##// END OF EJS Templates
schain
RSS
Add deflip to correct Julia data
jespinoza -
r1415:4d2e5108a3e8
parent child
Show More
Side by Side Unified
Context file:
r1415:4d2e5108a3e8 Loading diff...:
Show file before | Show file after | Hide comments
@@ -1,898 +1,922
1 1 # Copyright (c) 2012-2020 Jicamarca Radio Observatory
2 2 # All rights reserved.
3 3 #
4 4 # Distributed under the terms of the BSD 3-clause license.
5 5 """Spectra processing Unit and operations
6 6
7 7 Here you will find the processing unit `SpectraProc` and several operations
8 8 to work with Spectra data type
9 9 """
10 10
11 11 import time
12 12 import itertools
13 13
14 14 import numpy
15 15
16 16 from schainpy.model.proc.jroproc_base import ProcessingUnit, MPDecorator, Operation
17 17 from schainpy.model.data.jrodata import Spectra
18 18 from schainpy.model.data.jrodata import hildebrand_sekhon
19 19 from schainpy.utils import log
20 20
21 21
22 22 class SpectraProc(ProcessingUnit):
23 23
24 24 def __init__(self):
25 25
26 26 ProcessingUnit.__init__(self)
27 27
28 28 self.buffer = None
29 29 self.firstdatatime = None
30 30 self.profIndex = 0
31 31 self.dataOut = Spectra()
32 32 self.id_min = None
33 33 self.id_max = None
34 34 self.setupReq = False # Agregar a todas las unidades de proc
35 35
36 36 def __updateSpecFromVoltage(self):
37 37
38 38 self.dataOut.timeZone = self.dataIn.timeZone
39 39 self.dataOut.dstFlag = self.dataIn.dstFlag
40 40 self.dataOut.errorCount = self.dataIn.errorCount
41 41 self.dataOut.useLocalTime = self.dataIn.useLocalTime
42 42 try:
43 43 self.dataOut.processingHeaderObj = self.dataIn.processingHeaderObj.copy()
44 44 except:
45 45 pass
46 46 self.dataOut.radarControllerHeaderObj = self.dataIn.radarControllerHeaderObj.copy()
47 47 self.dataOut.systemHeaderObj = self.dataIn.systemHeaderObj.copy()
48 48 self.dataOut.channelList = self.dataIn.channelList
49 49 self.dataOut.heightList = self.dataIn.heightList
50 50 self.dataOut.dtype = numpy.dtype([('real', '<f4'), ('imag', '<f4')])
51 51 self.dataOut.nProfiles = self.dataOut.nFFTPoints
52 52 self.dataOut.flagDiscontinuousBlock = self.dataIn.flagDiscontinuousBlock
53 53 self.dataOut.utctime = self.firstdatatime
54 54 self.dataOut.flagDecodeData = self.dataIn.flagDecodeData
55 55 self.dataOut.flagDeflipData = self.dataIn.flagDeflipData
56 56 self.dataOut.flagShiftFFT = False
57 57 self.dataOut.nCohInt = self.dataIn.nCohInt
58 58 self.dataOut.nIncohInt = 1
59 59 self.dataOut.windowOfFilter = self.dataIn.windowOfFilter
60 60 self.dataOut.frequency = self.dataIn.frequency
61 61 self.dataOut.realtime = self.dataIn.realtime
62 62 self.dataOut.azimuth = self.dataIn.azimuth
63 63 self.dataOut.zenith = self.dataIn.zenith
64 64 self.dataOut.beam.codeList = self.dataIn.beam.codeList
65 65 self.dataOut.beam.azimuthList = self.dataIn.beam.azimuthList
66 66 self.dataOut.beam.zenithList = self.dataIn.beam.zenithList
67 67
68 68 def __getFft(self):
69 69 """
70 70 Convierte valores de Voltaje a Spectra
71 71
72 72 Affected:
73 73 self.dataOut.data_spc
74 74 self.dataOut.data_cspc
75 75 self.dataOut.data_dc
76 76 self.dataOut.heightList
77 77 self.profIndex
78 78 self.buffer
79 79 self.dataOut.flagNoData
80 80 """
81 81 fft_volt = numpy.fft.fft(
82 82 self.buffer, n=self.dataOut.nFFTPoints, axis=1)
83 83 fft_volt = fft_volt.astype(numpy.dtype('complex'))
84 84 dc = fft_volt[:, 0, :]
85 85
86 86 # calculo de self-spectra
87 87 fft_volt = numpy.fft.fftshift(fft_volt, axes=(1,))
88 88 spc = fft_volt * numpy.conjugate(fft_volt)
89 89 spc = spc.real
90 90
91 91 blocksize = 0
92 92 blocksize += dc.size
93 93 blocksize += spc.size
94 94
95 95 cspc = None
96 96 pairIndex = 0
97 97 if self.dataOut.pairsList != None:
98 98 # calculo de cross-spectra
99 99 cspc = numpy.zeros(
100 100 (self.dataOut.nPairs, self.dataOut.nFFTPoints, self.dataOut.nHeights), dtype='complex')
101 101 for pair in self.dataOut.pairsList:
102 102 if pair[0] not in self.dataOut.channelList:
