.cos(bet) - w_e*numpy.cos(alp))/(numpy.sin(alp)*numpy.cos(bet) - numpy.sin(bet)*numpy.cos(alp)) |
|
|
5649 | 5649 | |
|
|
5650 | 5650 | w_err = numpy.sqrt((w_w_err*numpy.sin(bet))**2.+(w_e_err*numpy.sin(alp))**2.)/ numpy.absolute(numpy.cos(alp)*numpy.sin(bet)-numpy.cos(bet)*numpy.sin(alp)) |
|
|
5651 | 5651 | u_err = numpy.sqrt((w_w_err*numpy.cos(bet))**2.+(w_e_err*numpy.cos(alp))**2.)/ numpy.absolute(numpy.cos(alp)*numpy.sin(bet)-numpy.cos(bet)*numpy.sin(alp)) |
|
|
5652 | 5652 | |
|
|
5653 | 5653 | winds = numpy.vstack((w,u)) |
|
|
5654 | 5654 | |
|
|
5655 | 5655 | dataOut.heightList = heiRang1 |
|
|
5656 | 5656 | dataOut.data_output = winds |
|
|
5657 | 5657 | |
|
|
5658 | 5658 | snr1 = 10*numpy.log10(SNR1[0]) |
|
|
5659 | 5659 | dataOut.data_snr1 = numpy.reshape(snr1,(1,snr1.shape[0])) |
|
|
5660 | 5660 | dataOut.utctimeInit = dataOut.utctime |
|
|
5661 | 5661 | dataOut.outputInterval = dataOut.timeInterval |
|
|
5662 | 5662 | |
|
|
5663 | 5663 | hei_aver0 = 218 |
|
|
5664 | 5664 | jrange = 450 #900 para HA drifts |
|
|
5665 | 5665 | deltah = 15.0 #dataOut.spacing(0) |
|
|
5666 | 5666 | h0 = 0.0 #dataOut.first_height(0) |
|
|
5667 | 5667 | heights = dataOut.heightList |
|
|
5668 | 5668 | nhei = len(heights) |
|
|
5669 | 5669 | |
|
|
5670 | 5670 | range1 = numpy.arange(nhei) * deltah + h0 |
|
|
5671 | 5671 | |
|
|
5672 | 5672 | #jhei = WHERE(range1 GE hei_aver0 , jcount) |
|
|
5673 | 5673 | jhei = (range1 >= hei_aver0).nonzero() |
|
|
5674 | 5674 | if len(jhei[0]) > 0 : |
|
|
5675 | 5675 | h0_index = jhei[0][0] # Initial height for getting averages 218km |
|
|
5676 | 5676 | |
|
|
5677 | 5677 | mynhei = 7 |
|
|
5678 | 5678 | nhei_avg = int(jrange/deltah) |
|
|
5679 | 5679 | h_avgs = int(nhei_avg/mynhei) |
|
|
5680 | 5680 | nhei_avg = h_avgs*(mynhei-1)+mynhei |
|
|
5681 | 5681 | |
|
|
5682 | 5682 | navgs = numpy.zeros(mynhei,dtype='float') |
|
|
5683 | 5683 | delta_h = numpy.zeros(mynhei,dtype='float') |
|
|
5684 | 5684 | range_aver = numpy.zeros(mynhei,dtype='float') |
|
|
5685 | 5685 | for ih in range( mynhei-1 ): |
|
|
5686 | 5686 | range_aver[ih] = numpy.sum(range1[h0_index+h_avgs*ih:h0_index+h_avgs*(ih+1)-0])/h_avgs |
|
|
5687 | 5687 | navgs[ih] = h_avgs |
|
|
5688 | 5688 | delta_h[ih] = deltah*h_avgs |
|
|
5689 | 5689 | |
|
|
5690 | 5690 | range_aver[mynhei-1] = numpy.sum(range1[h0_index:h0_index+6*h_avgs-0])/(6*h_avgs) |
|
|
5691 | 5691 | navgs[mynhei-1] = 6*h_avgs |
|
|
5692 | 5692 | delta_h[mynhei-1] = deltah*6*h_avgs |
|
|
5693 | 5693 | |
|
|
5694 | 5694 | wA = w[h0_index:h0_index+nhei_avg-0] |
|
|
5695 | 5695 | wA_err = w_err[h0_index:h0_index+nhei_avg-0] |
|
|
5696 | 5696 | |
|
|
5697 | 5697 | for i in range(5) : |
|
|
5698 | 5698 | vals = wA[i*h_avgs:(i+1)*h_avgs-0] |
|
|
5699 | 5699 | errs = wA_err[i*h_avgs:(i+1)*h_avgs-0] |
|
|
5700 | 5700 | avg = numpy.nansum(vals/errs**2.)/numpy.nansum(1./errs**2.) |
|
|
5701 | 5701 | sigma = numpy.sqrt(1./numpy.nansum(1./errs**2.)) |
|
|
5702 | 5702 | wA[6*h_avgs+i] = avg |
|
|
5703 | 5703 | wA_err[6*h_avgs+i] = sigma |
|
|
5704 | 5704 | |
|
|
5705 | 5705 | |
|
|
5706 | 5706 | vals = wA[0:6*h_avgs-0] |
|
|
5707 | 5707 | errs=wA_err[0:6*h_avgs-0] |
|
|
5708 | 5708 | avg = numpy.nansum(vals/errs**2.)/numpy.nansum(1./errs**2) |
|
|
5709 | 5709 | sigma = numpy.sqrt(1./numpy.nansum(1./errs**2.)) |
|
|
5710 | 5710 | wA[nhei_avg-1] = avg |
|
|
5711 | 5711 | wA_err[nhei_avg-1] = sigma |
|
|
5712 | 5712 | |
|
|
5713 | 5713 | wA = wA[6*h_avgs:nhei_avg-0] |
|
|
5714 | 5714 | wA_err=wA_err[6*h_avgs:nhei_avg-0] |
|
|
5715 | 5715 | if my_nbeams == 2 : |
|
|
5716 | 5716 | |
|
|
5717 | 5717 | uA = u[h0_index:h0_index+nhei_avg] |
|
|
5718 | 5718 | uA_err=u_err[h0_index:h0_index+nhei_avg] |
|
|
5719 | 5719 | |
|
|
5720 | 5720 | for i in range(5) : |
|
|
5721 | 5721 | vals = uA[i*h_avgs:(i+1)*h_avgs-0] |
|
|
5722 | 5722 | errs=uA_err[i*h_avgs:(i+1)*h_avgs-0] |
|
|
5723 | 5723 | avg = numpy.nansum(vals/errs**2.)/numpy.nansum(1./errs**2.) |
|
|
5724 | 5724 | sigma = numpy.sqrt(1./numpy.nansum(1./errs**2.)) |
|
|
5725 | 5725 | uA[6*h_avgs+i] = avg |
|
|
5726 | 5726 | uA_err[6*h_avgs+i]=sigma |
|
|
5727 | 5727 | |
|
|
5728 | 5728 | vals = uA[0:6*h_avgs-0] |
|
|
5729 | 5729 | errs = uA_err[0:6*h_avgs-0] |
|
|
5730 | 5730 | avg = numpy.nansum(vals/errs**2.)/numpy.nansum(1./errs**2.) |
|
|
5731 | 5731 | sigma = numpy.sqrt(1./numpy.nansum(1./errs**2.)) |
|
|
5732 | 5732 | uA[nhei_avg-1] = avg |
|
|
5733 | 5733 | uA_err[nhei_avg-1] = sigma |
|
|
5734 | 5734 | uA = uA[6*h_avgs:nhei_avg-0] |
|
|
5735 | 5735 | uA_err = uA_err[6*h_avgs:nhei_avg-0] |
|
|
5736 | 5736 | |
|
|
5737 | 5737 | dataOut.drifts_avg = numpy.vstack((wA,uA)) |
|
|
5738 | 5738 | |
|
|
5739 | 5739 | tini=time.localtime(dataOut.utctime) |
|
|
5740 | 5740 | datefile= str(tini[0]).zfill(4)+str(tini[1]).zfill(2)+str(tini[2]).zfill(2) |
|
|
5741 | 5741 | nfile = '/home/pcondor/Database/ewdriftsschain2019/jro'+datefile+'drifts_sch3.txt' |
|
|
5742 | 5742 | |
|
|
5743 | 5743 | f1 = open(nfile,'a') |
|
|
5744 | 5744 | |
|
|
5745 | 5745 | datedriftavg=str(tini[0])+' '+str(tini[1])+' '+str(tini[2])+' '+str(tini[3])+' '+str(tini[4]) |
|
|
5746 | 5746 | driftavgstr=str(dataOut.drifts_avg) |
|
|
5747 | 5747 | |
|
|
5748 | 5748 | numpy.savetxt(f1,numpy.column_stack([tini[0],tini[1],tini[2],tini[3],tini[4]]),fmt='%4i') |
|
|
5749 | 5749 | numpy.savetxt(f1,dataOut.drifts_avg,fmt='%10.2f') |
|
|
5750 | 5750 | f1.close() |
|
|
5751 | 5751 | |
|
|
5752 | 5752 | return dataOut |
|
|
5753 | 5753 | |
|
|
5754 | 5754 | #--------------- Non Specular Meteor ---------------- |
|
|
5755 | 5755 | |
|
|
5756 | 5756 | class NonSpecularMeteorDetection(Operation): |
|
|
5757 | 5757 | |
|
|
5758 | 5758 | def run(self, dataOut, mode, SNRthresh=8, phaseDerThresh=0.5, cohThresh=0.8, allData = False): |
|
|
5759 | 5759 | data_acf = dataOut.data_pre[0] |
|
|
5760 | 5760 | data_ccf = dataOut.data_pre[1] |
|
|
5761 | 5761 | pairsList = dataOut.groupList[1] |
|
|
5762 | 5762 | |
|
|
5763 | 5763 | lamb = dataOut.C/dataOut.frequency |
|
|
5764 | 5764 | tSamp = dataOut.ippSeconds*dataOut.nCohInt |
|
|
5765 | 5765 | paramInterval = dataOut.paramInterval |
|
|
5766 | 5766 | |
|
|
5767 | 5767 | nChannels = data_acf.shape[0] |
|
|
5768 | 5768 | nLags = data_acf.shape[1] |
|
|
5769 | 5769 | nProfiles = data_acf.shape[2] |
|
|
5770 | 5770 | nHeights = dataOut.nHeights |
|
|
5771 | 5771 | nCohInt = dataOut.nCohInt |
|
|
5772 | 5772 | sec = numpy.round(nProfiles/dataOut.paramInterval) |
|
|
5773 | 5773 | heightList = dataOut.heightList |
|
|
5774 | 5774 | ippSeconds = dataOut.ippSeconds*dataOut.nCohInt*dataOut.nAvg |
|
|
5775 | 5775 | utctime = dataOut.utctime |
|
|
5776 | 5776 | |
|
|
5777 | 5777 | dataOut.abscissaList = numpy.arange(0,paramInterval+ippSeconds,ippSeconds) |
|
|
5778 | 5778 | |
|
|
5779 | 5779 | #------------------------ SNR -------------------------------------- |
|
|
5780 | 5780 | power = data_acf[:,0,:,:].real |
|
|
5781 | 5781 | noise = numpy.zeros(nChannels) |
|
|
5782 | 5782 | SNR = numpy.zeros(power.shape) |
|
|
5783 | 5783 | for i in range(nChannels): |
|
|
5784 | 5784 | noise[i] = hildebrand_sekhon(power[i,:], nCohInt) |
|
|
5785 | 5785 | SNR[i] = (power[i]-noise[i])/noise[i] |
|
|
5786 | 5786 | SNRm = numpy.nanmean(SNR, axis = 0) |
|
|
5787 | 5787 | SNRdB = 10*numpy.log10(SNR) |
|
|
5788 | 5788 | |
|
|
5789 | 5789 | if mode == 'SA': |
|
|
5790 | 5790 | dataOut.groupList = dataOut.groupList[1] |
|
|
5791 | 5791 | nPairs = data_ccf.shape[0] |
|
|
5792 | 5792 | #---------------------- Coherence and Phase -------------------------- |
|
|
5793 | 5793 | phase = numpy.zeros(data_ccf[:,0,:,:].shape) |
|
|
5794 | 5794 | # phase1 = numpy.copy(phase) |
|
|
5795 | 5795 | coh1 = numpy.zeros(data_ccf[:,0,:,:].shape) |
|
|
5796 | 5796 | |
|
|
5797 | 5797 | for p in range(nPairs): |
|
|
5798 | 5798 | ch0 = pairsList[p][0] |
|
|
5799 | 5799 | ch1 = pairsList[p][1] |
|
|
5800 | 5800 | ccf = data_ccf[p,0,:,:]/numpy.sqrt(data_acf[ch0,0,:,:]*data_acf[ch1,0,:,:]) |
|
|
5801 | 5801 | phase[p,:,:] = ndimage.median_filter(numpy.angle(ccf), size = (5,1)) #median filter |
|
|
5802 | 5802 | # phase1[p,:,:] = numpy.angle(ccf) #median filter |
|
|
5803 | 5803 | coh1[p,:,:] = ndimage.median_filter(numpy.abs(ccf), 5) #median filter |
|
|
5804 | 5804 | # coh1[p,:,:] = numpy.abs(ccf) #median filter |
|
|
5805 | 5805 | coh = numpy.nanmax(coh1, axis = 0) |
|
|
5806 | 5806 | # struc = numpy.ones((5,1)) |
|
|
5807 | 5807 | # coh = ndimage.morphology.grey_dilation(coh, size=(10,1)) |
|
|
5808 | 5808 | #---------------------- Radial Velocity ---------------------------- |
|
|
5809 | 5809 | phaseAux = numpy.mean(numpy.angle(data_acf[:,1,:,:]), axis = 0) |
|
|
5810 | 5810 | velRad = phaseAux*lamb/(4*numpy.pi*tSamp) |
|
|
5811 | 5811 | |
|
|
5812 | 5812 | if allData: |
|
|
5813 | 5813 | boolMetFin = ~numpy.isnan(SNRm) |
|
|
5814 | 5814 | # coh[:-1,:] = numpy.nanmean(numpy.abs(phase[:,1:,:] - phase[:,:-1,:]),axis=0) |
|
|
5815 | 5815 | else: |
|
|
5816 | 5816 | #------------------------ Meteor mask --------------------------------- |
|
|
5817 | 5817 | # #SNR mask |
|
|
5818 | 5818 | # boolMet = (SNRdB>SNRthresh)#|(~numpy.isnan(SNRdB)) |
|
|
5819 | 5819 | # |
|
|
5820 | 5820 | # #Erase small objects |
|
|
5821 | 5821 | # boolMet1 = self.__erase_small(boolMet, 2*sec, 5) |
|
|
5822 | 5822 | # |
|
|
5823 | 5823 | # auxEEJ = numpy.sum(boolMet1,axis=0) |
|
|
5824 | 5824 | # indOver = auxEEJ>nProfiles*0.8 #Use this later |
|
|
5825 | 5825 | # indEEJ = numpy.where(indOver)[0] |
|
|
5826 | 5826 | # indNEEJ = numpy.where(~indOver)[0] |
|
|
5827 | 5827 | # |
|
|
5828 | 5828 | # boolMetFin = boolMet1 |
|
|
5829 | 5829 | # |
|
|
5830 | 5830 | # if indEEJ.size > 0: |
|
|
5831 | 5831 | # boolMet1[:,indEEJ] = False #Erase heights with EEJ |
|
|
5832 | 5832 | # |
|
|
5833 | 5833 | # boolMet2 = coh > cohThresh |
|
|
5834 | 5834 | # boolMet2 = self.__erase_small(boolMet2, 2*sec,5) |
|
|
5835 | 5835 | # |
|
|
5836 | 5836 | # #Final Meteor mask |
|
|
5837 | 5837 | # boolMetFin = boolMet1|boolMet2 |
|
|
5838 | 5838 | |
|
|
5839 | 5839 | #Coherence mask |
|
|
5840 | 5840 | boolMet1 = coh > 0.75 |
|
|
5841 | 5841 | struc = numpy.ones((30,1)) |
|
|
5842 | 5842 | boolMet1 = ndimage.morphology.binary_dilation(boolMet1, structure=struc) |
|
|
5843 | 5843 | |
|
|
5844 | 5844 | #Derivative mask |
|
|
5845 | 5845 | derPhase = numpy.nanmean(numpy.abs(phase[:,1:,:] - phase[:,:-1,:]),axis=0) |
|
|
5846 | 5846 | boolMet2 = derPhase < 0.2 |
|
|
5847 | 5847 | # boolMet2 = ndimage.morphology.binary_opening(boolMet2) |
|
|
5848 | 5848 | # boolMet2 = ndimage.morphology.binary_closing(boolMet2, structure = numpy.ones((10,1))) |
|
|
5849 | 5849 | boolMet2 = ndimage.median_filter(boolMet2,size=5) |
|
|
5850 | 5850 | boolMet2 = numpy.vstack((boolMet2,numpy.full((1,nHeights), True, dtype=bool))) |
|
|
5851 | 5851 | # #Final mask |
|
|
5852 | 5852 | # boolMetFin = boolMet2 |
|
|
5853 | 5853 | boolMetFin = boolMet1&boolMet2 |
|
|
5854 | 5854 | # boolMetFin = ndimage.morphology.binary_dilation(boolMetFin) |
|
|
5855 | 5855 | #Creating data_param |
|
|
5856 | 5856 | coordMet = numpy.where(boolMetFin) |
|
|
5857 | 5857 | |
|
|
5858 | 5858 | tmet = coordMet[0] |
|
|
5859 | 5859 | hmet = coordMet[1] |
|
|
5860 | 5860 | |
|
|
5861 | 5861 | data_param = numpy.zeros((tmet.size, 6 + nPairs)) |
|
|
5862 | 5862 | data_param[:,0] = utctime |
|
|
5863 | 5863 | data_param[:,1] = tmet |
|
|
5864 | 5864 | data_param[:,2] = hmet |
|
|
5865 | 5865 | data_param[:,3] = SNRm[tmet,hmet] |
|
|
5866 | 5866 | data_param[:,4] = velRad[tmet,hmet] |
|
|
5867 | 5867 | data_param[:,5] = coh[tmet,hmet] |
|
|
5868 | 5868 | data_param[:,6:] = phase[:,tmet,hmet].T |
|
|
5869 | 5869 | |
|
|
5870 | 5870 | elif mode == 'DBS': |
|
|
5871 | 5871 | dataOut.groupList = numpy.arange(nChannels) |
|
|
5872 | 5872 | |
|
|
5873 | 5873 | #Radial Velocities |
|
|
5874 | 5874 | phase = numpy.angle(data_acf[:,1,:,:]) |
|
|
5875 | 5875 | # phase = ndimage.median_filter(numpy.angle(data_acf[:,1,:,:]), size = (1,5,1)) |
|
|
5876 | 5876 | velRad = phase*lamb/(4*numpy.pi*tSamp) |
|
|
5877 | 5877 | |
|
|
5878 | 5878 | #Spectral width |
|
|