103 103 raise ValueError("Error getting CrossSpectra: pair 0 of %s is not in channelList = %s" % (
104 104 str(pair), str(self.dataOut.channelList)))
105 105 if pair[1] not in self.dataOut.channelList:
106 106 raise ValueError("Error getting CrossSpectra: pair 1 of %s is not in channelList = %s" % (
107 107 str(pair), str(self.dataOut.channelList)))
108 108
109 109 cspc[pairIndex, :, :] = fft_volt[pair[0], :, :] * \
110 110 numpy.conjugate(fft_volt[pair[1], :, :])
111 111 pairIndex += 1
112 112 blocksize += cspc.size
113 113
114 114 self.dataOut.data_spc = spc
115 115 self.dataOut.data_cspc = cspc
116 116 self.dataOut.data_dc = dc
117 117 self.dataOut.blockSize = blocksize
118 118 self.dataOut.flagShiftFFT = False
119 119
120 120 def run(self, nProfiles=None, nFFTPoints=None, pairsList=None, ippFactor=None, shift_fft=False):
121 121
122 122 if self.dataIn.type == "Spectra":
123 123 self.dataOut.copy(self.dataIn)
124 124 if shift_fft:
125 125 # desplaza a la derecha en el eje 2 determinadas posiciones
126 126 shift = int(self.dataOut.nFFTPoints / 2)
127 127 self.dataOut.data_spc = numpy.roll(self.dataOut.data_spc, shift , axis=1)
128 128
129 129 if self.dataOut.data_cspc is not None:
130 130 # desplaza a la derecha en el eje 2 determinadas posiciones
131 131 self.dataOut.data_cspc = numpy.roll(self.dataOut.data_cspc, shift, axis=1)
132 132 if pairsList:
133 133 self.__selectPairs(pairsList)
134 134
135 135 elif self.dataIn.type == "Voltage":
136 136
137 137 self.dataOut.flagNoData = True
138 138
139 139 if nFFTPoints == None:
140 140 raise ValueError("This SpectraProc.run() need nFFTPoints input variable")
141 141
142 142 if nProfiles == None:
143 143 nProfiles = nFFTPoints
144 144
145 145 if ippFactor == None:
146 146 self.dataOut.ippFactor = 1
147 147
148 148 self.dataOut.nFFTPoints = nFFTPoints
149 149
150 150 if self.buffer is None:
151 151 self.buffer = numpy.zeros((self.dataIn.nChannels,
152 152 nProfiles,
153 153 self.dataIn.nHeights),
154 154 dtype='complex')
155 155
156 156 if self.dataIn.flagDataAsBlock:
157 157 nVoltProfiles = self.dataIn.data.shape[1]
158 158
159 159 if nVoltProfiles == nProfiles:
160 160 self.buffer = self.dataIn.data.copy()
161 161 self.profIndex = nVoltProfiles
162 162
163 163 elif nVoltProfiles < nProfiles:
164 164
165 165 if self.profIndex == 0:
166 166 self.id_min = 0
167 167 self.id_max = nVoltProfiles
168 168
169 169 self.buffer[:, self.id_min:self.id_max,
170 170 :] = self.dataIn.data
171 171 self.profIndex += nVoltProfiles
172 172 self.id_min += nVoltProfiles
173 173 self.id_max += nVoltProfiles
174 174 else:
175 175 raise ValueError("The type object %s has %d profiles, it should just has %d profiles" % (
176 176 self.dataIn.type, self.dataIn.data.shape[1], nProfiles))
177 177 self.dataOut.flagNoData = True
178 178 else:
179 179 self.buffer[:, self.profIndex, :] = self.dataIn.data.copy()
180 180 self.profIndex += 1
181 181
182 182 if self.firstdatatime == None:
183 183 self.firstdatatime = self.dataIn.utctime
184 184
185 185 if self.profIndex == nProfiles:
186 186 self.__updateSpecFromVoltage()
187 187 if pairsList == None:
188 188 self.dataOut.pairsList = [pair for pair in itertools.combinations(self.dataOut.channelList, 2)]
189 189 else:
190 190 self.dataOut.pairsList = pairsList
191 191 self.__getFft()
192 192 self.dataOut.flagNoData = False
193 193 self.firstdatatime = None
194 194 self.profIndex = 0
195 195 else:
196 196 raise ValueError("The type of input object '%s' is not valid".format(
197 197 self.dataIn.type))
198 198
199 199 def __selectPairs(self, pairsList):
200 200
201 201 if not pairsList:
202 202 return
203 203
204 204 pairs = []
205 205 pairsIndex = []
206 206
207 207 for pair in pairsList:
208 208 if pair[0] not in self.dataOut.channelList or pair[1] not in self.dataOut.channelList:
209 209 continue
210 210 pairs.append(pair)
211 211 pairsIndex.append(pairs.index(pair))
212 212
213 213 self.dataOut.data_cspc = self.dataOut.data_cspc[pairsIndex]