5879 | 5879 | # acf1 = ndimage.median_filter(numpy.abs(data_acf[:,1,:,:]), size = (1,5,1)) |
|
|
5880 | 5880 | # acf2 = ndimage.median_filter(numpy.abs(data_acf[:,2,:,:]), size = (1,5,1)) |
|
|
5881 | 5881 | acf1 = data_acf[:,1,:,:] |
|
|
5882 | 5882 | acf2 = data_acf[:,2,:,:] |
|
|
5883 | 5883 | |
|
|
5884 | 5884 | spcWidth = (lamb/(2*numpy.sqrt(6)*numpy.pi*tSamp))*numpy.sqrt(numpy.log(acf1/acf2)) |
|
|
5885 | 5885 | # velRad = ndimage.median_filter(velRad, size = (1,5,1)) |
|
|
5886 | 5886 | if allData: |
|
|
5887 | 5887 | boolMetFin = ~numpy.isnan(SNRdB) |
|
|
5888 | 5888 | else: |
|
|
5889 | 5889 | #SNR |
|
|
5890 | 5890 | boolMet1 = (SNRdB>SNRthresh) #SNR mask |
|
|
5891 | 5891 | boolMet1 = ndimage.median_filter(boolMet1, size=(1,5,5)) |
|
|
5892 | 5892 | |
|
|
5893 | 5893 | #Radial velocity |
|
|
5894 | 5894 | boolMet2 = numpy.abs(velRad) < 20 |
|
|
5895 | 5895 | boolMet2 = ndimage.median_filter(boolMet2, (1,5,5)) |
|
|
5896 | 5896 | |
|
|
5897 | 5897 | #Spectral Width |
|
|
5898 | 5898 | boolMet3 = spcWidth < 30 |
|
|
5899 | 5899 | boolMet3 = ndimage.median_filter(boolMet3, (1,5,5)) |
|
|
5900 | 5900 | # boolMetFin = self.__erase_small(boolMet1, 10,5) |
|
|
5901 | 5901 | boolMetFin = boolMet1&boolMet2&boolMet3 |
|
|
5902 | 5902 | |
|
|
5903 | 5903 | #Creating data_param |
|
|
5904 | 5904 | coordMet = numpy.where(boolMetFin) |
|
|
5905 | 5905 | |
|
|
5906 | 5906 | cmet = coordMet[0] |
|
|
5907 | 5907 | tmet = coordMet[1] |
|
|
5908 | 5908 | hmet = coordMet[2] |
|
|
5909 | 5909 | |
|
|
5910 | 5910 | data_param = numpy.zeros((tmet.size, 7)) |
|
|
5911 | 5911 | data_param[:,0] = utctime |
|
|
5912 | 5912 | data_param[:,1] = cmet |
|
|
5913 | 5913 | data_param[:,2] = tmet |
|
|
5914 | 5914 | data_param[:,3] = hmet |
|
|
5915 | 5915 | data_param[:,4] = SNR[cmet,tmet,hmet].T |
|
|
5916 | 5916 | data_param[:,5] = velRad[cmet,tmet,hmet].T |
|
|
5917 | 5917 | data_param[:,6] = spcWidth[cmet,tmet,hmet].T |
|
|
5918 | 5918 | |
|
|
5919 | 5919 | # self.dataOut.data_param = data_int |
|
|
5920 | 5920 | if len(data_param) == 0: |
|
|
5921 | 5921 | dataOut.flagNoData = True |
|
|
5922 | 5922 | else: |
|
|
5923 | 5923 | dataOut.data_param = data_param |
|
|
5924 | 5924 | |
|
|
5925 | 5925 | def __erase_small(self, binArray, threshX, threshY): |
|
|
5926 | 5926 | labarray, numfeat = ndimage.measurements.label(binArray) |
|
|
5927 | 5927 | binArray1 = numpy.copy(binArray) |
|
|
5928 | 5928 | |
|
|
5929 | 5929 | for i in range(1,numfeat + 1): |
|
|
5930 | 5930 | auxBin = (labarray==i) |
|
|
5931 | 5931 | auxSize = auxBin.sum() |
|
|
5932 | 5932 | |
|
|
5933 | 5933 | x,y = numpy.where(auxBin) |
|
|
5934 | 5934 | widthX = x.max() - x.min() |
|
|
5935 | 5935 | widthY = y.max() - y.min() |
|
|
5936 | 5936 | |
|
|
5937 | 5937 | #width X: 3 seg -> 12.5*3 |
|
|
5938 | 5938 | #width Y: |
|
|
5939 | 5939 | |
|
|
5940 | 5940 | if (auxSize < 50) or (widthX < threshX) or (widthY < threshY): |
|
|
5941 | 5941 | binArray1[auxBin] = False |
|
|
5942 | 5942 | |
|
|
5943 | 5943 | return binArray1 |
|
|
5944 | 5944 | |
|
|
5945 | 5945 | #--------------- Specular Meteor ---------------- |
|
|
5946 | 5946 | |
|
|
5947 | 5947 | class SMDetection(Operation): |
|
|
5948 | 5948 | ''' |
|
|
5949 | 5949 | Function DetectMeteors() |
|
|
5950 | 5950 | Project developed with paper: |
|
|
5951 | 5951 | HOLDSWORTH ET AL. 2004 |
|
|
5952 | 5952 | |
|
|
5953 | 5953 | Input: |
|
|
5954 | 5954 | self.dataOut.data_pre |
|
|
5955 | 5955 | |
|
|
5956 | 5956 | centerReceiverIndex: From the channels, which is the center receiver |
|
|
5957 | 5957 | |
|
|
5958 | 5958 | hei_ref: Height reference for the Beacon signal extraction |
|
|
5959 | 5959 | tauindex: |
|
|
5960 | 5960 | predefinedPhaseShifts: Predefined phase offset for the voltge signals |
|
|
5961 | 5961 | |
|
|
5962 | 5962 | cohDetection: Whether to user Coherent detection or not |
|
|
5963 | 5963 | cohDet_timeStep: Coherent Detection calculation time step |
|
|
5964 | 5964 | cohDet_thresh: Coherent Detection phase threshold to correct phases |
|
|
5965 | 5965 | |
|
|
5966 | 5966 | noise_timeStep: Noise calculation time step |
|
|
5967 | 5967 | noise_multiple: Noise multiple to define signal threshold |
|
|
5968 | 5968 | |
|
|
5969 | 5969 | multDet_timeLimit: Multiple Detection Removal time limit in seconds |
|
|
5970 | 5970 | multDet_rangeLimit: Multiple Detection Removal range limit in km |
|
|
5971 | 5971 | |
|
|
5972 | 5972 | phaseThresh: Maximum phase difference between receiver to be consider a meteor |
|
|
5973 | 5973 | SNRThresh: Minimum SNR threshold of the meteor signal to be consider a meteor |
|
|
5974 | 5974 | |
|
|
5975 | 5975 | hmin: Minimum Height of the meteor to use it in the further wind estimations |
|
|
5976 | 5976 | hmax: Maximum Height of the meteor to use it in the further wind estimations |
|
|
5977 | 5977 | azimuth: Azimuth angle correction |
|
|
5978 | 5978 | |
|
|
5979 | 5979 | Affected: |
|
|
5980 | 5980 | self.dataOut.data_param |
|
|
5981 | 5981 | |
|
|
5982 | 5982 | Rejection Criteria (Errors): |
|
|
5983 | 5983 | 0: No error; analysis OK |
|
|
5984 | 5984 | 1: SNR < SNR threshold |
|
|
5985 | 5985 | 2: angle of arrival (AOA) ambiguously determined |
|
|
5986 | 5986 | 3: AOA estimate not feasible |
|
|
5987 | 5987 | 4: Large difference in AOAs obtained from different antenna baselines |
|
|
5988 | 5988 | 5: echo at start or end of time series |
|
|
5989 | 5989 | 6: echo less than 5 examples long; too short for analysis |
|
|
5990 | 5990 | 7: echo rise exceeds 0.3s |
|
|
5991 | 5991 | 8: echo decay time less than twice rise time |
|
|
5992 | 5992 | 9: large power level before echo |
|
|
5993 | 5993 | 10: large power level after echo |
|
|
5994 | 5994 | 11: poor fit to amplitude for estimation of decay time |
|
|
5995 | 5995 | 12: poor fit to CCF phase variation for estimation of radial drift velocity |
|
|
5996 | 5996 | 13: height unresolvable echo: not valid height within 70 to 110 km |
|
|
5997 | 5997 | 14: height ambiguous echo: more then one possible height within 70 to 110 km |
|
|
5998 | 5998 | 15: radial drift velocity or projected horizontal velocity exceeds 200 m/s |
|
|
5999 | 5999 | 16: oscilatory echo, indicating event most likely not an underdense echo |
|
|
6000 | 6000 | |
|
|
6001 | 6001 | 17: phase difference in meteor Reestimation |
|
|
6002 | 6002 | |
|
|
6003 | 6003 | Data Storage: |
|
|
6004 | 6004 | Meteors for Wind Estimation (8): |
|
|
6005 | 6005 | Utc Time | Range Height |
|
|
6006 | 6006 | Azimuth Zenith errorCosDir |
|
|
6007 | 6007 | VelRad errorVelRad |
|
|
6008 | 6008 | Phase0 Phase1 Phase2 Phase3 |
|
|
6009 | 6009 | TypeError |
|
|
6010 | 6010 | |
|
|
6011 | 6011 | ''' |
|
|
6012 | 6012 | |
|
|
6013 | 6013 | def run(self, dataOut, hei_ref = None, tauindex = 0, |
|
|
6014 | 6014 | phaseOffsets = None, |
|
|
6015 | 6015 | cohDetection = False, cohDet_timeStep = 1, cohDet_thresh = 25, |
|
|
6016 | 6016 | noise_timeStep = 4, noise_multiple = 4, |
|
|
6017 | 6017 | multDet_timeLimit = 1, multDet_rangeLimit = 3, |
|
|
6018 | 6018 | phaseThresh = 20, SNRThresh = 5, |
|
|
6019 | 6019 | hmin = 50, hmax=150, azimuth = 0, |
|
|
6020 | 6020 | channelPositions = None) : |
|
|
6021 | 6021 | |
|
|
6022 | 6022 | |
|
|
6023 | 6023 | #Getting Pairslist |
|
|
6024 | 6024 | if channelPositions is None: |
|
|
6025 | 6025 | # channelPositions = [(2.5,0), (0,2.5), (0,0), (0,4.5), (-2,0)] #T |
|
|
6026 | 6026 | channelPositions = [(4.5,2), (2,4.5), (2,2), (2,0), (0,2)] #Estrella |
|
|
6027 | 6027 | meteorOps = SMOperations() |
|
|
6028 | 6028 | pairslist0, distances = meteorOps.getPhasePairs(channelPositions) |
|
|
6029 | 6029 | heiRang = dataOut.heightList |
|
|
6030 | 6030 | #Get Beacon signal - No Beacon signal anymore |
|
|
6031 | 6031 | # newheis = numpy.where(self.dataOut.heightList>self.dataOut.radarControllerHeaderObj.Taus[tauindex]) |
|
|
6032 | 6032 | # |
|
|
6033 | 6033 | # if hei_ref != None: |
|
|
6034 | 6034 | # newheis = numpy.where(self.dataOut.heightList>hei_ref) |
|
|
6035 | 6035 | # |
|
|
6036 | 6036 | |
|
|
6037 | 6037 | |
|
|
6038 | 6038 | #****************REMOVING HARDWARE PHASE DIFFERENCES*************** |
|
|
6039 | 6039 | # see if the user put in pre defined phase shifts |
|
|
6040 | 6040 | voltsPShift = dataOut.data_pre.copy() |
|
|
6041 | 6041 | |
|
|
6042 | 6042 | # if predefinedPhaseShifts != None: |
|
|
6043 | 6043 | # hardwarePhaseShifts = numpy.array(predefinedPhaseShifts)*numpy.pi/180 |
|
|
6044 | 6044 | # |
|
|
6045 | 6045 | # # elif beaconPhaseShifts: |
|
|
6046 | 6046 | # # #get hardware phase shifts using beacon signal |
|
|
6047 | 6047 | # # hardwarePhaseShifts = self.__getHardwarePhaseDiff(self.dataOut.data_pre, pairslist, newheis, 10) |
|
|
6048 | 6048 | # # hardwarePhaseShifts = numpy.insert(hardwarePhaseShifts,centerReceiverIndex,0) |
|
|
6049 | 6049 | # |
|
|
6050 | 6050 | # else: |
|
|
6051 | 6051 | # hardwarePhaseShifts = numpy.zeros(5) |
|
|
6052 | 6052 | # |
|
|
6053 | 6053 | # voltsPShift = numpy.zeros((self.dataOut.data_pre.shape[0],self.dataOut.data_pre.shape[1],self.dataOut.data_pre.shape[2]), dtype = 'complex') |
|
|
6054 | 6054 | # for i in range(self.dataOut.data_pre.shape[0]): |
|
|
6055 | 6055 | # voltsPShift[i,:,:] = self.__shiftPhase(self.dataOut.data_pre[i,:,:], hardwarePhaseShifts[i]) |
|
|
6056 | 6056 | |
|
|
6057 | 6057 | #******************END OF REMOVING HARDWARE PHASE DIFFERENCES********* |
|
|
6058 | 6058 | |
|
|
6059 | 6059 | #Remove DC |
|
|
6060 | 6060 | voltsDC = numpy.mean(voltsPShift,1) |
|
|
6061 | 6061 | voltsDC = numpy.mean(voltsDC,1) |
|
|
6062 | 6062 | for i in range(voltsDC.shape[0]): |
|
|
6063 | 6063 | voltsPShift[i] = voltsPShift[i] - voltsDC[i] |
|
|
6064 | 6064 | |
|
|
6065 | 6065 | #Don't considerate last heights, theyre used to calculate Hardware Phase Shift |
|
|
6066 | 6066 | # voltsPShift = voltsPShift[:,:,:newheis[0][0]] |
|
|
6067 | 6067 | |
|
|
6068 | 6068 | #************ FIND POWER OF DATA W/COH OR NON COH DETECTION (3.4) ********** |
|
|
6069 | 6069 | #Coherent Detection |
|
|
6070 | 6070 | if cohDetection: |
|
|
6071 | 6071 | #use coherent detection to get the net power |
|
|
6072 | 6072 | cohDet_thresh = cohDet_thresh*numpy.pi/180 |
|
|
6073 | 6073 | voltsPShift = self.__coherentDetection(voltsPShift, cohDet_timeStep, dataOut.timeInterval, pairslist0, cohDet_thresh) |
|
|
6074 | 6074 | |
|
|
6075 | 6075 | #Non-coherent detection! |
|
|
6076 | 6076 | powerNet = numpy.nansum(numpy.abs(voltsPShift[:,:,:])**2,0) |
|
|
6077 | 6077 | #********** END OF COH/NON-COH POWER CALCULATION********************** |
|
|
6078 | 6078 | |
|
|
6079 | 6079 | #********** FIND THE NOISE LEVEL AND POSSIBLE METEORS **************** |
|
|
6080 | 6080 | #Get noise |
|
|
6081 | 6081 | noise, noise1 = self.__getNoise(powerNet, noise_timeStep, dataOut.timeInterval) |
|
|
6082 | 6082 | # noise = self.getNoise1(powerNet, noise_timeStep, self.dataOut.timeInterval) |
|
|
6083 | 6083 | #Get signal threshold |
|
|
6084 | 6084 | signalThresh = noise_multiple*noise |
|
|
6085 | 6085 | #Meteor echoes detection |
|
|
6086 | 6086 | listMeteors = self.__findMeteors(powerNet, signalThresh) |
|
|
6087 | 6087 | #******* END OF NOISE LEVEL AND POSSIBLE METEORS CACULATION ********** |
|
|
6088 | 6088 | |
|
|
6089 | 6089 | #************** REMOVE MULTIPLE DETECTIONS (3.5) *************************** |
|
|
6090 | 6090 | #Parameters |
|
|
6091 | 6091 | heiRange = dataOut.heightList |
|
|
6092 | 6092 | rangeInterval = heiRange[1] - heiRange[0] |
|
|
6093 | 6093 | rangeLimit = multDet_rangeLimit/rangeInterval |
|
|
6094 | 6094 | timeLimit = multDet_timeLimit/dataOut.timeInterval |
|
|
6095 | 6095 | #Multiple detection removals |
|
|
6096 | 6096 | listMeteors1 = self.__removeMultipleDetections(listMeteors, rangeLimit, timeLimit) |
|
|
6097 | 6097 | #************ END OF REMOVE MULTIPLE DETECTIONS ********************** |
|
|
6098 | 6098 | |
|
|
6099 | 6099 | #********************* METEOR REESTIMATION (3.7, 3.8, 3.9, 3.10) ******************** |
|
|
6100 | 6100 | #Parameters |
|
|
6101 | 6101 | phaseThresh = phaseThresh*numpy.pi/180 |
|
|
6102 | 6102 | thresh = [phaseThresh, noise_multiple, SNRThresh] |
|
|