214 214 self.dataOut.pairsList = pairs
215 215
216 216 return
217 217
218 218 def selectFFTs(self, minFFT, maxFFT):
219 219 """
220 220 Selecciona un bloque de datos en base a un grupo de valores de puntos FFTs segun el rango
221 221 minFFT<= FFT <= maxFFT
222 222 """
223 223
224 224 if (minFFT > maxFFT):
225 225 raise ValueError("Error selecting heights: Height range (%d,%d) is not valid" % (minFFT, maxFFT))
226 226
227 227 if (minFFT < self.dataOut.getFreqRange()[0]):
228 228 minFFT = self.dataOut.getFreqRange()[0]
229 229
230 230 if (maxFFT > self.dataOut.getFreqRange()[-1]):
231 231 maxFFT = self.dataOut.getFreqRange()[-1]
232 232
233 233 minIndex = 0
234 234 maxIndex = 0
235 235 FFTs = self.dataOut.getFreqRange()
236 236
237 237 inda = numpy.where(FFTs >= minFFT)
238 238 indb = numpy.where(FFTs <= maxFFT)
239 239
240 240 try:
241 241 minIndex = inda[0][0]
242 242 except:
243 243 minIndex = 0
244 244
245 245 try:
246 246 maxIndex = indb[0][-1]
247 247 except:
248 248 maxIndex = len(FFTs)
249 249
250 250 self.selectFFTsByIndex(minIndex, maxIndex)
251 251
252 252 return 1
253 253
254 254 def getBeaconSignal(self, tauindex=0, channelindex=0, hei_ref=None):
255 255 newheis = numpy.where(
256 256 self.dataOut.heightList > self.dataOut.radarControllerHeaderObj.Taus[tauindex])
257 257
258 258 if hei_ref != None:
259 259 newheis = numpy.where(self.dataOut.heightList > hei_ref)
260 260
261 261 minIndex = min(newheis[0])
262 262 maxIndex = max(newheis[0])
263 263 data_spc = self.dataOut.data_spc[:, :, minIndex:maxIndex + 1]
264 264 heightList = self.dataOut.heightList[minIndex:maxIndex + 1]
265 265
266 266 # determina indices
267 267 nheis = int(self.dataOut.radarControllerHeaderObj.txB /
268 268 (self.dataOut.heightList[1] - self.dataOut.heightList[0]))
269 269 avg_dB = 10 * \
270 270 numpy.log10(numpy.sum(data_spc[channelindex, :, :], axis=0))
271 271 beacon_dB = numpy.sort(avg_dB)[-nheis:]
272 272 beacon_heiIndexList = []
273 273 for val in avg_dB.tolist():
274 274 if val >= beacon_dB[0]:
275 275 beacon_heiIndexList.append(avg_dB.tolist().index(val))
276 276
277 277 # data_spc = data_spc[:,:,beacon_heiIndexList]
278 278 data_cspc = None
279 279 if self.dataOut.data_cspc is not None:
280 280 data_cspc = self.dataOut.data_cspc[:, :, minIndex:maxIndex + 1]
281 281 # data_cspc = data_cspc[:,:,beacon_heiIndexList]
282 282
283 283 data_dc = None
284 284 if self.dataOut.data_dc is not None:
285 285 data_dc = self.dataOut.data_dc[:, minIndex:maxIndex + 1]
286 286 # data_dc = data_dc[:,beacon_heiIndexList]
287 287
288 288 self.dataOut.data_spc = data_spc
289 289 self.dataOut.data_cspc = data_cspc
290 290 self.dataOut.data_dc = data_dc
291 291 self.dataOut.heightList = heightList
292 292 self.dataOut.beacon_heiIndexList = beacon_heiIndexList
293 293
294 294 return 1
295 295
296 296 def selectFFTsByIndex(self, minIndex, maxIndex):
297 297 """
298 298
299 299 """
300 300
301 301 if (minIndex < 0) or (minIndex > maxIndex):
302 302 raise ValueError("Error selecting heights: Index range (%d,%d) is not valid" % (minIndex, maxIndex))
303 303
304 304 if (maxIndex >= self.dataOut.nProfiles):
305 305 maxIndex = self.dataOut.nProfiles - 1
306 306
307 307 # Spectra
308 308 data_spc = self.dataOut.data_spc[:, minIndex:maxIndex + 1, :]
309 309
310 310 data_cspc = None
311 311 if self.dataOut.data_cspc is not None:
312 312 data_cspc = self.dataOut.data_cspc[:, minIndex:maxIndex + 1, :]
313 313
314 314 data_dc = None
315 315 if self.dataOut.data_dc is not None:
316 316 data_dc = self.dataOut.data_dc[minIndex:maxIndex + 1, :]
317 317
318 318 self.dataOut.data_spc = data_spc
319 319 self.dataOut.data_cspc = data_cspc
320 320 self.dataOut.data_dc = data_dc
321 321
322 322 self.dataOut.ippSeconds = self.dataOut.ippSeconds * (self.dataOut.nFFTPoints / numpy.shape(data_cspc)[1])
323 323 self.dataOut.nFFTPoints = numpy.shape(data_cspc)[1]
324 324 self.dataOut.profilesPerBlock = numpy.shape(data_cspc)[1]
325 325
326 326 return 1
327 327
328 328 def getNoise(self, minHei=None, maxHei=None, minVel=None, maxVel=None):
329 329 # validacion de rango
330 330 if minHei == None:
331 331 minHei = self.dataOut.heightList[0]
332 332
333 333 if maxHei == None:
334 334 maxHei = self.dataOut.heightList[-1]
335 335
336 336 if (minHei < self.dataOut.heightList[0]) or (minHei > maxHei):
337 337 print('minHei: %.2f is out of the heights range' % (minHei))
338 338 print('minHei is setting to %.2f' % (self.dataOut.heightList[0]))
339 339 minHei = self.dataOut.heightList[0]
340 340
341 341 if (maxHei > self.dataOut.heightList[-1]) or (maxHei < minHei):
342 342 print('maxHei: %.2f is out of the heights range' % (maxHei))
343 343 print('maxHei is setting to %.2f' % (self.dataOut.heightList[-1]))
344 344 maxHei = self.dataOut.heightList[-1]
345 345
346 346 # validacion de velocidades
347 347 velrange = self.dataOut.getVelRange(1)
348 348
349 349 if minVel == None:
350 350 minVel = velrange[0]
351 351
352 352 if maxVel == None:
353 353 maxVel = velrange[-1]
354 354
355 355 if (minVel < velrange[0]) or (minVel > maxVel):
356 356 print('minVel: %.2f is out of the velocity range' % (minVel))
357 357 print('minVel is setting to %.2f' % (velrange[0]))
358 358 minVel = velrange[0]
359 359
360 360 if (maxVel > velrange[-1]) or (maxVel < minVel):
361 361 print('maxVel: %.2f is out of the velocity range' % (maxVel))
362 362 print('maxVel is setting to %.2f' % (velrange[-1]))
363 363 maxVel = velrange[-1]
364 364
365 365 # seleccion de indices para rango
366 366 minIndex = 0
367 367 maxIndex = 0
368 368 heights = self.dataOut.heightList
369 369
370 370 inda = numpy.where(heights >= minHei)
371 371 indb = numpy.where(heights <= maxHei)
372 372
373 373 try:
374 374 minIndex = inda[0][0]
375 375 except:
376 376 minIndex = 0
377 377
378 378 try:
379 379 maxIndex = indb[0][-1]
380 380 except:
381 381 maxIndex = len(heights)
382 382
383 383 if (minIndex < 0) or (minIndex > maxIndex):
384 384 raise ValueError("some value in (%d,%d) is not valid" % (
385 385 minIndex, maxIndex))
386 386
387 387 if (maxIndex >= self.dataOut.nHeights):
388 388 maxIndex = self.dataOut.nHeights - 1
389 389
390 390 # seleccion de indices para velocidades
391 391 indminvel = numpy.where(velrange >= minVel)
392 392 indmaxvel = numpy.where(velrange <= maxVel)
393 393 try:
394 394 minIndexVel = indminvel[0][0]
395 395 except:
396 396 minIndexVel = 0
397 397
398 398 try:
399 399 maxIndexVel = indmaxvel[0][-1]
400 400 except:
401 401 maxIndexVel = len(velrange)
402 402
403 403 # seleccion del espectro
404 404 data_spc = self.dataOut.data_spc[:,
405 405 minIndexVel:maxIndexVel + 1, minIndex:maxIndex + 1]
406 406 # estimacion de ruido
407 407 noise = numpy.zeros(self.dataOut.nChannels)
408 408
409 409 for channel in range(self.dataOut.nChannels):
410 410 daux = data_spc[channel, :, :]
411 411 sortdata = numpy.sort(daux, axis=None)
412 412 noise[channel] = hildebrand_sekhon(sortdata, self.dataOut.nIncohInt)
413 413
414 414 self.dataOut.noise_estimation = noise.copy()
415 415
416 416 return 1
417 417
418 418 class removeDC(Operation):
419 419
420 420 def run(self, dataOut, mode=2):
421 421 self.dataOut = dataOut
422 422 jspectra = self.dataOut.data_spc
423 423 jcspectra = self.dataOut.data_cspc
424 424
425 425 num_chan = jspectra.shape[0]
426 426 num_hei = jspectra.shape[2]
427 427
428 428 if jcspectra is not None:
429 429 jcspectraExist = True
430 430 num_pairs = jcspectra.shape[0]
431 431 else:
432 432 jcspectraExist = False
433 433
434 434 freq_dc = int(jspectra.shape[1] / 2)
435 435 ind_vel = numpy.array([-2, -1, 1, 2]) + freq_dc
436 436 ind_vel = ind_vel.astype(int)
437 437
438 438 if ind_vel[0] < 0:
439 439 ind_vel[list(range(0, 1))] = ind_vel[list(range(0, 1))] + self.num_prof
440 440
441 441 if mode == 1:
442 442 jspectra[:, freq_dc, :] = (
443 443 jspectra[:, ind_vel[1], :] + jspectra[:, ind_vel[2], :]) / 2 # CORRECCION