6103 | 6103 | #Meteor reestimation (Errors N 1, 6, 12, 17) |
|
|
6104 | 6104 | listMeteors2, listMeteorsPower, listMeteorsVolts = self.__meteorReestimation(listMeteors1, voltsPShift, pairslist0, thresh, noise, dataOut.timeInterval, dataOut.frequency) |
|
|
6105 | 6105 | # listMeteors2, listMeteorsPower, listMeteorsVolts = self.meteorReestimation3(listMeteors2, listMeteorsPower, listMeteorsVolts, voltsPShift, pairslist, thresh, noise) |
|
|
6106 | 6106 | #Estimation of decay times (Errors N 7, 8, 11) |
|
|
6107 | 6107 | listMeteors3 = self.__estimateDecayTime(listMeteors2, listMeteorsPower, dataOut.timeInterval, dataOut.frequency) |
|
|
6108 | 6108 | #******************* END OF METEOR REESTIMATION ******************* |
|
|
6109 | 6109 | |
|
|
6110 | 6110 | #********************* METEOR PARAMETERS CALCULATION (3.11, 3.12, 3.13) ************************** |
|
|
6111 | 6111 | #Calculating Radial Velocity (Error N 15) |
|
|
6112 | 6112 | radialStdThresh = 10 |
|
|
6113 | 6113 | listMeteors4 = self.__getRadialVelocity(listMeteors3, listMeteorsVolts, radialStdThresh, pairslist0, dataOut.timeInterval) |
|
|
6114 | 6114 | |
|
|
6115 | 6115 | if len(listMeteors4) > 0: |
|
|
6116 | 6116 | #Setting New Array |
|
|
6117 | 6117 | date = dataOut.utctime |
|
|
6118 | 6118 | arrayParameters = self.__setNewArrays(listMeteors4, date, heiRang) |
|
|
6119 | 6119 | |
|
|
6120 | 6120 | #Correcting phase offset |
|
|
6121 | 6121 | if phaseOffsets != None: |
|
|
6122 | 6122 | phaseOffsets = numpy.array(phaseOffsets)*numpy.pi/180 |
|
|
6123 | 6123 | arrayParameters[:,8:12] = numpy.unwrap(arrayParameters[:,8:12] + phaseOffsets) |
|
|
6124 | 6124 | |
|
|
6125 | 6125 | #Second Pairslist |
|
|
6126 | 6126 | pairsList = [] |
|
|
6127 | 6127 | pairx = (0,1) |
|
|
6128 | 6128 | pairy = (2,3) |
|
|
6129 | 6129 | pairsList.append(pairx) |
|
|
6130 | 6130 | pairsList.append(pairy) |
|
|
6131 | 6131 | |
|
|
6132 | 6132 | jph = numpy.array([0,0,0,0]) |
|
|
6133 | 6133 | h = (hmin,hmax) |
|
|
6134 | 6134 | arrayParameters = meteorOps.getMeteorParams(arrayParameters, azimuth, h, pairsList, distances, jph) |
|
|
6135 | 6135 | |
|
|
6136 | 6136 | # #Calculate AOA (Error N 3, 4) |
|
|
6137 | 6137 | # #JONES ET AL. 1998 |
|
|
6138 | 6138 | # error = arrayParameters[:,-1] |
|
|
6139 | 6139 | # AOAthresh = numpy.pi/8 |
|
|
6140 | 6140 | # phases = -arrayParameters[:,9:13] |
|
|
6141 | 6141 | # arrayParameters[:,4:7], arrayParameters[:,-1] = meteorOps.getAOA(phases, pairsList, error, AOAthresh, azimuth) |
|
|
6142 | 6142 | # |
|
|
6143 | 6143 | # #Calculate Heights (Error N 13 and 14) |
|
|
6144 | 6144 | # error = arrayParameters[:,-1] |
|
|
6145 | 6145 | # Ranges = arrayParameters[:,2] |
|
|
6146 | 6146 | # zenith = arrayParameters[:,5] |
|
|
6147 | 6147 | # arrayParameters[:,3], arrayParameters[:,-1] = meteorOps.getHeights(Ranges, zenith, error, hmin, hmax) |
|
|
6148 | 6148 | # error = arrayParameters[:,-1] |
|
|
6149 | 6149 | #********************* END OF PARAMETERS CALCULATION ************************** |
|
|
6150 | 6150 | |
|
|
6151 | 6151 | #***************************+ PASS DATA TO NEXT STEP ********************** |
|
|
6152 | 6152 | # arrayFinal = arrayParameters.reshape((1,arrayParameters.shape[0],arrayParameters.shape[1])) |
|
|
6153 | 6153 | dataOut.data_param = arrayParameters |
|
|
6154 | 6154 | |
|
|
6155 | 6155 | if arrayParameters is None: |
|
|
6156 | 6156 | dataOut.flagNoData = True |
|
|
6157 | 6157 | else: |
|
|
6158 | 6158 | dataOut.flagNoData = True |
|
|
6159 | 6159 | |
|
|
6160 | 6160 | return |
|
|
6161 | 6161 | |
|
|
6162 | 6162 | def __getHardwarePhaseDiff(self, voltage0, pairslist, newheis, n): |
|
|
6163 | 6163 | |
|
|
6164 | 6164 | minIndex = min(newheis[0]) |
|
|
6165 | 6165 | maxIndex = max(newheis[0]) |
|
|
6166 | 6166 | |
|
|
6167 | 6167 | voltage = voltage0[:,:,minIndex:maxIndex+1] |
|
|
6168 | 6168 | nLength = voltage.shape[1]/n |
|
|
6169 | 6169 | nMin = 0 |
|
|
6170 | 6170 | nMax = 0 |
|
|
6171 | 6171 | phaseOffset = numpy.zeros((len(pairslist),n)) |
|
|
6172 | 6172 | |
|
|
6173 | 6173 | for i in range(n): |
|
|
6174 | 6174 | nMax += nLength |
|
|
6175 | 6175 | phaseCCF = -numpy.angle(self.__calculateCCF(voltage[:,nMin:nMax,:], pairslist, [0])) |
|
|
6176 | 6176 | phaseCCF = numpy.mean(phaseCCF, axis = 2) |
|
|
6177 | 6177 | phaseOffset[:,i] = phaseCCF.transpose() |
|
|
6178 | 6178 | nMin = nMax |
|
|
6179 | 6179 | # phaseDiff, phaseArrival = self.estimatePhaseDifference(voltage, pairslist) |
|
|
6180 | 6180 | |
|
|
6181 | 6181 | #Remove Outliers |
|
|
6182 | 6182 | factor = 2 |
|
|
6183 | 6183 | wt = phaseOffset - signal.medfilt(phaseOffset,(1,5)) |
|
|
6184 | 6184 | dw = numpy.std(wt,axis = 1) |
|
|
6185 | 6185 | dw = dw.reshape((dw.size,1)) |
|
|
6186 | 6186 | ind = numpy.where(numpy.logical_or(wt>dw*factor,wt<-dw*factor)) |
|
|
6187 | 6187 | phaseOffset[ind] = numpy.nan |
|
|
6188 | 6188 | phaseOffset = stats.nanmean(phaseOffset, axis=1) |
|
|
6189 | 6189 | |
|
|
6190 | 6190 | return phaseOffset |
|
|
6191 | 6191 | |
|
|
6192 | 6192 | def __shiftPhase(self, data, phaseShift): |
|
|
6193 | 6193 | #this will shift the phase of a complex number |
|
|
6194 | 6194 | dataShifted = numpy.abs(data) * numpy.exp((numpy.angle(data)+phaseShift)*1j) |
|
|
6195 | 6195 | return dataShifted |
|
|
6196 | 6196 | |
|
|
6197 | 6197 | def __estimatePhaseDifference(self, array, pairslist): |
|
|
6198 | 6198 | nChannel = array.shape[0] |
|
|
6199 | 6199 | nHeights = array.shape[2] |
|
|
6200 | 6200 | numPairs = len(pairslist) |
|
|
6201 | 6201 | # phaseCCF = numpy.zeros((nChannel, 5, nHeights)) |
|
|
6202 | 6202 | phaseCCF = numpy.angle(self.__calculateCCF(array, pairslist, [-2,-1,0,1,2])) |
|
|
6203 | 6203 | |
|
|
6204 | 6204 | #Correct phases |
|
|
6205 | 6205 | derPhaseCCF = phaseCCF[:,1:,:] - phaseCCF[:,0:-1,:] |
|
|
6206 | 6206 | indDer = numpy.where(numpy.abs(derPhaseCCF) > numpy.pi) |
|
|
6207 | 6207 | |
|
|
6208 | 6208 | if indDer[0].shape[0] > 0: |
|
|
6209 | 6209 | for i in range(indDer[0].shape[0]): |
|
|
6210 | 6210 | signo = -numpy.sign(derPhaseCCF[indDer[0][i],indDer[1][i],indDer[2][i]]) |
|
|
6211 | 6211 | phaseCCF[indDer[0][i],indDer[1][i]+1:,:] += signo*2*numpy.pi |
|
|
6212 | 6212 | |
|
|
6213 | 6213 | # for j in range(numSides): |
|
|
6214 | 6214 | # phaseCCFAux = self.calculateCCF(arrayCenter, arraySides[j,:,:], [-2,1,0,1,2]) |
|
|
6215 | 6215 | # phaseCCF[j,:,:] = numpy.angle(phaseCCFAux) |
|
|
6216 | 6216 | # |
|
|
6217 | 6217 | #Linear |
|
|
6218 | 6218 | phaseInt = numpy.zeros((numPairs,1)) |
|
|
6219 | 6219 | angAllCCF = phaseCCF[:,[0,1,3,4],0] |
|
|
6220 | 6220 | for j in range(numPairs): |
|
|
6221 | 6221 | fit = stats.linregress([-2,-1,1,2],angAllCCF[j,:]) |
|
|
6222 | 6222 | phaseInt[j] = fit[1] |
|
|
6223 | 6223 | #Phase Differences |
|
|
6224 | 6224 | phaseDiff = phaseInt - phaseCCF[:,2,:] |
|
|
6225 | 6225 | phaseArrival = phaseInt.reshape(phaseInt.size) |
|
|
6226 | 6226 | |
|
|
6227 | 6227 | #Dealias |
|
|
6228 | 6228 | phaseArrival = numpy.angle(numpy.exp(1j*phaseArrival)) |
|
|
6229 | 6229 | # indAlias = numpy.where(phaseArrival > numpy.pi) |
|
|
6230 | 6230 | # phaseArrival[indAlias] -= 2*numpy.pi |
|
|
6231 | 6231 | # indAlias = numpy.where(phaseArrival < -numpy.pi) |
|
|
6232 | 6232 | # phaseArrival[indAlias] += 2*numpy.pi |
|
|
6233 | 6233 | |
|
|
6234 | 6234 | return phaseDiff, phaseArrival |
|
|
6235 | 6235 | |
|
|
6236 | 6236 | def __coherentDetection(self, volts, timeSegment, timeInterval, pairslist, thresh): |
|
|
6237 | 6237 | #this function will run the coherent detection used in Holdworth et al. 2004 and return the net power |
|
|
6238 | 6238 | #find the phase shifts of each channel over 1 second intervals |
|
|
6239 | 6239 | #only look at ranges below the beacon signal |
|
|
6240 | 6240 | numProfPerBlock = numpy.ceil(timeSegment/timeInterval) |
|
|
6241 | 6241 | numBlocks = int(volts.shape[1]/numProfPerBlock) |
|
|
6242 | 6242 | numHeights = volts.shape[2] |
|
|
6243 | 6243 | nChannel = volts.shape[0] |
|
|
6244 | 6244 | voltsCohDet = volts.copy() |
|
|
6245 | 6245 | |
|
|
6246 | 6246 | pairsarray = numpy.array(pairslist) |
|
|
6247 | 6247 | indSides = pairsarray[:,1] |
|
|
6248 | 6248 | # indSides = numpy.array(range(nChannel)) |
|
|
6249 | 6249 | # indSides = numpy.delete(indSides, indCenter) |
|
|
6250 | 6250 | # |
|
|
6251 | 6251 | # listCenter = numpy.array_split(volts[indCenter,:,:], numBlocks, 0) |
|
|
6252 | 6252 | listBlocks = numpy.array_split(volts, numBlocks, 1) |
|
|
6253 | 6253 | |
|
|
6254 | 6254 | startInd = 0 |
|
|
6255 | 6255 | endInd = 0 |
|
|
6256 | 6256 | |
|
|
6257 | 6257 | for i in range(numBlocks): |
|
|
6258 | 6258 | startInd = endInd |
|
|
6259 | 6259 | endInd = endInd + listBlocks[i].shape[1] |
|
|
6260 | 6260 | |
|
|
6261 | 6261 | arrayBlock = listBlocks[i] |
|
|
6262 | 6262 | # arrayBlockCenter = listCenter[i] |
|
|
6263 | 6263 | |
|
|
6264 | 6264 | #Estimate the Phase Difference |
|
|
6265 | 6265 | phaseDiff, aux = self.__estimatePhaseDifference(arrayBlock, pairslist) |
|
|
6266 | 6266 | #Phase Difference RMS |
|
|
6267 | 6267 | arrayPhaseRMS = numpy.abs(phaseDiff) |
|
|
6268 | 6268 | phaseRMSaux = numpy.sum(arrayPhaseRMS < thresh,0) |
|
|
6269 | 6269 | indPhase = numpy.where(phaseRMSaux==4) |
|
|
6270 | 6270 | #Shifting |
|
|
6271 | 6271 | if indPhase[0].shape[0] > 0: |
|
|
6272 | 6272 | for j in range(indSides.size): |
|
|
6273 | 6273 | arrayBlock[indSides[j],:,indPhase] = self.__shiftPhase(arrayBlock[indSides[j],:,indPhase], phaseDiff[j,indPhase].transpose()) |
|
|
6274 | 6274 | voltsCohDet[:,startInd:endInd,:] = arrayBlock |
|
|
6275 | 6275 | |
|
|
6276 | 6276 | return voltsCohDet |
|
|
6277 | 6277 | |
|
|
6278 | 6278 | def __calculateCCF(self, volts, pairslist ,laglist): |
|
|
6279 | 6279 | |
|
|
6280 | 6280 | nHeights = volts.shape[2] |
|
|
6281 | 6281 | nPoints = volts.shape[1] |
|
|
6282 | 6282 | voltsCCF = numpy.zeros((len(pairslist), len(laglist), nHeights),dtype = 'complex') |
|
|
6283 | 6283 | |
|
|
6284 | 6284 | for i in range(len(pairslist)): |
|
|
6285 | 6285 | volts1 = volts[pairslist[i][0]] |
|
|
6286 | 6286 | volts2 = volts[pairslist[i][1]] |
|
|
6287 | 6287 | |
|
|
6288 | 6288 | for t in range(len(laglist)): |
|
|
6289 | 6289 | idxT = laglist[t] |
|
|
6290 | 6290 | if idxT >= 0: |
|
|
6291 | 6291 | vStacked = numpy.vstack((volts2[idxT:,:], |
|
|
6292 | 6292 | numpy.zeros((idxT, nHeights),dtype='complex'))) |
|
|
6293 | 6293 | else: |
|
|
6294 | 6294 | vStacked = numpy.vstack((numpy.zeros((-idxT, nHeights),dtype='complex'), |
|
|
6295 | 6295 | volts2[:(nPoints + idxT),:])) |
|
|
6296 | 6296 | voltsCCF[i,t,:] = numpy.sum((numpy.conjugate(volts1)*vStacked),axis=0) |
|
|
6297 | 6297 | |
|
|
6298 | 6298 | vStacked = None |
|
|
6299 | 6299 | return voltsCCF |
|
|
6300 | 6300 | |
|
|
6301 | 6301 | def __getNoise(self, power, timeSegment, timeInterval): |
|
|
6302 | 6302 | numProfPerBlock = numpy.ceil(timeSegment/timeInterval) |
|
|
6303 | 6303 | numBlocks = int(power.shape[0]/numProfPerBlock) |
|
|
6304 | 6304 | numHeights = power.shape[1] |
|
|
6305 | 6305 | |
|
|
6306 | 6306 | listPower = numpy.array_split(power, numBlocks, 0) |
|
|
6307 | 6307 | noise = numpy.zeros((power.shape[0], power.shape[1])) |
|
|
6308 | 6308 | noise1 = numpy.zeros((power.shape[0], power.shape[1])) |
|
|
6309 | 6309 | |
|
|
6310 | 6310 | startInd = 0 |
|
|
6311 | 6311 | endInd = 0 |
|
|
6312 | 6312 | |
|
|
6313 | 6313 | for i in range(numBlocks): #split por canal |
|
|
6314 | 6314 | startInd = endInd |
|
|
6315 | 6315 | endInd = endInd + listPower[i].shape[0] |
|
|
6316 | 6316 | |
|
|
6317 | 6317 | arrayBlock = listPower[i] |
|
|
6318 | 6318 | noiseAux = numpy.mean(arrayBlock, 0) |
|
|
6319 | 6319 | # noiseAux = numpy.median(noiseAux) |
|
|
6320 | 6320 | # noiseAux = numpy.mean(arrayBlock) |
|
|
6321 | 6321 | noise[startInd:endInd,:] = noise[startInd:endInd,:] + noiseAux |
|
|
6322 | 6322 | |
|
|
6323 | 6323 | noiseAux1 = numpy.mean(arrayBlock) |
|
|
6324 | 6324 | noise1[startInd:endInd,:] = noise1[startInd:endInd,:] + noiseAux1 |
|
|
6325 | 6325 | |
|
|
6326 | 6326 | return noise, noise1 |
|
|
6327 | 6327 | |
|
|
6328 | 6328 | def __findMeteors(self, power, thresh): |
|
|