444 444
445 445 if jcspectraExist:
446 446 jcspectra[:, freq_dc, :] = (
447 447 jcspectra[:, ind_vel[1], :] + jcspectra[:, ind_vel[2], :]) / 2
448 448
449 449 if mode == 2:
450 450
451 451 vel = numpy.array([-2, -1, 1, 2])
452 452 xx = numpy.zeros([4, 4])
453 453
454 454 for fil in range(4):
455 455 xx[fil, :] = vel[fil] ** numpy.asarray(list(range(4)))
456 456
457 457 xx_inv = numpy.linalg.inv(xx)
458 458 xx_aux = xx_inv[0, :]
459 459
460 460 for ich in range(num_chan):
461 461 yy = jspectra[ich, ind_vel, :]
462 462 jspectra[ich, freq_dc, :] = numpy.dot(xx_aux, yy)
463 463
464 464 junkid = jspectra[ich, freq_dc, :] <= 0
465 465 cjunkid = sum(junkid)
466 466
467 467 if cjunkid.any():
468 468 jspectra[ich, freq_dc, junkid.nonzero()] = (
469 469 jspectra[ich, ind_vel[1], junkid] + jspectra[ich, ind_vel[2], junkid]) / 2
470 470
471 471 if jcspectraExist:
472 472 for ip in range(num_pairs):
473 473 yy = jcspectra[ip, ind_vel, :]
474 474 jcspectra[ip, freq_dc, :] = numpy.dot(xx_aux, yy)
475 475
476 476 self.dataOut.data_spc = jspectra
477 477 self.dataOut.data_cspc = jcspectra
478 478
479 479 return self.dataOut
480 480
481 481 class removeInterference(Operation):
482 482
483 483 def removeInterference2(self):
484 484
485 485 cspc = self.dataOut.data_cspc
486 486 spc = self.dataOut.data_spc
487 487 Heights = numpy.arange(cspc.shape[2])
488 488 realCspc = numpy.abs(cspc)
489 489
490 490 for i in range(cspc.shape[0]):
491 491 LinePower = numpy.sum(realCspc[i], axis=0)
492 492 Threshold = numpy.amax(LinePower) - numpy.sort(LinePower)[len(Heights) - int(len(Heights) * 0.1)]
493 493 SelectedHeights = Heights[ numpy.where(LinePower < Threshold) ]
494 494 InterferenceSum = numpy.sum(realCspc[i, :, SelectedHeights], axis=0)
495 495 InterferenceThresholdMin = numpy.sort(InterferenceSum)[int(len(InterferenceSum) * 0.98)]
496 496 InterferenceThresholdMax = numpy.sort(InterferenceSum)[int(len(InterferenceSum) * 0.99)]
497 497
498 498
499 499 InterferenceRange = numpy.where(([InterferenceSum > InterferenceThresholdMin])) # , InterferenceSum < InterferenceThresholdMax]) )
500 500 # InterferenceRange = numpy.where( ([InterferenceRange < InterferenceThresholdMax]))
501 501 if len(InterferenceRange) < int(cspc.shape[1] * 0.3):
502 502 cspc[i, InterferenceRange, :] = numpy.NaN
503 503
504 504 self.dataOut.data_cspc = cspc
505 505
506 506 def removeInterference(self, interf=2, hei_interf=None, nhei_interf=None, offhei_interf=None):
507 507
508 508 jspectra = self.dataOut.data_spc
509 509 jcspectra = self.dataOut.data_cspc
510 510 jnoise = self.dataOut.getNoise()
511 511 num_incoh = self.dataOut.nIncohInt
512 512
513 513 num_channel = jspectra.shape[0]
514 514 num_prof = jspectra.shape[1]
515 515 num_hei = jspectra.shape[2]
516 516
517 517 # hei_interf
518 518 if hei_interf is None:
519 519 count_hei = int(num_hei / 2)
520 520 hei_interf = numpy.asmatrix(list(range(count_hei))) + num_hei - count_hei
521 521 hei_interf = numpy.asarray(hei_interf)[0]
522 522 # nhei_interf
523 523 if (nhei_interf == None):
524 524 nhei_interf = 5
525 525 if (nhei_interf < 1):
526 526 nhei_interf = 1
527 527 if (nhei_interf > count_hei):
528 528 nhei_interf = count_hei
529 529 if (offhei_interf == None):
530 530 offhei_interf = 0
531 531
532 532 ind_hei = list(range(num_hei))
533 533 # mask_prof = numpy.asarray(range(num_prof - 2)) + 1
534 534 # mask_prof[range(num_prof/2 - 1,len(mask_prof))] += 1
535 535 mask_prof = numpy.asarray(list(range(num_prof)))
536 536 num_mask_prof = mask_prof.size
537 537 comp_mask_prof = [0, num_prof / 2]
538 538
539 539 # noise_exist: Determina si la variable jnoise ha sido definida y contiene la informacion del ruido de cada canal
540 540 if (jnoise.size < num_channel or numpy.isnan(jnoise).any()):
541 541 jnoise = numpy.nan
542 542 noise_exist = jnoise[0] < numpy.Inf
543 543
544 544 # Subrutina de Remocion de la Interferencia
545 545 for ich in range(num_channel):