6329 | 6329 | nProf = power.shape[0] |
|
|
6330 | 6330 | nHeights = power.shape[1] |
|
|
6331 | 6331 | listMeteors = [] |
|
|
6332 | 6332 | |
|
|
6333 | 6333 | for i in range(nHeights): |
|
|
6334 | 6334 | powerAux = power[:,i] |
|
|
6335 | 6335 | threshAux = thresh[:,i] |
|
|
6336 | 6336 | |
|
|
6337 | 6337 | indUPthresh = numpy.where(powerAux > threshAux)[0] |
|
|
6338 | 6338 | indDNthresh = numpy.where(powerAux <= threshAux)[0] |
|
|
6339 | 6339 | |
|
|
6340 | 6340 | j = 0 |
|
|
6341 | 6341 | |
|
|
6342 | 6342 | while (j < indUPthresh.size - 2): |
|
|
6343 | 6343 | if (indUPthresh[j + 2] == indUPthresh[j] + 2): |
|
|
6344 | 6344 | indDNAux = numpy.where(indDNthresh > indUPthresh[j]) |
|
|
6345 | 6345 | indDNthresh = indDNthresh[indDNAux] |
|
|
6346 | 6346 | |
|
|
6347 | 6347 | if (indDNthresh.size > 0): |
|
|
6348 | 6348 | indEnd = indDNthresh[0] - 1 |
|
|
6349 | 6349 | indInit = indUPthresh[j] |
|
|
6350 | 6350 | |
|
|
6351 | 6351 | meteor = powerAux[indInit:indEnd + 1] |
|
|
6352 | 6352 | indPeak = meteor.argmax() + indInit |
|
|
6353 | 6353 | FLA = sum(numpy.conj(meteor)*numpy.hstack((meteor[1:],0))) |
|
|
6354 | 6354 | |
|
|
6355 | 6355 | listMeteors.append(numpy.array([i,indInit,indPeak,indEnd,FLA])) #CHEQUEAR!!!!! |
|
|
6356 | 6356 | j = numpy.where(indUPthresh == indEnd)[0] + 1 |
|
|
6357 | 6357 | else: j+=1 |
|
|
6358 | 6358 | else: j+=1 |
|
|
6359 | 6359 | |
|
|
6360 | 6360 | return listMeteors |
|
|
6361 | 6361 | |
|
|
6362 | 6362 | def __removeMultipleDetections(self,listMeteors, rangeLimit, timeLimit): |
|
|
6363 | 6363 | |
|
|
6364 | 6364 | arrayMeteors = numpy.asarray(listMeteors) |
|
|
6365 | 6365 | listMeteors1 = [] |
|
|
6366 | 6366 | |
|
|
6367 | 6367 | while arrayMeteors.shape[0] > 0: |
|
|
6368 | 6368 | FLAs = arrayMeteors[:,4] |
|
|
6369 | 6369 | maxFLA = FLAs.argmax() |
|
|
6370 | 6370 | listMeteors1.append(arrayMeteors[maxFLA,:]) |
|
|
6371 | 6371 | |
|
|
6372 | 6372 | MeteorInitTime = arrayMeteors[maxFLA,1] |
|
|
6373 | 6373 | MeteorEndTime = arrayMeteors[maxFLA,3] |
|
|
6374 | 6374 | MeteorHeight = arrayMeteors[maxFLA,0] |
|
|
6375 | 6375 | |
|
|
6376 | 6376 | #Check neighborhood |
|
|
6377 | 6377 | maxHeightIndex = MeteorHeight + rangeLimit |
|
|
6378 | 6378 | minHeightIndex = MeteorHeight - rangeLimit |
|
|
6379 | 6379 | minTimeIndex = MeteorInitTime - timeLimit |
|
|
6380 | 6380 | maxTimeIndex = MeteorEndTime + timeLimit |
|
|
6381 | 6381 | |
|
|
6382 | 6382 | #Check Heights |
|
|
6383 | 6383 | indHeight = numpy.logical_and(arrayMeteors[:,0] >= minHeightIndex, arrayMeteors[:,0] <= maxHeightIndex) |
|
|
6384 | 6384 | indTime = numpy.logical_and(arrayMeteors[:,3] >= minTimeIndex, arrayMeteors[:,1] <= maxTimeIndex) |
|
|
6385 | 6385 | indBoth = numpy.where(numpy.logical_and(indTime,indHeight)) |
|
|
6386 | 6386 | |
|
|
6387 | 6387 | arrayMeteors = numpy.delete(arrayMeteors, indBoth, axis = 0) |
|
|
6388 | 6388 | |
|
|
6389 | 6389 | return listMeteors1 |
|
|
6390 | 6390 | |
|
|
6391 | 6391 | def __meteorReestimation(self, listMeteors, volts, pairslist, thresh, noise, timeInterval,frequency): |
|
|
6392 | 6392 | numHeights = volts.shape[2] |
|
|
6393 | 6393 | nChannel = volts.shape[0] |
|
|
6394 | 6394 | |
|
|
6395 | 6395 | thresholdPhase = thresh[0] |
|
|
6396 | 6396 | thresholdNoise = thresh[1] |
|
|
6397 | 6397 | thresholdDB = float(thresh[2]) |
|
|
6398 | 6398 | |
|
|
6399 | 6399 | thresholdDB1 = 10**(thresholdDB/10) |
|
|
6400 | 6400 | pairsarray = numpy.array(pairslist) |
|
|
6401 | 6401 | indSides = pairsarray[:,1] |
|
|
6402 | 6402 | |
|
|
6403 | 6403 | pairslist1 = list(pairslist) |
|
|
6404 | 6404 | pairslist1.append((0,1)) |
|
|
6405 | 6405 | pairslist1.append((3,4)) |
|
|
6406 | 6406 | |
|
|
6407 | 6407 | listMeteors1 = [] |
|
|
6408 | 6408 | listPowerSeries = [] |
|
|
6409 | 6409 | listVoltageSeries = [] |
|
|
6410 | 6410 | #volts has the war data |
|
|
6411 | 6411 | |
|
|
6412 | 6412 | if frequency == 30e6: |
|
|
6413 | 6413 | timeLag = 45*10**-3 |
|
|
6414 | 6414 | else: |
|
|
6415 | 6415 | timeLag = 15*10**-3 |
|
|
6416 | 6416 | lag = numpy.ceil(timeLag/timeInterval) |
|
|
6417 | 6417 | |
|
|
6418 | 6418 | for i in range(len(listMeteors)): |
|
|
6419 | 6419 | |
|
|
6420 | 6420 | ###################### 3.6 - 3.7 PARAMETERS REESTIMATION ######################### |
|
|
6421 | 6421 | meteorAux = numpy.zeros(16) |
|
|
6422 | 6422 | |
|
|
6423 | 6423 | #Loading meteor Data (mHeight, mStart, mPeak, mEnd) |
|
|
6424 | 6424 | mHeight = listMeteors[i][0] |
|
|
6425 | 6425 | mStart = listMeteors[i][1] |
|
|
6426 | 6426 | mPeak = listMeteors[i][2] |
|
|
6427 | 6427 | mEnd = listMeteors[i][3] |
|
|
6428 | 6428 | |
|
|
6429 | 6429 | #get the volt data between the start and end times of the meteor |
|
|
6430 | 6430 | meteorVolts = volts[:,mStart:mEnd+1,mHeight] |
|
|
6431 | 6431 | meteorVolts = meteorVolts.reshape(meteorVolts.shape[0], meteorVolts.shape[1], 1) |
|
|
6432 | 6432 | |
|
|
6433 | 6433 | #3.6. Phase Difference estimation |
|
|
6434 | 6434 | phaseDiff, aux = self.__estimatePhaseDifference(meteorVolts, pairslist) |
|
|
6435 | 6435 | |
|
|
6436 | 6436 | #3.7. Phase difference removal & meteor start, peak and end times reestimated |
|
|
6437 | 6437 | #meteorVolts0.- all Channels, all Profiles |
|
|
6438 | 6438 | meteorVolts0 = volts[:,:,mHeight] |
|
|
6439 | 6439 | meteorThresh = noise[:,mHeight]*thresholdNoise |
|
|
6440 | 6440 | meteorNoise = noise[:,mHeight] |
|
|
6441 | 6441 | meteorVolts0[indSides,:] = self.__shiftPhase(meteorVolts0[indSides,:], phaseDiff) #Phase Shifting |
|
|
6442 | 6442 | powerNet0 = numpy.nansum(numpy.abs(meteorVolts0)**2, axis = 0) #Power |
|
|
6443 | 6443 | |
|
|
6444 | 6444 | #Times reestimation |
|
|
6445 | 6445 | mStart1 = numpy.where(powerNet0[:mPeak] < meteorThresh[:mPeak])[0] |
|
|
6446 | 6446 | if mStart1.size > 0: |
|
|
6447 | 6447 | mStart1 = mStart1[-1] + 1 |
|
|
6448 | 6448 | |
|
|
6449 | 6449 | else: |
|
|
6450 | 6450 | mStart1 = mPeak |
|
|
6451 | 6451 | |
|
|
6452 | 6452 | mEnd1 = numpy.where(powerNet0[mPeak:] < meteorThresh[mPeak:])[0][0] + mPeak - 1 |
|
|
6453 | 6453 | mEndDecayTime1 = numpy.where(powerNet0[mPeak:] < meteorNoise[mPeak:])[0] |
|
|
6454 | 6454 | if mEndDecayTime1.size == 0: |
|
|
6455 | 6455 | mEndDecayTime1 = powerNet0.size |
|
|
6456 | 6456 | else: |
|
|
6457 | 6457 | mEndDecayTime1 = mEndDecayTime1[0] + mPeak - 1 |
|
|
6458 | 6458 | # mPeak1 = meteorVolts0[mStart1:mEnd1 + 1].argmax() |
|
|
6459 | 6459 | |
|
|
6460 | 6460 | #meteorVolts1.- all Channels, from start to end |
|
|
6461 | 6461 | meteorVolts1 = meteorVolts0[:,mStart1:mEnd1 + 1] |
|
|
6462 | 6462 | meteorVolts2 = meteorVolts0[:,mPeak + lag:mEnd1 + 1] |
|
|
6463 | 6463 | if meteorVolts2.shape[1] == 0: |
|
|
6464 | 6464 | meteorVolts2 = meteorVolts0[:,mPeak:mEnd1 + 1] |
|
|
6465 | 6465 | meteorVolts1 = meteorVolts1.reshape(meteorVolts1.shape[0], meteorVolts1.shape[1], 1) |
|
|
6466 | 6466 | meteorVolts2 = meteorVolts2.reshape(meteorVolts2.shape[0], meteorVolts2.shape[1], 1) |
|
|
6467 | 6467 | ##################### END PARAMETERS REESTIMATION ######################### |
|
|
6468 | 6468 | |
|
|
6469 | 6469 | ##################### 3.8 PHASE DIFFERENCE REESTIMATION ######################## |
|
|
6470 | 6470 | # if mEnd1 - mStart1 > 4: #Error Number 6: echo less than 5 samples long; too short for analysis |
|
|
6471 | 6471 | if meteorVolts2.shape[1] > 0: |
|
|
6472 | 6472 | #Phase Difference re-estimation |
|
|
6473 | 6473 | phaseDiff1, phaseDiffint = self.__estimatePhaseDifference(meteorVolts2, pairslist1) #Phase Difference Estimation |
|
|
6474 | 6474 | # phaseDiff1, phaseDiffint = self.estimatePhaseDifference(meteorVolts2, pairslist) |
|
|
6475 | 6475 | meteorVolts2 = meteorVolts2.reshape(meteorVolts2.shape[0], meteorVolts2.shape[1]) |
|
|
6476 | 6476 | phaseDiff11 = numpy.reshape(phaseDiff1, (phaseDiff1.shape[0],1)) |
|
|
6477 | 6477 | meteorVolts2[indSides,:] = self.__shiftPhase(meteorVolts2[indSides,:], phaseDiff11[0:4]) #Phase Shifting |
|
|
6478 | 6478 | |
|
|
6479 | 6479 | #Phase Difference RMS |
|
|
6480 | 6480 | phaseRMS1 = numpy.sqrt(numpy.mean(numpy.square(phaseDiff1))) |
|
|
6481 | 6481 | powerNet1 = numpy.nansum(numpy.abs(meteorVolts1[:,:])**2,0) |
|
|
6482 | 6482 | #Data from Meteor |
|
|
6483 | 6483 | mPeak1 = powerNet1.argmax() + mStart1 |
|
|
6484 | 6484 | mPeakPower1 = powerNet1.max() |
|
|
6485 | 6485 | noiseAux = sum(noise[mStart1:mEnd1 + 1,mHeight]) |
|
|
6486 | 6486 | mSNR1 = (sum(powerNet1)-noiseAux)/noiseAux |
|
|
6487 | 6487 | Meteor1 = numpy.array([mHeight, mStart1, mPeak1, mEnd1, mPeakPower1, mSNR1, phaseRMS1]) |
|
|
6488 | 6488 | Meteor1 = numpy.hstack((Meteor1,phaseDiffint)) |
|
|
6489 | 6489 | PowerSeries = powerNet0[mStart1:mEndDecayTime1 + 1] |
|
|
6490 | 6490 | #Vectorize |
|
|
6491 | 6491 | meteorAux[0:7] = [mHeight, mStart1, mPeak1, mEnd1, mPeakPower1, mSNR1, phaseRMS1] |
|
|
6492 | 6492 | meteorAux[7:11] = phaseDiffint[0:4] |
|
|
6493 | 6493 | |
|
|
6494 | 6494 | #Rejection Criterions |
|
|
6495 | 6495 | if phaseRMS1 > thresholdPhase: #Error Number 17: Phase variation |
|
|
6496 | 6496 | meteorAux[-1] = 17 |
|
|
6497 | 6497 | elif mSNR1 < thresholdDB1: #Error Number 1: SNR < threshold dB |
|
|
6498 | 6498 | meteorAux[-1] = 1 |
|
|
6499 | 6499 | |
|
|
6500 | 6500 | |
|
|
6501 | 6501 | else: |
|
|
6502 | 6502 | meteorAux[0:4] = [mHeight, mStart, mPeak, mEnd] |
|
|
6503 | 6503 | meteorAux[-1] = 6 #Error Number 6: echo less than 5 samples long; too short for analysis |
|
|
6504 | 6504 | PowerSeries = 0 |
|
|
6505 | 6505 | |
|
|
6506 | 6506 | listMeteors1.append(meteorAux) |
|
|
6507 | 6507 | listPowerSeries.append(PowerSeries) |
|
|
6508 | 6508 | listVoltageSeries.append(meteorVolts1) |
|
|
6509 | 6509 | |
|
|
6510 | 6510 | return listMeteors1, listPowerSeries, listVoltageSeries |
|
|
6511 | 6511 | |
|
|
6512 | 6512 | def __estimateDecayTime(self, listMeteors, listPower, timeInterval, frequency): |
|
|
6513 | 6513 | |
|
|
6514 | 6514 | threshError = 10 |
|
|
6515 | 6515 | #Depending if it is 30 or 50 MHz |
|
|
6516 | 6516 | if frequency == 30e6: |
|
|
6517 | 6517 | timeLag = 45*10**-3 |
|
|
6518 | 6518 | else: |
|
|
6519 | 6519 | timeLag = 15*10**-3 |
|
|
6520 | 6520 | lag = numpy.ceil(timeLag/timeInterval) |
|
|
6521 | 6521 | |
|
|
6522 | 6522 | listMeteors1 = [] |
|
|
6523 | 6523 | |
|
|
6524 | 6524 | for i in range(len(listMeteors)): |
|
|
6525 | 6525 | meteorPower = listPower[i] |
|
|
6526 | 6526 | meteorAux = listMeteors[i] |
|
|
6527 | 6527 | |
|
|
6528 | 6528 | if meteorAux[-1] == 0: |
|
|
6529 | 6529 | |
|
|
6530 | 6530 | try: |
|
|
6531 | 6531 | indmax = meteorPower.argmax() |
|
|
6532 | 6532 | indlag = indmax + lag |
|
|
6533 | 6533 | |
|
|
6534 | 6534 | y = meteorPower[indlag:] |
|
|
6535 | 6535 | x = numpy.arange(0, y.size)*timeLag |
|
|
6536 | 6536 | |
|
|
6537 | 6537 | #first guess |
|
|
6538 | 6538 | a = y[0] |
|
|
6539 | 6539 | tau = timeLag |
|
|
6540 | 6540 | #exponential fit |
|
|
6541 | 6541 | popt, pcov = optimize.curve_fit(self.__exponential_function, x, y, p0 = [a, tau]) |
|
|
6542 | 6542 | y1 = self.__exponential_function(x, *popt) |
|
|
6543 | 6543 | #error estimation |
|
|
6544 | 6544 | error = sum((y - y1)**2)/(numpy.var(y)*(y.size - popt.size)) |
|
|
6545 | 6545 | |
|
|
6546 | 6546 | decayTime = popt[1] |
|
|
6547 | 6547 | riseTime = indmax*timeInterval |
|
|
6548 | 6548 | meteorAux[11:13] = [decayTime, error] |
|
|
6549 | 6549 | |
|
|
6550 | 6550 | #Table items 7, 8 and 11 |
|
|
6551 | 6551 | if (riseTime > 0.3): #Number 7: Echo rise exceeds 0.3s |
|
|
6552 | 6552 | meteorAux[-1] = 7 |
|
|
6553 | 6553 | elif (decayTime < 2*riseTime) : #Number 8: Echo decay time less than than twice rise time |
|
|
6554 | 6554 | meteorAux[-1] = 8 |
|
|
6555 | 6555 | if (error > threshError): #Number 11: Poor fit to amplitude for estimation of decay time |
|
|
6556 | 6556 | meteorAux[-1] = 11 |
|
|
6557 | 6557 | |
|
|
6558 | 6558 | |
|
|
6559 | 6559 | except: |
|
|
6560 | 6560 | meteorAux[-1] = 11 |
|
|
6561 | 6561 | |
|
|
6562 | 6562 | |
|
|
6563 | 6563 | listMeteors1.append(meteorAux) |
|
|
6564 | 6564 | |
|
|
6565 | 6565 | return listMeteors1 |
|
|
6566 | 6566 | |
|
|
6567 | 6567 | #Exponential Function |
|
|
6568 | 6568 | |
|
|
6569 | 6569 | def __exponential_function(self, x, a, tau): |
|
|
6570 | 6570 | y = a*numpy.exp(-x/tau) |
|
|
6571 | 6571 | return y |
|
|
6572 | 6572 | |
|
|
6573 | 6573 | def __getRadialVelocity(self, listMeteors, listVolts, radialStdThresh, pairslist, timeInterval): |
|
|
6574 | 6574 | |
|
|
6575 | 6575 | pairslist1 = list(pairslist) |
|
|
6576 | 6576 | pairslist1.append((0,1)) |
|
|
6577 | 6577 | pairslist1.append((3,4)) |
|
|
6578 | 6578 | numPairs = len(pairslist1) |
|
|
6579 | 6579 | #Time Lag |
|
|
6580 | 6580 | timeLag = 45*10**-3 |
|
|
6581 | 6581 | c = 3e8 |
|
|
6582 | 6582 | lag = numpy.ceil(timeLag/timeInterval) |
|
|
6583 | 6583 | freq = 30e6 |
|
|
6584 | 6584 | |
|
|
6585 | 6585 | listMeteors1 = [] |
|
|
6586 | 6586 | |
|
|
6587 | 6587 | for i in range(len(listMeteors)): |
|
|
6588 | 6588 | meteorAux = listMeteors[i] |
|
|
6589 | 6589 | if meteorAux[-1] == 0: |
|
|
6590 | 6590 | mStart = listMeteors[i][1] |
|
|
6591 | 6591 | mPeak = listMeteors[i][2] |
|
|
6592 | 6592 | mLag = mPeak - mStart + lag |
|
|
6593 | 6593 | |
|
|
6594 | 6594 | #get the volt data between the start and end times of the meteor |
|
|
6595 | 6595 | meteorVolts = listVolts[i] |
|
|
6596 | 6596 | meteorVolts = meteorVolts.reshape(meteorVolts.shape[0], meteorVolts.shape[1], 1) |
|
|
6597 | 6597 | |
|
|
6598 | 6598 | #Get CCF |
|
|
6599 | 6599 | allCCFs = self.__calculateCCF(meteorVolts, pairslist1, [-2,-1,0,1,2]) |
|
|
6600 | 6600 | |
|
|
6601 | 6601 | #Method 2 |
|
|
6602 | 6602 | slopes = numpy.zeros(numPairs) |
|
|
6603 | 6603 | time = numpy.array([-2,-1,1,2])*timeInterval |
|
|
6604 | 6604 | angAllCCF = numpy.angle(allCCFs[:,[0,1,3,4],0]) |
|
|
6605 | 6605 | |
|
|
6606 | 6606 | #Correct phases |
|
|
6607 | 6607 | derPhaseCCF = angAllCCF[:,1:] - angAllCCF[:,0:-1] |
|
|
6608 | 6608 | indDer = numpy.where(numpy.abs(derPhaseCCF) > numpy.pi) |
|
|
6609 | 6609 | |
|
|
6610 | 6610 | if indDer[0].shape[0] > 0: |
|
|
6611 | 6611 | for i in range(indDer[0].shape[0]): |
|
|
6612 | 6612 | signo = -numpy.sign(derPhaseCCF[indDer[0][i],indDer[1][i]]) |
|
|
6613 | 6613 | angAllCCF[indDer[0][i],indDer[1][i]+1:] += signo*2*numpy.pi |
|
|
6614 | 6614 | |
|
|
6615 | 6615 | # fit = scipy.stats.linregress(numpy.array([-2,-1,1,2])*timeInterval, numpy.array([phaseLagN2s[i],phaseLagN1s[i],phaseLag1s[i],phaseLag2s[i]])) |
|
|
6616 | 6616 | for j in range(numPairs): |
|
|
6617 | 6617 | fit = stats.linregress(time, angAllCCF[j,:]) |
|
|
6618 | 6618 | slopes[j] = fit[0] |
|
|
6619 | 6619 | |
|
|
6620 | 6620 | #Remove Outlier |
|
|
6621 | 6621 | # indOut = numpy.argmax(numpy.abs(slopes - numpy.mean(slopes))) |
|
|
6622 | 6622 | # slopes = numpy.delete(slopes,indOut) |
|
|
6623 | 6623 | # indOut = numpy.argmax(numpy.abs(slopes - numpy.mean(slopes))) |
|
|
6624 | 6624 | # slopes = numpy.delete(slopes,indOut) |
|
|
6625 | 6625 | |
|
|
6626 | 6626 | radialVelocity = -numpy.mean(slopes)*(0.25/numpy.pi)*(c/freq) |
|
|
6627 | 6627 | radialError = numpy.std(slopes)*(0.25/numpy.pi)*(c/freq) |
|
|
6628 | 6628 | meteorAux[-2] = radialError |
|
|
6629 | 6629 | meteorAux[-3] = radialVelocity |
|
|
6630 | 6630 | |
|
|
6631 | 6631 | #Setting Error |
|
|
6632 | 6632 | #Number 15: Radial Drift velocity or projected horizontal velocity exceeds 200 m/s |
|
|
6633 | 6633 | if numpy.abs(radialVelocity) > 200: |
|
|
6634 | 6634 | meteorAux[-1] = 15 |
|
|
6635 | 6635 | #Number 12: Poor fit to CCF variation for estimation of radial drift velocity |
|
|
6636 | 6636 | elif radialError > radialStdThresh: |
|
|
6637 | 6637 | meteorAux[-1] = 12 |
|
|
6638 | 6638 | |
|
|
6639 | 6639 | listMeteors1.append(meteorAux) |
|
|
6640 | 6640 | return listMeteors1 |
|
|
6641 | 6641 | |
|
|
6642 | 6642 | def __setNewArrays(self, listMeteors, date, heiRang): |
|
|
6643 | 6643 | |
|
|
6644 | 6644 | #New arrays |
|
|
6645 | 6645 | arrayMeteors = numpy.array(listMeteors) |
|
|
6646 | 6646 | arrayParameters = numpy.zeros((len(listMeteors), 13)) |
|
|
6647 | 6647 | |
|
|
6648 | 6648 | #Date inclusion |
|
|
6649 | 6649 | # date = re.findall(r'\((.*?)\)', date) |
|
|
6650 | 6650 | # date = date[0].split(',') |
|
|
6651 | 6651 | # date = map(int, date) |
|
|
6652 | 6652 | # |
|
|
6653 | 6653 | # if len(date)<6: |
|
|
6654 | 6654 | # date.append(0) |
|
|
6655 | 6655 | # |
|
|
6656 | 6656 | # date = [date[0]*10000 + date[1]*100 + date[2], date[3]*10000 + date[4]*100 + date[5]] |
|
|
6657 | 6657 | # arrayDate = numpy.tile(date, (len(listMeteors), 1)) |
|
|
6658 | 6658 | arrayDate = numpy.tile(date, (len(listMeteors))) |
|
|
6659 | 6659 | |
|
|
6660 | 6660 | #Meteor array |
|
|
6661 | 6661 | # arrayMeteors[:,0] = heiRang[arrayMeteors[:,0].astype(int)] |
|
|
6662 | 6662 | # arrayMeteors = numpy.hstack((arrayDate, arrayMeteors)) |
|
|
6663 | 6663 | |
|
|
6664 | 6664 | #Parameters Array |
|
|
6665 | 6665 | arrayParameters[:,0] = arrayDate #Date |
|
|
6666 | 6666 | arrayParameters[:,1] = heiRang[arrayMeteors[:,0].astype(int)] #Range |
|
|
6667 | 6667 | arrayParameters[:,6:8] = arrayMeteors[:,-3:-1] #Radial velocity and its error |
|
|
6668 | 6668 | arrayParameters[:,8:12] = arrayMeteors[:,7:11] #Phases |
|
|
6669 | 6669 | arrayParameters[:,-1] = arrayMeteors[:,-1] #Error |
|
|
6670 | 6670 | |
|
|
6671 | 6671 | |
|
|
6672 | 6672 | return arrayParameters |
|
|
6673 | 6673 | |
|
|
6674 | 6674 | class CorrectSMPhases(Operation): |
|
|
6675 | 6675 | |
|
|
6676 | 6676 | def run(self, dataOut, phaseOffsets, hmin = 50, hmax = 150, azimuth = 45, channelPositions = None): |
|
|
6677 | 6677 | |
|
|
6678 | 6678 | arrayParameters = dataOut.data_param |
|
|
6679 | 6679 | pairsList = [] |
|
|
6680 | 6680 | pairx = (0,1) |
|
|
6681 | 6681 | pairy = (2,3) |
|
|
6682 | 6682 | pairsList.append(pairx) |
|
|
6683 | 6683 | pairsList.append(pairy) |
|
|
6684 | 6684 | jph = numpy.zeros(4) |
|
|
6685 | 6685 | |
|
|
6686 | 6686 | phaseOffsets = numpy.array(phaseOffsets)*numpy.pi/180 |
|
|
6687 | 6687 | # arrayParameters[:,8:12] = numpy.unwrap(arrayParameters[:,8:12] + phaseOffsets) |
|
|
6688 | 6688 | arrayParameters[:,8:12] = numpy.angle(numpy.exp(1j*(arrayParameters[:,8:12] + phaseOffsets))) |
|
|
6689 | 6689 | |
|
|
6690 | 6690 | meteorOps = SMOperations() |
|
|
6691 | 6691 | if channelPositions is None: |
|
|
6692 | 6692 | # channelPositions = [(2.5,0), (0,2.5), (0,0), (0,4.5), (-2,0)] #T |
|
|
6693 | 6693 | channelPositions = [(4.5,2), (2,4.5), (2,2), (2,0), (0,2)] #Estrella |
|
|
6694 | 6694 | |
|
|
6695 | 6695 | pairslist0, distances = meteorOps.getPhasePairs(channelPositions) |
|
|
6696 | 6696 | h = (hmin,hmax) |
|
|
6697 | 6697 | |
|
|
6698 | 6698 | arrayParameters = meteorOps.getMeteorParams(arrayParameters, azimuth, h, pairsList, distances, jph) |
|
|
6699 | 6699 | |
|
|
6700 | 6700 | dataOut.data_param = arrayParameters |
|
|
6701 | 6701 | return |
|
|
6702 | 6702 | |
|
|
6703 | 6703 | class SMPhaseCalibration(Operation): |
|
|
6704 | 6704 | |
|
|
6705 | 6705 | __buffer = None |
|
|
6706 | 6706 | |
|
|
6707 | 6707 | __initime = None |
|
|
6708 | 6708 | |
|
|
6709 | 6709 | __dataReady = False |
|
|
6710 | 6710 | |
|
|
6711 | 6711 | __isConfig = False |
|
|
6712 | 6712 | |
|
|
6713 | 6713 | def __checkTime(self, currentTime, initTime, paramInterval, outputInterval): |
|
|
6714 | 6714 | |
|
|
6715 | 6715 | dataTime = currentTime + paramInterval |
|
|
6716 | 6716 | deltaTime = dataTime - initTime |
|
|
6717 | 6717 | |
|
|
6718 | 6718 | if deltaTime >= outputInterval or deltaTime < 0: |
|
|
6719 | 6719 | return True |
|
|
6720 | 6720 | |
|
|
6721 | 6721 | return False |
|
|
6722 | 6722 | |
|
|
6723 | 6723 | def __getGammas(self, pairs, d, phases): |
|
|
6724 | 6724 | gammas = numpy.zeros(2) |
|
|
6725 | 6725 | |
|
|
6726 | 6726 | for i in range(len(pairs)): |
|
|
6727 | 6727 | |
|
|
6728 | 6728 | pairi = pairs[i] |
|
|
6729 | 6729 | |
|
|
6730 | 6730 | phip3 = phases[:,pairi[0]] |
|
|
6731 | 6731 | d3 = d[pairi[0]] |
|
|
6732 | 6732 | phip2 = phases[:,pairi[1]] |
|
|
6733 | 6733 | d2 = d[pairi[1]] |
|
|
6734 | 6734 | #Calculating gamma |
|
|
6735 | 6735 | # jdcos = alp1/(k*d1) |
|
|
6736 | 6736 | # jgamma = numpy.angle(numpy.exp(1j*(d0*alp1/d1 - alp0))) |
|
|
6737 | 6737 | jgamma = -phip2*d3/d2 - phip3 |
|
|
6738 | 6738 | jgamma = numpy.angle(numpy.exp(1j*jgamma)) |
|
|
6739 | 6739 | # jgamma[jgamma>numpy.pi] -= 2*numpy.pi |
|
|
6740 | 6740 | # jgamma[jgamma<-numpy.pi] += 2*numpy.pi |
|
|
6741 | 6741 | |
|
|
6742 | 6742 | #Revised distribution |
|
|
6743 | 6743 | jgammaArray = numpy.hstack((jgamma,jgamma+0.5*numpy.pi,jgamma-0.5*numpy.pi)) |
|
|
6744 | 6744 | |
|
|
6745 | 6745 | #Histogram |
|
|
6746 | 6746 | nBins = 64 |
|
|
6747 | 6747 | rmin = -0.5*numpy.pi |
|
|
6748 | 6748 | rmax = 0.5*numpy.pi |
|
|
6749 | 6749 | phaseHisto = numpy.histogram(jgammaArray, bins=nBins, range=(rmin,rmax)) |
|
|
6750 | 6750 | |
|
|
6751 | 6751 | meteorsY = phaseHisto[0] |
|
|
6752 | 6752 | phasesX = phaseHisto[1][:-1] |
|
|
6753 | 6753 | width = phasesX[1] - phasesX[0] |
|
|
6754 | 6754 | phasesX += width/2 |
|
|
6755 | 6755 | |
|
|
6756 | 6756 | #Gaussian aproximation |
|
|
6757 | 6757 | bpeak = meteorsY.argmax() |
|
|
6758 | 6758 | peak = meteorsY.max() |
|
|
6759 | 6759 | jmin = bpeak - 5 |
|
|
6760 | 6760 | jmax = bpeak + 5 + 1 |
|
|
6761 | 6761 | |
|
|
6762 | 6762 | if jmin<0: |
|
|
6763 | 6763 | jmin = 0 |
|
|
6764 | 6764 | jmax = 6 |
|
|
6765 | 6765 | elif jmax > meteorsY.size: |
|
|
6766 | 6766 | jmin = meteorsY.size - 6 |
|
|
6767 | 6767 | jmax = meteorsY.size |
|
|
6768 | 6768 | |
|
|
6769 | 6769 | x0 = numpy.array([peak,bpeak,50]) |
|
|
6770 | 6770 | coeff = optimize.leastsq(self.__residualFunction, x0, args=(meteorsY[jmin:jmax], phasesX[jmin:jmax])) |
|
|
6771 | 6771 | |
|
|
6772 | 6772 | #Gammas |
|
|
6773 | 6773 | gammas[i] = coeff[0][1] |
|
|
6774 | 6774 | |
|
|
6775 | 6775 | return gammas |
|
|
6776 | 6776 | |
|
|
6777 | 6777 | def __residualFunction(self, coeffs, y, t): |
|
|
6778 | 6778 | |
|
|
6779 | 6779 | return y - self.__gauss_function(t, coeffs) |
|
|
6780 | 6780 | |
|
|
6781 | 6781 | def __gauss_function(self, t, coeffs): |
|
|
6782 | 6782 | |
|
|
6783 | 6783 | return coeffs[0]*numpy.exp(-0.5*((t - coeffs[1]) / coeffs[2])**2) |
|
|
6784 | 6784 | |
|
|
6785 | 6785 | def __getPhases(self, azimuth, h, pairsList, d, gammas, meteorsArray): |
|
|
6786 | 6786 | meteorOps = SMOperations() |
|
|
6787 | 6787 | nchan = 4 |
|
|
6788 | 6788 | pairx = pairsList[0] #x es 0 |
|
|
6789 | 6789 | pairy = pairsList[1] #y es 1 |
|
|
6790 | 6790 | center_xangle = 0 |
|
|
6791 | 6791 | center_yangle = 0 |
|
|
6792 | 6792 | range_angle = numpy.array([10*numpy.pi,numpy.pi,numpy.pi/2,numpy.pi/4]) |
|
|
6793 | 6793 | ntimes = len(range_angle) |
|
|
6794 | 6794 | |
|
|
6795 | 6795 | nstepsx = 20 |
|
|
6796 | 6796 | nstepsy = 20 |
|
|
6797 | 6797 | |
|
|
6798 | 6798 | for iz in range(ntimes): |
|
|
6799 | 6799 | min_xangle = -range_angle[iz]/2 + center_xangle |
|
|
6800 | 6800 | max_xangle = range_angle[iz]/2 + center_xangle |
|
|
6801 | 6801 | min_yangle = -range_angle[iz]/2 + center_yangle |
|
|
6802 | 6802 | max_yangle = range_angle[iz]/2 + center_yangle |
|
|
6803 | 6803 | |
|
|
6804 | 6804 | inc_x = (max_xangle-min_xangle)/nstepsx |
|
|
6805 | 6805 | inc_y = (max_yangle-min_yangle)/nstepsy |
|
|
6806 | 6806 | |
|
|
6807 | 6807 | alpha_y = numpy.arange(nstepsy)*inc_y + min_yangle |
|
|
6808 | 6808 | alpha_x = numpy.arange(nstepsx)*inc_x + min_xangle |
|
|
6809 | 6809 | penalty = numpy.zeros((nstepsx,nstepsy)) |
|
|
6810 | 6810 | jph_array = numpy.zeros((nchan,nstepsx,nstepsy)) |
|
|
6811 | 6811 | jph = numpy.zeros(nchan) |
|
|
6812 | 6812 | |
|
|
6813 | 6813 | # Iterations looking for the offset |
|
|
6814 | 6814 | for iy in range(int(nstepsy)): |
|
|
6815 | 6815 | for ix in range(int(nstepsx)): |
|
|
6816 | 6816 | d3 = d[pairsList[1][0]] |
|
|
6817 | 6817 | d2 = d[pairsList[1][1]] |
|
|
6818 | 6818 | d5 = d[pairsList[0][0]] |
|
|
6819 | 6819 | d4 = d[pairsList[0][1]] |
|
|
6820 | 6820 | |
|
|
6821 | 6821 | alp2 = alpha_y[iy] #gamma 1 |
|
|
6822 | 6822 | alp4 = alpha_x[ix] #gamma 0 |
|
|
6823 | 6823 | |
|
|
6824 | 6824 | alp3 = -alp2*d3/d2 - gammas[1] |
|
|
6825 | 6825 | alp5 = -alp4*d5/d4 - gammas[0] |