546 546 # Se ordena los espectros segun su potencia (menor a mayor)
547 547 power = jspectra[ich, mask_prof, :]
548 548 power = power[:, hei_interf]
549 549 power = power.sum(axis=0)
550 550 psort = power.ravel().argsort()
551 551
552 552 # Se estima la interferencia promedio en los Espectros de Potencia empleando
553 553 junkspc_interf = jspectra[ich, :, hei_interf[psort[list(range(
554 554 offhei_interf, nhei_interf + offhei_interf))]]]
555 555
556 556 if noise_exist:
557 557 # tmp_noise = jnoise[ich] / num_prof
558 558 tmp_noise = jnoise[ich]
559 559 junkspc_interf = junkspc_interf - tmp_noise
560 560 # junkspc_interf[:,comp_mask_prof] = 0
561 561
562 562 jspc_interf = junkspc_interf.sum(axis=0) / nhei_interf
563 563 jspc_interf = jspc_interf.transpose()
564 564 # Calculando el espectro de interferencia promedio
565 565 noiseid = numpy.where(
566 566 jspc_interf <= tmp_noise / numpy.sqrt(num_incoh))
567 567 noiseid = noiseid[0]
568 568 cnoiseid = noiseid.size
569 569 interfid = numpy.where(
570 570 jspc_interf > tmp_noise / numpy.sqrt(num_incoh))
571 571 interfid = interfid[0]
572 572 cinterfid = interfid.size
573 573
574 574 if (cnoiseid > 0):
575 575 jspc_interf[noiseid] = 0
576 576
577 577 # Expandiendo los perfiles a limpiar
578 578 if (cinterfid > 0):
579 579 new_interfid = (
580 580 numpy.r_[interfid - 1, interfid, interfid + 1] + num_prof) % num_prof
581 581 new_interfid = numpy.asarray(new_interfid)
582 582 new_interfid = {x for x in new_interfid}
583 583 new_interfid = numpy.array(list(new_interfid))
584 584 new_cinterfid = new_interfid.size
585 585 else:
586 586 new_cinterfid = 0
587 587
588 588 for ip in range(new_cinterfid):
589 589 ind = junkspc_interf[:, new_interfid[ip]].ravel().argsort()
590 590 jspc_interf[new_interfid[ip]
591 591 ] = junkspc_interf[ind[nhei_interf // 2], new_interfid[ip]]
592 592
593 593 jspectra[ich, :, ind_hei] = jspectra[ich, :,
594 594 ind_hei] - jspc_interf # Corregir indices
595 595
596 596 # Removiendo la interferencia del punto de mayor interferencia
597 597 ListAux = jspc_interf[mask_prof].tolist()
598 598 maxid = ListAux.index(max(ListAux))
599 599
600 600 if cinterfid > 0:
601 601 for ip in range(cinterfid * (interf == 2) - 1):
602 602 ind = (jspectra[ich, interfid[ip], :] < tmp_noise *
603 603 (1 + 1 / numpy.sqrt(num_incoh))).nonzero()
604 604 cind = len(ind)
605 605
606 606 if (cind > 0):
607 607 jspectra[ich, interfid[ip], ind] = tmp_noise * \
608 608 (1 + (numpy.random.uniform(cind) - 0.5) /
609 609 numpy.sqrt(num_incoh))
610 610
611 611 ind = numpy.array([-2, -1, 1, 2])
612 612 xx = numpy.zeros([4, 4])
613 613
614 614 for id1 in range(4):
615 615 xx[:, id1] = ind[id1] ** numpy.asarray(list(range(4)))
616 616
617 617 xx_inv = numpy.linalg.inv(xx)
618 618 xx = xx_inv[:, 0]
619 619 ind = (ind + maxid + num_mask_prof) % num_mask_prof
620 620 yy = jspectra[ich, mask_prof[ind], :]
621 621 jspectra[ich, mask_prof[maxid], :] = numpy.dot(
622 622 yy.transpose(), xx)
623 623
624 624 indAux = (jspectra[ich, :, :] < tmp_noise *
625 625 (1 - 1 / numpy.sqrt(num_incoh))).nonzero()
626 626 jspectra[ich, indAux[0], indAux[1]] = tmp_noise * \
627 627 (1 - 1 / numpy.sqrt(num_incoh))
628 628
629 629 # Remocion de Interferencia en el Cross Spectra
630 630 if jcspectra is None:
631 631 return jspectra, jcspectra
632 632 num_pairs = int(jcspectra.size / (num_prof * num_hei))
633 633 jcspectra = jcspectra.reshape(num_pairs, num_prof, num_hei)
634 634
635 635 for ip in range(num_pairs):
636 636
637 637 #-------------------------------------------
638 638
639 639 cspower = numpy.abs(jcspectra[ip, mask_prof, :])
640 640 cspower = cspower[:, hei_interf]
641 641 cspower = cspower.sum(axis=0)
642 642
643 643 cspsort = cspower.ravel().argsort()
644 644 junkcspc_interf = jcspectra[ip, :, hei_interf[cspsort[list(range(
645 645 offhei_interf, nhei_interf + offhei_interf))]]]
646 646 junkcspc_interf = junkcspc_interf.transpose()
647 647 jcspc_interf = junkcspc_interf.sum(axis=1) / nhei_interf