|
|
6826 | 6826 | # jph[pairy[1]] = alpha_y[iy] |
|
|
6827 | 6827 | # jph[pairy[0]] = -gammas[1] - alpha_y[iy]*d[pairy[1]]/d[pairy[0]] |
|
|
6828 | 6828 | |
|
|
6829 | 6829 | # jph[pairx[1]] = alpha_x[ix] |
|
|
6830 | 6830 | # jph[pairx[0]] = -gammas[0] - alpha_x[ix]*d[pairx[1]]/d[pairx[0]] |
|
|
6831 | 6831 | jph[pairsList[0][1]] = alp4 |
|
|
6832 | 6832 | jph[pairsList[0][0]] = alp5 |
|
|
6833 | 6833 | jph[pairsList[1][0]] = alp3 |
|
|
6834 | 6834 | jph[pairsList[1][1]] = alp2 |
|
|
6835 | 6835 | jph_array[:,ix,iy] = jph |
|
|
6836 | 6836 | # d = [2.0,2.5,2.5,2.0] |
|
|
6837 | 6837 | #falta chequear si va a leer bien los meteoros |
|
|
6838 | 6838 | meteorsArray1 = meteorOps.getMeteorParams(meteorsArray, azimuth, h, pairsList, d, jph) |
|
|
6839 | 6839 | error = meteorsArray1[:,-1] |
|
|
6840 | 6840 | ind1 = numpy.where(error==0)[0] |
|
|
6841 | 6841 | penalty[ix,iy] = ind1.size |
|
|
6842 | 6842 | |
|
|
6843 | 6843 | i,j = numpy.unravel_index(penalty.argmax(), penalty.shape) |
|
|
6844 | 6844 | phOffset = jph_array[:,i,j] |
|
|
6845 | 6845 | |
|
|
6846 | 6846 | center_xangle = phOffset[pairx[1]] |
|
|
6847 | 6847 | center_yangle = phOffset[pairy[1]] |
|
|
6848 | 6848 | |
|
|
6849 | 6849 | phOffset = numpy.angle(numpy.exp(1j*jph_array[:,i,j])) |
|
|
6850 | 6850 | phOffset = phOffset*180/numpy.pi |
|
|
6851 | 6851 | return phOffset |
|
|
6852 | 6852 | |
|
|
6853 | 6853 | |
|
|
6854 | 6854 | def run(self, dataOut, hmin, hmax, channelPositions=None, nHours = 1): |
|
|
6855 | 6855 | |
|
|
6856 | 6856 | dataOut.flagNoData = True |
|
|
6857 | 6857 | self.__dataReady = False |
|
|
6858 | 6858 | dataOut.outputInterval = nHours*3600 |
|
|
6859 | 6859 | |
|
|
6860 | 6860 | if self.__isConfig == False: |
|
|
6861 | 6861 | # self.__initime = dataOut.datatime.replace(minute = 0, second = 0, microsecond = 03) |
|
|
6862 | 6862 | #Get Initial LTC time |
|
|
6863 | 6863 | self.__initime = datetime.datetime.utcfromtimestamp(dataOut.utctime) |
|
|
6864 | 6864 | self.__initime = (self.__initime.replace(minute = 0, second = 0, microsecond = 0) - datetime.datetime(1970, 1, 1)).total_seconds() |
|
|
6865 | 6865 | |
|
|
6866 | 6866 | self.__isConfig = True |
|
|
6867 | 6867 | |
|
|
6868 | 6868 | if self.__buffer is None: |
|
|
6869 | 6869 | self.__buffer = dataOut.data_param.copy() |
|
|
6870 | 6870 | |
|
|
6871 | 6871 | else: |
|
|
6872 | 6872 | self.__buffer = numpy.vstack((self.__buffer, dataOut.data_param)) |
|
|
6873 | 6873 | |
|
|
6874 | 6874 | self.__dataReady = self.__checkTime(dataOut.utctime, self.__initime, dataOut.paramInterval, dataOut.outputInterval) #Check if the buffer is ready |
|
|
6875 | 6875 | |
|
|
6876 | 6876 | if self.__dataReady: |
|
|
6877 | 6877 | dataOut.utctimeInit = self.__initime |
|
|
6878 | 6878 | self.__initime += dataOut.outputInterval #to erase time offset |
|
|
6879 | 6879 | |
|
|
6880 | 6880 | freq = dataOut.frequency |
|
|
6881 | 6881 | c = dataOut.C #m/s |
|
|
6882 | 6882 | lamb = c/freq |
|
|
6883 | 6883 | k = 2*numpy.pi/lamb |
|
|
6884 | 6884 | azimuth = 0 |
|
|
6885 | 6885 | h = (hmin, hmax) |
|
|
6886 | 6886 | # pairs = ((0,1),(2,3)) #Estrella |
|
|
6887 | 6887 | # pairs = ((1,0),(2,3)) #T |
|
|
6888 | 6888 | |
|
|
6889 | 6889 | if channelPositions is None: |
|
|
6890 | 6890 | # channelPositions = [(2.5,0), (0,2.5), (0,0), (0,4.5), (-2,0)] #T |
|
|
6891 | 6891 | channelPositions = [(4.5,2), (2,4.5), (2,2), (2,0), (0,2)] #Estrella |
|
|
6892 | 6892 | meteorOps = SMOperations() |
|
|
6893 | 6893 | pairslist0, distances = meteorOps.getPhasePairs(channelPositions) |
|
|
6894 | 6894 | |
|
|
6895 | 6895 | #Checking correct order of pairs |
|
|
6896 | 6896 | pairs = [] |
|
|
6897 | 6897 | if distances[1] > distances[0]: |
|
|
6898 | 6898 | pairs.append((1,0)) |
|
|
6899 | 6899 | else: |
|
|
6900 | 6900 | pairs.append((0,1)) |
|
|
6901 | 6901 | |
|
|
6902 | 6902 | if distances[3] > distances[2]: |
|
|
6903 | 6903 | pairs.append((3,2)) |
|
|
6904 | 6904 | else: |
|
|
6905 | 6905 | pairs.append((2,3)) |
|
|
6906 | 6906 | # distances1 = [-distances[0]*lamb, distances[1]*lamb, -distances[2]*lamb, distances[3]*lamb] |
|
|
6907 | 6907 | |
|
|
6908 | 6908 | meteorsArray = self.__buffer |
|
|
6909 | 6909 | error = meteorsArray[:,-1] |
|
|
6910 | 6910 | boolError = (error==0)|(error==3)|(error==4)|(error==13)|(error==14) |
|
|
6911 | 6911 | ind1 = numpy.where(boolError)[0] |
|
|
6912 | 6912 | meteorsArray = meteorsArray[ind1,:] |
|
|
6913 | 6913 | meteorsArray[:,-1] = 0 |
|
|
6914 | 6914 | phases = meteorsArray[:,8:12] |
|
|
6915 | 6915 | |
|
|
6916 | 6916 | #Calculate Gammas |
|
|
6917 | 6917 | gammas = self.__getGammas(pairs, distances, phases) |
|
|
6918 | 6918 | # gammas = numpy.array([-21.70409463,45.76935864])*numpy.pi/180 |
|
|
6919 | 6919 | #Calculate Phases |
|
|
6920 | 6920 | phasesOff = self.__getPhases(azimuth, h, pairs, distances, gammas, meteorsArray) |
|
|
6921 | 6921 | phasesOff = phasesOff.reshape((1,phasesOff.size)) |
|
|
6922 | 6922 | dataOut.data_output = -phasesOff |
|
|
6923 | 6923 | dataOut.flagNoData = False |
|
|
6924 | 6924 | self.__buffer = None |
|
|
6925 | 6925 | |
|
|
6926 | 6926 | |
|
|
6927 | 6927 | return |
|
|
6928 | 6928 | |
|
|
6929 | 6929 | class SMOperations(): |
|
|
6930 | 6930 | |
|
|
6931 | 6931 | def __init__(self): |
|
|
6932 | 6932 | |
|
|
6933 | 6933 | return |
|
|
6934 | 6934 | |
|
|
6935 | 6935 | def getMeteorParams(self, arrayParameters0, azimuth, h, pairsList, distances, jph): |
|
|
6936 | 6936 | |
|
|
6937 | 6937 | arrayParameters = arrayParameters0.copy() |
|
|
6938 | 6938 | hmin = h[0] |
|
|
6939 | 6939 | hmax = h[1] |
|
|
6940 | 6940 | |
|
|
6941 | 6941 | #Calculate AOA (Error N 3, 4) |
|
|
6942 | 6942 | #JONES ET AL. 1998 |
|
|
6943 | 6943 | AOAthresh = numpy.pi/8 |
|
|
6944 | 6944 | error = arrayParameters[:,-1] |
|
|
6945 | 6945 | phases = -arrayParameters[:,8:12] + jph |
|
|
6946 | 6946 | # phases = numpy.unwrap(phases) |
|
|
6947 | 6947 | arrayParameters[:,3:6], arrayParameters[:,-1] = self.__getAOA(phases, pairsList, distances, error, AOAthresh, azimuth) |
|
|
6948 | 6948 | |
|
|
6949 | 6949 | #Calculate Heights (Error N 13 and 14) |
|
|
6950 | 6950 | error = arrayParameters[:,-1] |
|
|
6951 | 6951 | Ranges = arrayParameters[:,1] |
|
|
6952 | 6952 | zenith = arrayParameters[:,4] |
|
|
6953 | 6953 | arrayParameters[:,2], arrayParameters[:,-1] = self.__getHeights(Ranges, zenith, error, hmin, hmax) |
|
|
6954 | 6954 | |
|
|
6955 | 6955 | #----------------------- Get Final data ------------------------------------ |
|
|
6956 | 6956 | # error = arrayParameters[:,-1] |
|
|
6957 | 6957 | # ind1 = numpy.where(error==0)[0] |
|
|
6958 | 6958 | # arrayParameters = arrayParameters[ind1,:] |
|
|
6959 | 6959 | |
|
|
6960 | 6960 | return arrayParameters |
|
|
6961 | 6961 | |
|
|
6962 | 6962 | def __getAOA(self, phases, pairsList, directions, error, AOAthresh, azimuth): |
|
|
6963 | 6963 | |
|
|
6964 | 6964 | arrayAOA = numpy.zeros((phases.shape[0],3)) |
|
|
6965 | 6965 | cosdir0, cosdir = self.__getDirectionCosines(phases, pairsList,directions) |
|
|
6966 | 6966 | |
|
|
6967 | 6967 | arrayAOA[:,:2] = self.__calculateAOA(cosdir, azimuth) |
|
|
6968 | 6968 | cosDirError = numpy.sum(numpy.abs(cosdir0 - cosdir), axis = 1) |
|
|
6969 | 6969 | arrayAOA[:,2] = cosDirError |
|
|
6970 | 6970 | |
|
|
6971 | 6971 | azimuthAngle = arrayAOA[:,0] |
|
|
6972 | 6972 | zenithAngle = arrayAOA[:,1] |
|
|
6973 | 6973 | |
|
|
6974 | 6974 | #Setting Error |
|
|
6975 | 6975 | indError = numpy.where(numpy.logical_or(error == 3, error == 4))[0] |
|
|
6976 | 6976 | error[indError] = 0 |
|
|
6977 | 6977 | #Number 3: AOA not fesible |
|
|
6978 | 6978 | indInvalid = numpy.where(numpy.logical_and((numpy.logical_or(numpy.isnan(zenithAngle), numpy.isnan(azimuthAngle))),error == 0))[0] |
|
|
6979 | 6979 | error[indInvalid] = 3 |
|
|
6980 | 6980 | #Number 4: Large difference in AOAs obtained from different antenna baselines |
|
|
6981 | 6981 | indInvalid = numpy.where(numpy.logical_and(cosDirError > AOAthresh,error == 0))[0] |
|
|
6982 | 6982 | error[indInvalid] = 4 |
|
|
6983 | 6983 | return arrayAOA, error |
|
|
6984 | 6984 | |
|
|
6985 | 6985 | def __getDirectionCosines(self, arrayPhase, pairsList, distances): |
|
|
6986 | 6986 | |
|
|
6987 | 6987 | #Initializing some variables |
|
|
6988 | 6988 | ang_aux = numpy.array([-8,-7,-6,-5,-4,-3,-2,-1,0,1,2,3,4,5,6,7,8])*2*numpy.pi |
|
|
6989 | 6989 | ang_aux = ang_aux.reshape(1,ang_aux.size) |
|
|
6990 | 6990 | |
|
|
6991 | 6991 | cosdir = numpy.zeros((arrayPhase.shape[0],2)) |
|
|
6992 | 6992 | cosdir0 = numpy.zeros((arrayPhase.shape[0],2)) |
|
|
6993 | 6993 | |
|
|
6994 | 6994 | |
|
|
6995 | 6995 | for i in range(2): |
|
|
6996 | 6996 | ph0 = arrayPhase[:,pairsList[i][0]] |
|
|
6997 | 6997 | ph1 = arrayPhase[:,pairsList[i][1]] |
|
|
6998 | 6998 | d0 = distances[pairsList[i][0]] |
|
|
6999 | 6999 | d1 = distances[pairsList[i][1]] |
|
|
7000 | 7000 | |
|
|
7001 | 7001 | ph0_aux = ph0 + ph1 |
|
|
7002 | 7002 | ph0_aux = numpy.angle(numpy.exp(1j*ph0_aux)) |
|
|
7003 | 7003 | # ph0_aux[ph0_aux > numpy.pi] -= 2*numpy.pi |
|
|
7004 | 7004 | # ph0_aux[ph0_aux < -numpy.pi] += 2*numpy.pi |
|
|
7005 | 7005 | #First Estimation |
|
|
7006 | 7006 | cosdir0[:,i] = (ph0_aux)/(2*numpy.pi*(d0 - d1)) |
|
|
7007 | 7007 | |
|
|
7008 | 7008 | #Most-Accurate Second Estimation |
|
|
7009 | 7009 | phi1_aux = ph0 - ph1 |
|
|
7010 | 7010 | phi1_aux = phi1_aux.reshape(phi1_aux.size,1) |
|
|
7011 | 7011 | #Direction Cosine 1 |
|
|
7012 | 7012 | cosdir1 = (phi1_aux + ang_aux)/(2*numpy.pi*(d0 + d1)) |
|
|
7013 | 7013 | |
|
|
7014 | 7014 | #Searching the correct Direction Cosine |
|
|
7015 | 7015 | cosdir0_aux = cosdir0[:,i] |
|
|
7016 | 7016 | cosdir0_aux = cosdir0_aux.reshape(cosdir0_aux.size,1) |
|
|
7017 | 7017 | #Minimum Distance |
|
|
7018 | 7018 | cosDiff = (cosdir1 - cosdir0_aux)**2 |
|
|
7019 | 7019 | indcos = cosDiff.argmin(axis = 1) |
|
|
7020 | 7020 | #Saving Value obtained |
|
|
7021 | 7021 | cosdir[:,i] = cosdir1[numpy.arange(len(indcos)),indcos] |
|
|
7022 | 7022 | |
|
|
7023 | 7023 | return cosdir0, cosdir |
|
|
7024 | 7024 | |
|
|
7025 | 7025 | def __calculateAOA(self, cosdir, azimuth): |
|
|
7026 | 7026 | cosdirX = cosdir[:,0] |
|
|
7027 | 7027 | cosdirY = cosdir[:,1] |
|
|
7028 | 7028 | |
|
|
7029 | 7029 | zenithAngle = numpy.arccos(numpy.sqrt(1 - cosdirX**2 - cosdirY**2))*180/numpy.pi |
|
|
7030 | 7030 | azimuthAngle = numpy.arctan2(cosdirX,cosdirY)*180/numpy.pi + azimuth#0 deg north, 90 deg east |
|
|
7031 | 7031 | angles = numpy.vstack((azimuthAngle, zenithAngle)).transpose() |
|
|
7032 | 7032 | |
|
|
7033 | 7033 | return angles |
|
|
7034 | 7034 | |
|
|
7035 | 7035 | def __getHeights(self, Ranges, zenith, error, minHeight, maxHeight): |
|
|
7036 | 7036 | |
|
|
7037 | 7037 | Ramb = 375 #Ramb = c/(2*PRF) |
|
|
7038 | 7038 | Re = 6371 #Earth Radius |
|
|
7039 | 7039 | heights = numpy.zeros(Ranges.shape) |
|
|
7040 | 7040 | |
|
|
7041 | 7041 | R_aux = numpy.array([0,1,2])*Ramb |
|
|
7042 | 7042 | R_aux = R_aux.reshape(1,R_aux.size) |
|
|
7043 | 7043 | |
|
|
7044 | 7044 | Ranges = Ranges.reshape(Ranges.size,1) |
|
|
7045 | 7045 | |
|
|
7046 | 7046 | Ri = Ranges + R_aux |
|
|
7047 | 7047 | hi = numpy.sqrt(Re**2 + Ri**2 + (2*Re*numpy.cos(zenith*numpy.pi/180)*Ri.transpose()).transpose()) - Re |
|
|
7048 | 7048 | |
|
|
7049 | 7049 | #Check if there is a height between 70 and 110 km |
|
|
7050 | 7050 | h_bool = numpy.sum(numpy.logical_and(hi > minHeight, hi < maxHeight), axis = 1) |
|
|
7051 | 7051 | ind_h = numpy.where(h_bool == 1)[0] |
|
|
7052 | 7052 | |
|
|
7053 | 7053 | hCorr = hi[ind_h, :] |
|
|
7054 | 7054 | ind_hCorr = numpy.where(numpy.logical_and(hi > minHeight, hi < maxHeight)) |
|
|
7055 | 7055 | |
|
|
7056 | 7056 | hCorr = hi[ind_hCorr][:len(ind_h)] |
|
|
7057 | 7057 | heights[ind_h] = hCorr |
|
|
7058 | 7058 | |
|
|
7059 | 7059 | #Setting Error |
|
|
7060 | 7060 | #Number 13: Height unresolvable echo: not valid height within 70 to 110 km |
|
|
7061 | 7061 | #Number 14: Height ambiguous echo: more than one possible height within 70 to 110 km |
|
|
7062 | 7062 | indError = numpy.where(numpy.logical_or(error == 13, error == 14))[0] |
|
|
7063 | 7063 | error[indError] = 0 |
|
|
7064 | 7064 | indInvalid2 = numpy.where(numpy.logical_and(h_bool > 1, error == 0))[0] |
|
|
7065 | 7065 | error[indInvalid2] = 14 |
|
|
7066 | 7066 | indInvalid1 = numpy.where(numpy.logical_and(h_bool == 0, error == 0))[0] |
|
|
7067 | 7067 | error[indInvalid1] = 13 |
|
|
7068 | 7068 | |
|
|
7069 | 7069 | return heights, error |
|
|
7070 | 7070 | |
|
|
7071 | 7071 | def getPhasePairs(self, channelPositions): |
|
|
7072 | 7072 | chanPos = numpy.array(channelPositions) |
|
|
7073 | 7073 | listOper = list(itertools.combinations(list(range(5)),2)) |
|
|
7074 | 7074 | |
|
|
7075 | 7075 | distances = numpy.zeros(4) |
|
|
7076 | 7076 | axisX = [] |
|
|
7077 | 7077 | axisY = [] |
|
|
7078 | 7078 | distX = numpy.zeros(3) |
|
|
7079 | 7079 | distY = numpy.zeros(3) |
|
|
7080 | 7080 | ix = 0 |
|
|
7081 | 7081 | iy = 0 |
|
|
7082 | 7082 | |
|
|
7083 | 7083 | pairX = numpy.zeros((2,2)) |
|
|
7084 | 7084 | pairY = numpy.zeros((2,2)) |
|
|
7085 | 7085 | |
|
|
7086 | 7086 | for i in range(len(listOper)): |
|
|
7087 | 7087 | pairi = listOper[i] |
|
|
7088 | 7088 | |
|
|
7089 | 7089 | posDif = numpy.abs(chanPos[pairi[0],:] - chanPos[pairi[1],:]) |
|
|
7090 | 7090 | |
|
|
7091 | 7091 | if posDif[0] == 0: |
|
|
7092 | 7092 | axisY.append(pairi) |
|
|
7093 | 7093 | distY[iy] = posDif[1] |
|
|
7094 | 7094 | iy += 1 |
|
|
7095 | 7095 | elif posDif[1] == 0: |
|
|
7096 | 7096 | axisX.append(pairi) |
|
|
7097 | 7097 | distX[ix] = posDif[0] |
|
|
7098 | 7098 | ix += 1 |
|
|
7099 | 7099 | |
|
|
7100 | 7100 | for i in range(2): |
|
|
7101 | 7101 | if i==0: |
|
|
7102 | 7102 | dist0 = distX |
|
|
7103 | 7103 | axis0 = axisX |
|
|
7104 | 7104 | else: |
|
|
7105 | 7105 | dist0 = distY |
|
|
7106 | 7106 | axis0 = axisY |
|
|
7107 | 7107 | |
|
|
7108 | 7108 | side = numpy.argsort(dist0)[:-1] |
|
|
7109 | 7109 | axis0 = numpy.array(axis0)[side,:] |
|
|
7110 | 7110 | chanC = int(numpy.intersect1d(axis0[0,:], axis0[1,:])[0]) |
|
|
7111 | 7111 | axis1 = numpy.unique(numpy.reshape(axis0,4)) |
|
|
7112 | 7112 | side = axis1[axis1 != chanC] |
|
|
7113 | 7113 | diff1 = chanPos[chanC,i] - chanPos[side[0],i] |
|
|
7114 | 7114 | diff2 = chanPos[chanC,i] - chanPos[side[1],i] |
|
|
7115 | 7115 | if diff1<0: |
|
|
7116 | 7116 | chan2 = side[0] |
|
|
7117 | 7117 | d2 = numpy.abs(diff1) |
|
|
7118 | 7118 | chan1 = side[1] |
|
|
7119 | 7119 | d1 = numpy.abs(diff2) |
|
|
7120 | 7120 | else: |
|
|
7121 | 7121 | chan2 = side[1] |
|
|
7122 | 7122 | d2 = numpy.abs(diff2) |
|
|
7123 | 7123 | chan1 = side[0] |
|
|
7124 | 7124 | d1 = numpy.abs(diff1) |
|
|
7125 | 7125 | |
|
|
7126 | 7126 | if i==0: |
|
|
7127 | 7127 | chanCX = chanC |
|
|
7128 | 7128 | chan1X = chan1 |
|
|
7129 | 7129 | chan2X = chan2 |
|
|
7130 | 7130 | distances[0:2] = numpy.array([d1,d2]) |
|
|
7131 | 7131 | else: |
|
|
7132 | 7132 | chanCY = chanC |
|
|
7133 | 7133 | chan1Y = chan1 |
|
|
7134 | 7134 | chan2Y = chan2 |
|
|
7135 | 7135 | distances[2:4] = numpy.array([d1,d2]) |
|
|
7136 | 7136 | # axisXsides = numpy.reshape(axisX[ix,:],4) |
|
|
7137 | 7137 | # |
|
|
7138 | 7138 | # channelCentX = int(numpy.intersect1d(pairX[0,:], pairX[1,:])[0]) |
|
|
7139 | 7139 | # channelCentY = int(numpy.intersect1d(pairY[0,:], pairY[1,:])[0]) |
|
|
7140 | 7140 | # |
|
|
7141 | 7141 | # ind25X = numpy.where(pairX[0,:] != channelCentX)[0][0] |
|
|
7142 | 7142 | # ind20X = numpy.where(pairX[1,:] != channelCentX)[0][0] |
|
|
7143 | 7143 | # channel25X = int(pairX[0,ind25X]) |
|
|
7144 | 7144 | # channel20X = int(pairX[1,ind20X]) |
|
|
7145 | 7145 | # ind25Y = numpy.where(pairY[0,:] != channelCentY)[0][0] |
|
|
7146 | 7146 | # ind20Y = numpy.where(pairY[1,:] != channelCentY)[0][0] |
|
|
7147 | 7147 | # channel25Y = int(pairY[0,ind25Y]) |
|
|
7148 | 7148 | # channel20Y = int(pairY[1,ind20Y]) |
|
|
7149 | 7149 | |
|
|
7150 | 7150 | # pairslist = [(channelCentX, channel25X),(channelCentX, channel20X),(channelCentY,channel25Y),(channelCentY, channel20Y)] |
|
|
7151 | 7151 | pairslist = [(chanCX, chan1X),(chanCX, chan2X),(chanCY,chan1Y),(chanCY, chan2Y)] |
|
|
7152 | 7152 | |
|
|
7153 | 7153 | return pairslist, distances |
|
|
7154 | 7154 | # def __getAOA(self, phases, pairsList, error, AOAthresh, azimuth): |
|
|
7155 | 7155 | # |
|
|
7156 | 7156 | # arrayAOA = numpy.zeros((phases.shape[0],3)) |
|
|
7157 | 7157 | # cosdir0, cosdir = self.__getDirectionCosines(phases, pairsList) |
|
|
7158 | 7158 | # |
|
|
7159 | 7159 | # arrayAOA[:,:2] = self.__calculateAOA(cosdir, azimuth) |
|
|
7160 | 7160 | # cosDirError = numpy.sum(numpy.abs(cosdir0 - cosdir), axis = 1) |
|
|
7161 | 7161 | # arrayAOA[:,2] = cosDirError |
|
|
7162 | 7162 | # |
|
|
7163 | 7163 | # azimuthAngle = arrayAOA[:,0] |
|
|
7164 | 7164 | # zenithAngle = arrayAOA[:,1] |
|
|
7165 | 7165 | # |
|
|
7166 | 7166 | # #Setting Error |
|
|
7167 | 7167 | # #Number 3: AOA not fesible |
|
|
7168 | 7168 | # indInvalid = numpy.where(numpy.logical_and((numpy.logical_or(numpy.isnan(zenithAngle), numpy.isnan(azimuthAngle))),error == 0))[0] |
|
|
7169 | 7169 | # error[indInvalid] = 3 |
|
|
7170 | 7170 | # #Number 4: Large difference in AOAs obtained from different antenna baselines |
|
|
7171 | 7171 | # indInvalid = numpy.where(numpy.logical_and(cosDirError > AOAthresh,error == 0))[0] |
|
|
7172 | 7172 | # error[indInvalid] = 4 |
|
|
7173 | 7173 | # return arrayAOA, error |
|
|
7174 | 7174 | # |
|
|
7175 | 7175 | # def __getDirectionCosines(self, arrayPhase, pairsList): |
|
|
7176 | 7176 | # |
|
|
7177 | 7177 | # #Initializing some variables |
|
|
7178 | 7178 | # ang_aux = numpy.array([-8,-7,-6,-5,-4,-3,-2,-1,0,1,2,3,4,5,6,7,8])*2*numpy.pi |
|
|
7179 | 7179 | # ang_aux = ang_aux.reshape(1,ang_aux.size) |
|
|
7180 | 7180 | # |
|
|
7181 | 7181 | # cosdir = numpy.zeros((arrayPhase.shape[0],2)) |
|
|
7182 | 7182 | # cosdir0 = numpy.zeros((arrayPhase.shape[0],2)) |
|
|
7183 | 7183 | # |
|
|
7184 | 7184 | # |
|
|
7185 | 7185 | # for i in range(2): |
|
|
7186 | 7186 | # #First Estimation |
|
|
7187 | 7187 | # phi0_aux = arrayPhase[:,pairsList[i][0]] + arrayPhase[:,pairsList[i][1]] |
|
|
7188 | 7188 | # #Dealias |
|
|
7189 | 7189 | # indcsi = numpy.where(phi0_aux > numpy.pi) |
|
|
7190 | 7190 | # phi0_aux[indcsi] -= 2*numpy.pi |
|
|
7191 | 7191 | # indcsi = numpy.where(phi0_aux < -numpy.pi) |
|
|
7192 | 7192 | # phi0_aux[indcsi] += 2*numpy.pi |
|
|
7193 | 7193 | # #Direction Cosine 0 |
|
|
7194 | 7194 | # cosdir0[:,i] = -(phi0_aux)/(2*numpy.pi*0.5) |
|
|
7195 | 7195 | # |
|
|
7196 | 7196 | # #Most-Accurate Second Estimation |
|
|
7197 | 7197 | # phi1_aux = arrayPhase[:,pairsList[i][0]] - arrayPhase[:,pairsList[i][1]] |
|
|
7198 | 7198 | # phi1_aux = phi1_aux.reshape(phi1_aux.size,1) |
|
|
7199 | 7199 | # #Direction Cosine 1 |
|
|
7200 | 7200 | # cosdir1 = -(phi1_aux + ang_aux)/(2*numpy.pi*4.5) |
|
|
7201 | 7201 | # |
|
|
7202 | 7202 | # #Searching the correct Direction Cosine |
|
|
7203 | 7203 | # cosdir0_aux = cosdir0[:,i] |
|
|
7204 | 7204 | # cosdir0_aux = cosdir0_aux.reshape(cosdir0_aux.size,1) |
|
|
7205 | 7205 | # #Minimum Distance |
|
|
7206 | 7206 | # cosDiff = (cosdir1 - cosdir0_aux)**2 |
|
|
7207 | 7207 | # indcos = cosDiff.argmin(axis = 1) |
|
|
7208 | 7208 | # #Saving Value obtained |
|
|
7209 | 7209 | # cosdir[:,i] = cosdir1[numpy.arange(len(indcos)),indcos] |
|
|
7210 | 7210 | # |
|
|
7211 | 7211 | # return cosdir0, cosdir |
|
|
7212 | 7212 | # |
|
|
7213 | 7213 | # def __calculateAOA(self, cosdir, azimuth): |
|
|
7214 | 7214 | # cosdirX = cosdir[:,0] |
|
|
7215 | 7215 | # cosdirY = cosdir[:,1] |
|
|
7216 | 7216 | # |
|
|
7217 | 7217 | # zenithAngle = numpy.arccos(numpy.sqrt(1 - cosdirX**2 - cosdirY**2))*180/numpy.pi |
|
|
7218 | 7218 | # azimuthAngle = numpy.arctan2(cosdirX,cosdirY)*180/numpy.pi + azimuth #0 deg north, 90 deg east |
|
|
7219 | 7219 | # angles = numpy.vstack((azimuthAngle, zenithAngle)).transpose() |
|
|
7220 | 7220 | # |
|
|
7221 | 7221 | # return angles |
|
|
7222 | 7222 | # |
|
|
7223 | 7223 | # def __getHeights(self, Ranges, zenith, error, minHeight, maxHeight): |
|
|
7224 | 7224 | # |
|
|
7225 | 7225 | # Ramb = 375 #Ramb = c/(2*PRF) |
|
|
7226 | 7226 | # Re = 6371 #Earth Radius |
|
|
7227 | 7227 | # heights = numpy.zeros(Ranges.shape) |
|
|
7228 | 7228 | # |
|
|
7229 | 7229 | # R_aux = numpy.array([0,1,2])*Ramb |
|
|
7230 | 7230 | # R_aux = R_aux.reshape(1,R_aux.size) |
|
|
7231 | 7231 | # |
|
|
7232 | 7232 | # Ranges = Ranges.reshape(Ranges.size,1) |
|
|
7233 | 7233 | # |
|
|
7234 | 7234 | # Ri = Ranges + R_aux |
|
|
7235 | 7235 | # hi = numpy.sqrt(Re**2 + Ri**2 + (2*Re*numpy.cos(zenith*numpy.pi/180)*Ri.transpose()).transpose()) - Re |
|
|
7236 | 7236 | # |
|
|
7237 | 7237 | # #Check if there is a height between 70 and 110 km |
|
|
7238 | 7238 | # h_bool = numpy.sum(numpy.logical_and(hi > minHeight, hi < maxHeight), axis = 1) |
|
|
7239 | 7239 | # ind_h = numpy.where(h_bool == 1)[0] |
|
|
7240 | 7240 | # |
|
|
7241 | 7241 | # hCorr = hi[ind_h, :] |
|
|
7242 | 7242 | # ind_hCorr = numpy.where(numpy.logical_and(hi > minHeight, hi < maxHeight)) |
|
|
7243 | 7243 | # |
|
|
7244 | 7244 | # hCorr = hi[ind_hCorr] |
|
|
7245 | 7245 | # heights[ind_h] = hCorr |
|
|
7246 | 7246 | # |
|
|
7247 | 7247 | # #Setting Error |
|
|
7248 | 7248 | # #Number 13: Height unresolvable echo: not valid height within 70 to 110 km |
|
|
7249 | 7249 | # #Number 14: Height ambiguous echo: more than one possible height within 70 to 110 km |
|
|
7250 | 7250 | # |
|
|
7251 | 7251 | # indInvalid2 = numpy.where(numpy.logical_and(h_bool > 1, error == 0))[0] |
|
|
7252 | 7252 | # error[indInvalid2] = 14 |
|
|
7253 | 7253 | # indInvalid1 = numpy.where(numpy.logical_and(h_bool == 0, error == 0))[0] |
|
|
7254 | 7254 | # error[indInvalid1] = 13 |
|
|
7255 | 7255 | # |
|
|
7256 | 7256 | # return heights, error |
|
|
7257 | 7257 | |
|
|
7258 | 7258 | |
|
|
7259 | 7259 | |
|
|
7260 | 7260 | class IGRFModel(Operation): |
|
|
7261 | 7261 | ''' |
|
|
7262 | 7262 | Written by R. Flores |
|
|
7263 | 7263 | ''' |
|
|
7264 | 7264 | """Operation to calculate Geomagnetic parameters. |
|
|
7265 | 7265 | |
|
|
7266 | 7266 | Parameters: |
|
|
7267 | 7267 | ----------- |
|
|
7268 | 7268 | None |
|
|
7269 | 7269 | |
|
|
7270 | 7270 | Example |
|
|
7271 | 7271 | -------- |
|
|
7272 | 7272 | |
|
|
7273 | 7273 | op = proc_unit.addOperation(name='IGRFModel', optype='other') |
|
|
7274 | 7274 | |
|
|
7275 | 7275 | """ |
|
|
7276 | 7276 | |
|
|
7277 | 7277 | def __init__(self, **kwargs): |
|
|
7278 | 7278 | |
|
|
7279 | 7279 | Operation.__init__(self, **kwargs) |
|
|
7280 | 7280 | |
|
|
7281 | 7281 | self.aux=1 |
|
|
7282 | 7282 | |
|
|
7283 | 7283 | def run(self,dataOut): |
|
|
7284 | 7284 | |
|
|
7285 | 7285 | try: |
|
|
7286 | 7286 | from schainpy.model.proc import mkfact_short_2020_2 |
|
|
7287 | 7287 | except: |
|
|
7288 | 7288 | log.warning('You should install "mkfact_short_2020" module to process IGRF Model') |
|
|
7289 | 7289 | |
|
|
7290 | 7290 | if self.aux==1: |
|
|
7291 | 7291 | |
|
|
7292 | 7292 | #dataOut.TimeBlockSeconds_First_Time=time.mktime(time.strptime(dataOut.TimeBlockDate)) |
|
|
7293 | 7293 | #### we do not use dataOut.datatime.ctime() because it's the time of the second (next) block |
|
|
7294 | 7294 | dataOut.TimeBlockSeconds_First_Time=dataOut.TimeBlockSeconds |
|
|
7295 | 7295 | dataOut.bd_time=time.gmtime(dataOut.TimeBlockSeconds_First_Time) |
|
|
7296 | 7296 | dataOut.year=dataOut.bd_time.tm_year+(dataOut.bd_time.tm_yday-1)/364.0 |
|
|
7297 | 7297 | dataOut.ut=dataOut.bd_time.tm_hour+dataOut.bd_time.tm_min/60.0+dataOut.bd_time.tm_sec/3600.0 |
|
|
7298 | 7298 | |
|
|
7299 | 7299 | self.aux=0 |
|
|
7300 | 7300 | dh = dataOut.heightList[1]-dataOut.heightList[0] |
|
|
7301 | 7301 | dataOut.h=numpy.arange(0.0,dh*dataOut.MAXNRANGENDT,dh,dtype='float32') |
|
|
7302 | 7302 | dataOut.bfm=numpy.zeros(dataOut.MAXNRANGENDT,dtype='float32') |
|
|
7303 | 7303 | dataOut.bfm=numpy.array(dataOut.bfm,order='F') |
|
|
7304 | 7304 | dataOut.thb=numpy.zeros(dataOut.MAXNRANGENDT,dtype='float32') |
|
|
7305 | 7305 | dataOut.thb=numpy.array(dataOut.thb,order='F') |
|
|
7306 | 7306 | dataOut.bki=numpy.zeros(dataOut.MAXNRANGENDT,dtype='float32') |