648 648
649 649 ind = numpy.abs(jcspc_interf[mask_prof]).ravel().argsort()
650 650
651 651 median_real = int(numpy.median(numpy.real(
652 652 junkcspc_interf[mask_prof[ind[list(range(3 * num_prof // 4))]], :])))
653 653 median_imag = int(numpy.median(numpy.imag(
654 654 junkcspc_interf[mask_prof[ind[list(range(3 * num_prof // 4))]], :])))
655 655 comp_mask_prof = [int(e) for e in comp_mask_prof]
656 656 junkcspc_interf[comp_mask_prof, :] = numpy.complex(
657 657 median_real, median_imag)
658 658
659 659 for iprof in range(num_prof):
660 660 ind = numpy.abs(junkcspc_interf[iprof, :]).ravel().argsort()
661 661 jcspc_interf[iprof] = junkcspc_interf[iprof, ind[nhei_interf // 2]]
662 662
663 663 # Removiendo la Interferencia
664 664 jcspectra[ip, :, ind_hei] = jcspectra[ip,
665 665 :, ind_hei] - jcspc_interf
666 666
667 667 ListAux = numpy.abs(jcspc_interf[mask_prof]).tolist()
668 668 maxid = ListAux.index(max(ListAux))
669 669
670 670 ind = numpy.array([-2, -1, 1, 2])
671 671 xx = numpy.zeros([4, 4])
672 672
673 673 for id1 in range(4):
674 674 xx[:, id1] = ind[id1] ** numpy.asarray(list(range(4)))
675 675
676 676 xx_inv = numpy.linalg.inv(xx)
677 677 xx = xx_inv[:, 0]
678 678
679 679 ind = (ind + maxid + num_mask_prof) % num_mask_prof
680 680 yy = jcspectra[ip, mask_prof[ind], :]
681 681 jcspectra[ip, mask_prof[maxid], :] = numpy.dot(yy.transpose(), xx)
682 682
683 683 # Guardar Resultados
684 684 self.dataOut.data_spc = jspectra
685 685 self.dataOut.data_cspc = jcspectra
686 686
687 687 return 1
688 688
689 689 def run(self, dataOut, interf=2, hei_interf=None, nhei_interf=None, offhei_interf=None, mode=1):
690 690
691 691 self.dataOut = dataOut
692 692
693 693 if mode == 1:
694 694 self.removeInterference(interf=2, hei_interf=None, nhei_interf=None, offhei_interf=None)
695 695 elif mode == 2:
696 696 self.removeInterference2()
697 697
698 698 return self.dataOut
699 699
700 700
701 class deflip(Operation):
702
703 def run(self, dataOut):
704 # arreglo 1: (num_chan, num_profiles, num_heights)
705 self.dataOut = dataOut
706
707 # JULIA-oblicua, indice 2
708 # arreglo 2: (num_profiles, num_heights)
709 jspectra = self.dataOut.data_spc[2]
710 jspectra_tmp=numpy.zeros(jspectra.shape)
711 num_profiles=jspectra.shape[0]
712 freq_dc = int(num_profiles / 2)
713 # Flip con for
714 for j in range(num_profiles):
715 jspectra_tmp[num_profiles-j-1]= jspectra[j]
716 # Intercambio perfil de DC con perfil inmediato anterior
717 jspectra_tmp[freq_dc-1]= jspectra[freq_dc-1]
718 jspectra_tmp[freq_dc]= jspectra[freq_dc]
719 # canal modificado es re-escrito en el arreglo de canales
720 self.dataOut.data_spc[2] = jspectra_tmp
721
722 return self.dataOut
723
724
701 725 class IncohInt(Operation):
702 726
703 727 __profIndex = 0
704 728 __withOverapping = False
705 729
706 730 __byTime = False
707 731 __initime = None
708 732 __lastdatatime = None
709 733 __integrationtime = None
710 734
711 735 __buffer_spc = None
712 736 __buffer_cspc = None
713 737 __buffer_dc = None
714 738
715 739 __dataReady = False
716 740
717 741 __timeInterval = None
718 742
719 743 n = None
720 744
721 745 def __init__(self):
722 746
723 747 Operation.__init__(self)
724 748
725 749 def setup(self, n=None, timeInterval=None, overlapping=False):
726 750 """
727 751 Set the parameters of the integration class.
728 752
729 753 Inputs:
730 754
731 755 n : Number of coherent integrations
732 756 timeInterval : Time of integration. If the parameter "n" is selected this one does not work
733 757 overlapping :
734 758
735 759 """
736 760
737 761 self.__initime = None
738 762 self.__lastdatatime = 0
739 763
740 764 self.__buffer_spc = 0
741 765 self.__buffer_cspc = 0
742 766 self.__buffer_dc = 0
743 767
744 768 self.__profIndex = 0
745 769 self.__dataReady = False
746 770 self.__byTime = False
747 771
748 772 if n is None and timeInterval is None:
749 773 raise ValueError("n or timeInterval should be specified ...")