|
|
7307 | 7307 | dataOut.bki=numpy.array(dataOut.bki,order='F') |
|
|
7308 | 7308 | mkfact_short_2020_2.mkfact(dataOut.year,dataOut.h,dataOut.bfm,dataOut.thb,dataOut.bki,dataOut.MAXNRANGENDT) |
|
|
7309 | 7309 | |
|
|
7310 | 7310 | |
|
|
7311 | 7311 | return dataOut |
|
|
7312 | 7312 | |
|
|
7313 | 7313 | class MergeProc(ProcessingUnit): |
|
|
7314 | 7314 | |
|
|
7315 | 7315 | def __init__(self): |
|
|
7316 | 7316 | ProcessingUnit.__init__(self) |
|
|
7317 | 7317 | |
|
|
7318 | 7318 | def run(self, attr_data, attr_data_2 = None, attr_data_3 = None, attr_data_4 = None, attr_data_5 = None, mode=0): |
|
|
7319 | 7319 | #print("*****************************Merge***************") |
|
|
7320 | 7320 | |
|
|
7321 | 7321 | self.dataOut = getattr(self, self.inputs[0]) |
|
|
7322 | 7322 | data_inputs = [getattr(self, attr) for attr in self.inputs] |
|
|
7323 | 7323 | #print(data_inputs) |
|
|
7324 | 7324 | #print("Run: ",self.dataOut.runNextUnit) |
|
|
7325 | 7325 | #exit(1) |
|
|
7326 | 7326 | #print(self.dataOut.nHeights) |
|
|
7327 | 7327 | #exit(1) |
|
|
7328 | 7328 | #print("a:", [getattr(data, attr_data) for data in data_inputs][1]) |
|
|
7329 | 7329 | #exit(1) |
|
|
7330 | 7330 | if mode==0: |
|
|
7331 | 7331 | data = numpy.concatenate([getattr(data, attr_data) for data in data_inputs]) |
|
|
7332 | 7332 | setattr(self.dataOut, attr_data, data) |
|
|
7333 | 7333 | |
|
|
7334 | 7334 | if mode==1: #Hybrid |
|
|
7335 | 7335 | #data = numpy.concatenate([getattr(data, attr_data) for data in data_inputs],axis=1) |
|
|
7336 | 7336 | #setattr(self.dataOut, attr_data, data) |
|
|
7337 | 7337 | setattr(self.dataOut, 'dataLag_spc', [getattr(data, attr_data) for data in data_inputs][0]) |
|
|
7338 | 7338 | setattr(self.dataOut, 'dataLag_spc_LP', [getattr(data, attr_data) for data in data_inputs][1]) |
|
|
7339 | 7339 | setattr(self.dataOut, 'dataLag_cspc', [getattr(data, attr_data_2) for data in data_inputs][0]) |
|
|
7340 | 7340 | setattr(self.dataOut, 'dataLag_cspc_LP', [getattr(data, attr_data_2) for data in data_inputs][1]) |
|
|
7341 | 7341 | #setattr(self.dataOut, 'nIncohInt', [getattr(data, attr_data_3) for data in data_inputs][0]) |
|
|
7342 | 7342 | #setattr(self.dataOut, 'nIncohInt_LP', [getattr(data, attr_data_3) for data in data_inputs][1]) |
|
|
7343 | 7343 | ''' |
|
|
7344 | 7344 | print(self.dataOut.dataLag_spc_LP.shape) |
|
|
7345 | 7345 | print(self.dataOut.dataLag_cspc_LP.shape) |
|
|
7346 | 7346 | exit(1) |
|
|
7347 | 7347 | ''' |
|
|
7348 | 7348 | |
|
|
7349 | 7349 | #self.dataOut.dataLag_spc_LP = numpy.transpose(self.dataOut.dataLag_spc_LP[0],(2,0,1)) |
|
|
7350 | 7350 | #self.dataOut.dataLag_cspc_LP = numpy.transpose(self.dataOut.dataLag_cspc_LP,(3,1,2,0)) |
|
|
7351 | 7351 | ''' |
|
|
7352 | 7352 | print("Merge") |
|
|
7353 | 7353 | print(numpy.shape(self.dataOut.dataLag_spc)) |
|
|
7354 | 7354 | print(numpy.shape(self.dataOut.dataLag_spc_LP)) |
|
|
7355 | 7355 | print(numpy.shape(self.dataOut.dataLag_cspc)) |
|
|
7356 | 7356 | print(numpy.shape(self.dataOut.dataLag_cspc_LP)) |
|
|
7357 | 7357 | exit(1) |
|
|
7358 | 7358 | ''' |
|
|
7359 | 7359 | #print(numpy.sum(self.dataOut.dataLag_spc_LP[2,:,164])/128) |
|
|
7360 | 7360 | #print(numpy.sum(self.dataOut.dataLag_cspc_LP[0,:,30,1])/128) |
|
|
7361 | 7361 | #exit(1) |
|
|
7362 | 7362 | #print(self.dataOut.NDP) |
|
|
7363 | 7363 | #print(self.dataOut.nNoiseProfiles) |
|
|
7364 | 7364 | |
|
|
7365 | 7365 | #self.dataOut.nIncohInt_LP = 128 |
|
|
7366 | 7366 | self.dataOut.nProfiles_LP = 128#self.dataOut.nIncohInt_LP |
|
|
7367 | 7367 | self.dataOut.nIncohInt_LP = self.dataOut.nIncohInt |
|
|
7368 | 7368 | self.dataOut.NLAG = 16 |
|
|
7369 | 7369 | self.dataOut.NRANGE = 200 |
|
|
7370 | 7370 | self.dataOut.NSCAN = 128 |
|
|
7371 | 7371 | #print(numpy.shape(self.dataOut.data_spc)) |
|
|
7372 | 7372 | |
|
|
7373 | 7373 | #exit(1) |
|
|
7374 | 7374 | |
|
|
7375 | 7375 | if mode==2: #HAE 2022 |
|
|
7376 | 7376 | data = numpy.sum([getattr(data, attr_data) for data in data_inputs],axis=0) |
|
|
7377 | 7377 | setattr(self.dataOut, attr_data, data) |
|
|
7378 | 7378 | |
|
|
7379 | 7379 | self.dataOut.nIncohInt *= 2 |
|
|
7380 | 7380 | #meta = self.dataOut.getFreqRange(1)/1000. |
|
|
7381 | 7381 | self.dataOut.freqRange = self.dataOut.getFreqRange(1)/1000. |
|
|
7382 | 7382 | |
|
|
7383 | 7383 | #exit(1) |
|
|
7384 | 7384 | |
|
|
7385 | 7385 | if mode==4: #Hybrid LP-SSheightProfiles |
|
|
7386 | 7386 | #data = numpy.concatenate([getattr(data, attr_data) for data in data_inputs],axis=1) |
|
|
7387 | 7387 | #setattr(self.dataOut, attr_data, data) |
|
|
7388 | 7388 | setattr(self.dataOut, 'dataLag_spc', getattr(data_inputs[0], attr_data)) #DP |
|
|
7389 | 7389 | setattr(self.dataOut, 'dataLag_cspc', getattr(data_inputs[0], attr_data_2)) #DP |
|
|
7390 | 7390 | setattr(self.dataOut, 'dataLag_spc_LP', getattr(data_inputs[1], attr_data_3)) #LP |
|
|
7391 | 7391 | #setattr(self.dataOut, 'dataLag_cspc_LP', getattr(data_inputs[1], attr_data_4)) #LP |
|
|
7392 | 7392 | #setattr(self.dataOut, 'data_acf', getattr(data_inputs[1], attr_data_5)) #LP |
|
|
7393 | 7393 | setattr(self.dataOut, 'data_acf', getattr(data_inputs[1], attr_data_5)) #LP |
|
|
7394 | 7394 | #print("Merge data_acf: ",self.dataOut.data_acf.shape) |
|
|
7395 | 7395 | |
|
|
7396 | 7396 | |
|
|
7397 | 7397 | #self.dataOut.nIncohInt_LP = 128 |
|
|
7398 | 7398 | #self.dataOut.nProfiles_LP = 128#self.dataOut.nIncohInt_LP |
|
|
7399 | 7399 | self.dataOut.nProfiles_LP = 16#28#self.dataOut.nIncohInt_LP |
|
|
7400 | 7400 | self.dataOut.nProfiles_LP = self.dataOut.data_acf.shape[1]#28#self.dataOut.nIncohInt_LP |
|
|
7401 | 7401 | self.dataOut.NSCAN = 128 |
|
|
7402 | 7402 | self.dataOut.nIncohInt_LP = self.dataOut.nIncohInt*self.dataOut.NSCAN |
|
|
7403 | 7403 | #print("sahpi",self.dataOut.nIncohInt_LP) |
|
|
7404 | 7404 | #exit(1) |
|
|
7405 | 7405 | self.dataOut.NLAG = 16 |
|
|
7406 | 7406 | self.dataOut.NLAG = self.dataOut.data_acf.shape[1] |
|
|
7407 | 7407 | self.dataOut.NRANGE = self.dataOut.data_acf.shape[-1] |
|
|
7408 | 7408 | |
|
|
7409 | 7409 | #print(numpy.shape(self.dataOut.data_spc)) |
|
|
7410 | 7410 | |
|
|
7411 | 7411 | #exit(1) |
|
|
7412 | 7412 | if mode==5: |
|
|
7413 | 7413 | data = numpy.concatenate([getattr(data, attr_data) for data in data_inputs]) |
|
|
7414 | 7414 | setattr(self.dataOut, attr_data, data) |
|
|
7415 | 7415 | data = numpy.concatenate([getattr(data, attr_data_2) for data in data_inputs]) |
|
|
7416 | 7416 | setattr(self.dataOut, attr_data_2, data) |
|
|
7417 | 7417 | |
|
|
7418 | 7418 | if mode==6: #Hybrid Spectra-Voltage |
|
|
7419 | 7419 | #data = numpy.concatenate([getattr(data, attr_data) for data in data_inputs],axis=1) |
|
|
7420 | 7420 | #setattr(self.dataOut, attr_data, data) |
|
|
7421 | 7421 | setattr(self.dataOut, 'dataLag_spc', getattr(data_inputs[1], attr_data)) #DP |
|
|
7422 | 7422 | setattr(self.dataOut, 'dataLag_cspc', getattr(data_inputs[1], attr_data_2)) #DP |
|
|
7423 | 7423 | setattr(self.dataOut, 'output_LP_integrated', getattr(data_inputs[0], attr_data_3)) #LP |
|
|
7424 | 7424 | #setattr(self.dataOut, 'dataLag_cspc_LP', getattr(data_inputs[1], attr_data_4)) #LP |
|
|
7425 | 7425 | #setattr(self.dataOut, 'data_acf', getattr(data_inputs[1], attr_data_5)) #LP |
|
|
7426 | 7426 | #setattr(self.dataOut, 'data_acf', getattr(data_inputs[1], attr_data_5)) #LP |
|
|
7427 | 7427 | #print("Merge data_acf: ",self.dataOut.data_acf.shape) |
|
|
7428 | 7428 | #print(self.dataOut.NSCAN) |
|
|
7429 | 7429 | self.dataOut.nIncohInt = int(self.dataOut.NAVG * self.dataOut.nint) |
|
|
7430 | 7430 | #print(self.dataOut.dataLag_spc.shape) |
|
|
7431 | 7431 | self.dataOut.nProfiles = self.dataOut.nProfiles_DP = self.dataOut.dataLag_spc.shape[1] |
|
|
7432 | 7432 | ''' |
|
|
7433 | 7433 | #self.dataOut.nIncohInt_LP = 128 |
|
|
7434 | 7434 | #self.dataOut.nProfiles_LP = 128#self.dataOut.nIncohInt_LP |
|
|
7435 | 7435 | self.dataOut.nProfiles_LP = 16#28#self.dataOut.nIncohInt_LP |
|
|
7436 | 7436 | self.dataOut.nProfiles_LP = self.dataOut.data_acf.shape[1]#28#self.dataOut.nIncohInt_LP |
|
|
7437 | 7437 | self.dataOut.NSCAN = 128 |
|
|
7438 | 7438 | self.dataOut.nIncohInt_LP = self.dataOut.nIncohInt*self.dataOut.NSCAN |
|
|
7439 | 7439 | #print("sahpi",self.dataOut.nIncohInt_LP) |
|
|
7440 | 7440 | #exit(1) |
|
|
7441 | 7441 | self.dataOut.NLAG = 16 |
|
|
7442 | 7442 | self.dataOut.NLAG = self.dataOut.data_acf.shape[1] |
|
|
7443 | 7443 | self.dataOut.NRANGE = self.dataOut.data_acf.shape[-1] |
|
|
7444 | 7444 | ''' |
|
|
7445 | 7445 | #print(numpy.shape(self.dataOut.data_spc)) |
|
|
7446 | 7446 | #print("*************************GOOD*************************") |
|
|
7447 | 7447 | #exit(1) |
|
|
7448 | 7448 | |
|
|
7449 | 7449 | if mode==11: #MST ISR |
|
|
7450 | 7450 | #data = numpy.concatenate([getattr(data, attr_data) for data in data_inputs],axis=1) |
|
|
7451 | 7451 | #setattr(self.dataOut, attr_data, data) |
|
|
7452 | 7452 | #setattr(self.dataOut, 'ph2', [getattr(data, attr_data) for data in data_inputs][1]) |
|
|
7453 | 7453 | #setattr(self.dataOut, 'dphi', [getattr(data, attr_data_2) for data in data_inputs][1]) |
|
|
7454 | 7454 | #setattr(self.dataOut, 'sdp2', [getattr(data, attr_data_3) for data in data_inputs][1]) |
|
|
7455 | 7455 | |
|
|
7456 | 7456 | setattr(self.dataOut, 'ph2', getattr(data_inputs[1], attr_data)) #DP |
|
|
7457 | 7457 | setattr(self.dataOut, 'dphi', getattr(data_inputs[1], attr_data_2)) #DP |
|
|
7458 | 7458 | setattr(self.dataOut, 'sdp2', getattr(data_inputs[1], attr_data_3)) #DP |
|
|
7459 | 7459 | |
|
|
7460 | 7460 | print("MST Density", numpy.shape(self.dataOut.ph2)) |
|
|
7461 | 7461 | print("cf MST: ", self.dataOut.cf) |
|
|
7462 | 7462 | #exit(1) |
|
|
7463 | 7463 | #print("MST Density", self.dataOut.ph2[116:283]) |
|
|
7464 | 7464 | print("MST Density", self.dataOut.ph2[80:120]) |
|
|
7465 | 7465 | print("MST dPhi", self.dataOut.dphi[80:120]) |
|
|
7466 | 7466 | self.dataOut.ph2 *= self.dataOut.cf#0.0008136899 |
|
|
7467 | 7467 | #print("MST Density", self.dataOut.ph2[116:283]) |
|
|
7468 | 7468 | self.dataOut.sdp2 *= 0#self.dataOut.cf#0.0008136899 |
|
|
7469 | 7469 | #print("MST Density", self.dataOut.ph2[116:283]) |
|
|
7470 | 7470 | print("MST Density", self.dataOut.ph2[80:120]) |
|
|
7471 | 7471 | self.dataOut.NSHTS = int(numpy.shape(self.dataOut.ph2)[0]) |
|
|
7472 | 7472 | dH = self.dataOut.heightList[1]-self.dataOut.heightList[0] |
|
|
7473 | 7473 | dH /= self.dataOut.windowOfFilter |
|
|
7474 | 7474 | self.dataOut.heightList = numpy.arange(0,self.dataOut.NSHTS)*dH + dH |
|
|
7475 | 7475 | #print("heightList: ", self.dataOut.heightList) |
|
|
7476 | 7476 | self.dataOut.NDP = self.dataOut.NSHTS |
|
|
7477 | 7477 | #exit(1) |
|
|
7478 | 7478 | #print(self.dataOut.heightList) |
|
|
7479 | 7479 | |
|
|
7480 | 7480 | class MST_Den_Conv(Operation): |
|
|
7481 | 7481 | ''' |
|
|
7482 | 7482 | Written by R. Flores |
|
|
7483 | 7483 | ''' |
|
|
7484 | 7484 | """Operation to calculate Geomagnetic parameters. |
|
|
7485 | 7485 | |
|
|
7486 | 7486 | Parameters: |
|
|
7487 | 7487 | ----------- |
|
|
7488 | 7488 | None |
|
|
7489 | 7489 | |
|
|
7490 | 7490 | Example |
|
|
7491 | 7491 | -------- |
|
|
7492 | 7492 | |
|
|
7493 | 7493 | op = proc_unit.addOperation(name='MST_Den_Conv', optype='other') |
|
|
7494 | 7494 | |
|
|
7495 | 7495 | """ |
|
|
7496 | 7496 | |
|
|
7497 | 7497 | def __init__(self, **kwargs): |
|
|
7498 | 7498 | |
|
|
7499 | 7499 | Operation.__init__(self, **kwargs) |
|
|
7500 | 7500 | |
|
|
7501 | 7501 | def run(self,dataOut): |
|
|
7502 | 7502 | |
|
|
7503 | 7503 | dataOut.PowDen = numpy.zeros((1,dataOut.NDP)) |
|
|
7504 | 7504 | dataOut.PowDen[0] = numpy.copy(dataOut.ph2[:dataOut.NDP]) |
|
|
7505 | 7505 | |
|
|
7506 | 7506 | dataOut.FarDen = numpy.zeros((1,dataOut.NDP)) |
|
|
7507 | 7507 | dataOut.FarDen[0] = numpy.copy(dataOut.dphi[:dataOut.NDP]) |
|
|
7508 | 7508 | print("pow den shape", numpy.shape(dataOut.PowDen)) |
|
|
7509 | 7509 | print("far den shape", numpy.shape(dataOut.FarDen)) |
|
|
7510 | 7510 | return dataOut |
General Comments 0
You need to be logged in to leave comments.
Login now