750 774
751 775 if n is not None:
752 776 self.n = int(n)
753 777 else:
754 778
755 779 self.__integrationtime = int(timeInterval)
756 780 self.n = None
757 781 self.__byTime = True
758 782
759 783 def putData(self, data_spc, data_cspc, data_dc):
760 784 """
761 785 Add a profile to the __buffer_spc and increase in one the __profileIndex
762 786
763 787 """
764 788
765 789 self.__buffer_spc += data_spc
766 790
767 791 if data_cspc is None:
768 792 self.__buffer_cspc = None
769 793 else:
770 794 self.__buffer_cspc += data_cspc
771 795
772 796 if data_dc is None:
773 797 self.__buffer_dc = None
774 798 else:
775 799 self.__buffer_dc += data_dc
776 800
777 801 self.__profIndex += 1
778 802
779 803 return
780 804
781 805 def pushData(self):
782 806 """
783 807 Return the sum of the last profiles and the profiles used in the sum.
784 808
785 809 Affected:
786 810
787 811 self.__profileIndex
788 812
789 813 """
790 814
791 815 data_spc = self.__buffer_spc
792 816 data_cspc = self.__buffer_cspc
793 817 data_dc = self.__buffer_dc
794 818 n = self.__profIndex
795 819
796 820 self.__buffer_spc = 0
797 821 self.__buffer_cspc = 0
798 822 self.__buffer_dc = 0
799 823 self.__profIndex = 0
800 824
801 825 return data_spc, data_cspc, data_dc, n
802 826
803 827 def byProfiles(self, *args):
804 828
805 829 self.__dataReady = False
806 830 avgdata_spc = None
807 831 avgdata_cspc = None
808 832 avgdata_dc = None
809 833
810 834 self.putData(*args)
811 835
812 836 if self.__profIndex == self.n:
813 837
814 838 avgdata_spc, avgdata_cspc, avgdata_dc, n = self.pushData()
815 839 self.n = n
816 840 self.__dataReady = True
817 841
818 842 return avgdata_spc, avgdata_cspc, avgdata_dc
819 843
820 844 def byTime(self, datatime, *args):
821 845
822 846 self.__dataReady = False
823 847 avgdata_spc = None
824 848 avgdata_cspc = None
825 849 avgdata_dc = None
826 850
827 851 self.putData(*args)
828 852
829 853 if (datatime - self.__initime) >= self.__integrationtime:
830 854 avgdata_spc, avgdata_cspc, avgdata_dc, n = self.pushData()
831 855 self.n = n
832 856 self.__dataReady = True
833 857
834 858 return avgdata_spc, avgdata_cspc, avgdata_dc
835 859
836 860 def integrate(self, datatime, *args):
837 861
838 862 if self.__profIndex == 0:
839 863 self.__initime = datatime
840 864
841 865 if self.__byTime:
842 866 avgdata_spc, avgdata_cspc, avgdata_dc = self.byTime(
843 867 datatime, *args)
844 868 else:
845 869 avgdata_spc, avgdata_cspc, avgdata_dc = self.byProfiles(*args)
846 870
847 871 if not self.__dataReady:
848 872 return None, None, None, None
849 873
850 874 return self.__initime, avgdata_spc, avgdata_cspc, avgdata_dc
851 875
852 876 def run(self, dataOut, n=None, timeInterval=None, overlapping=False):
853 877 if n == 1:
854 878 return dataOut
855 879
856 880 dataOut.flagNoData = True
857 881
858 882 if not self.isConfig:
859 883 self.setup(n, timeInterval, overlapping)
860 884 self.isConfig = True
861 885
862 886 avgdatatime, avgdata_spc, avgdata_cspc, avgdata_dc = self.integrate(dataOut.utctime,
863 887 dataOut.data_spc,
864 888 dataOut.data_cspc,
865 889 dataOut.data_dc)
866 890
867 891 if self.__dataReady:
868 892
869 893 dataOut.data_spc = avgdata_spc
870 894 dataOut.data_cspc = avgdata_cspc
871 895 dataOut.data_dc = avgdata_dc
872 896 dataOut.nIncohInt *= self.n
873 897 dataOut.utctime = avgdatatime
874 898 dataOut.flagNoData = False
875 899
876 900 return dataOut
877 901
878 902 class dopplerFlip(Operation):
879 903
880 904 def run(self, dataOut):
881 905 # arreglo 1: (num_chan, num_profiles, num_heights)
882 906 self.dataOut = dataOut
883 907 # JULIA-oblicua, indice 2
884 908 # arreglo 2: (num_profiles, num_heights)
885 909 jspectra = self.dataOut.data_spc[2]
886 910 jspectra_tmp = numpy.zeros(jspectra.shape)
887 911 num_profiles = jspectra.shape[0]
888 912 freq_dc = int(num_profiles / 2)
889 913 # Flip con for
890 914 for j in range(num_profiles):
891 915 jspectra_tmp[num_profiles - j - 1] = jspectra[j]
892 916 # Intercambio perfil de DC con perfil inmediato anterior
893 917 jspectra_tmp[freq_dc - 1] = jspectra[freq_dc - 1]
894 918 jspectra_tmp[freq_dc] = jspectra[freq_dc]
895 919 # canal modificado es re-escrito en el arreglo de canales
896 920 self.dataOut.data_spc[2] = jspectra_tmp
897 921
898 922 return self.dataOut
General Comments 0
You need to be logged in to leave comments. Login now