Coverage for python/lsst/pipe/tasks/dcrAssembleCoadd.py: 14%

Shortcuts on this page

r m x p   toggle line displays

j k   next/prev highlighted chunk

0   (zero) top of page

1   (one) first highlighted chunk

458 statements  

1# This file is part of pipe_tasks. 

2# 

3# LSST Data Management System 

4# This product includes software developed by the 

5# LSST Project (http://www.lsst.org/). 

6# See COPYRIGHT file at the top of the source tree. 

7# 

8# This program is free software: you can redistribute it and/or modify 

9# it under the terms of the GNU General Public License as published by 

10# the Free Software Foundation, either version 3 of the License, or 

11# (at your option) any later version. 

12# 

13# This program is distributed in the hope that it will be useful, 

14# but WITHOUT ANY WARRANTY; without even the implied warranty of 

15# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

16# GNU General Public License for more details. 

17# 

18# You should have received a copy of the LSST License Statement and 

19# the GNU General Public License along with this program. If not, 

20# see <https://www.lsstcorp.org/LegalNotices/>. 

21# 

22 

23from math import ceil 

24import numpy as np 

25from scipy import ndimage 

26import lsst.geom as geom 

27import lsst.afw.image as afwImage 

28import lsst.afw.table as afwTable 

29import lsst.coadd.utils as coaddUtils 

30from lsst.daf.butler import DeferredDatasetHandle 

31from lsst.ip.diffim.dcrModel import applyDcr, calculateDcr, DcrModel 

32import lsst.meas.algorithms as measAlg 

33from lsst.meas.base import SingleFrameMeasurementTask 

34import lsst.pex.config as pexConfig 

35import lsst.pipe.base as pipeBase 

36import lsst.utils as utils 

37from lsst.utils.timer import timeMethod 

38from .assembleCoadd import (AssembleCoaddConnections, 

39 AssembleCoaddTask, 

40 CompareWarpAssembleCoaddConfig, 

41 CompareWarpAssembleCoaddTask) 

42from .coaddBase import makeSkyInfo 

43from .measurePsf import MeasurePsfTask 

44 

45__all__ = ["DcrAssembleCoaddConnections", "DcrAssembleCoaddTask", "DcrAssembleCoaddConfig"] 

46 

47 

48class DcrAssembleCoaddConnections(AssembleCoaddConnections, 

49 dimensions=("tract", "patch", "band", "skymap"), 

50 defaultTemplates={"inputWarpName": "deep", 

51 "inputCoaddName": "deep", 

52 "outputCoaddName": "dcr", 

53 "warpType": "direct", 

54 "warpTypeSuffix": "", 

55 "fakesType": ""}): 

56 inputWarps = pipeBase.connectionTypes.Input( 

57 doc=("Input list of warps to be assembled i.e. stacked." 

58 "Note that this will often be different than the inputCoaddName." 

59 "WarpType (e.g. direct, psfMatched) is controlled by the warpType config parameter"), 

60 name="{inputWarpName}Coadd_{warpType}Warp", 

61 storageClass="ExposureF", 

62 dimensions=("tract", "patch", "skymap", "visit", "instrument"), 

63 deferLoad=True, 

64 multiple=True 

65 ) 

66 templateExposure = pipeBase.connectionTypes.Input( 

67 doc="Input coadded exposure, produced by previous call to AssembleCoadd", 

68 name="{fakesType}{inputCoaddName}Coadd{warpTypeSuffix}", 

69 storageClass="ExposureF", 

70 dimensions=("tract", "patch", "skymap", "band"), 

71 ) 

72 dcrCoadds = pipeBase.connectionTypes.Output( 

73 doc="Output coadded exposure, produced by stacking input warps", 

74 name="{fakesType}{outputCoaddName}Coadd{warpTypeSuffix}", 

75 storageClass="ExposureF", 

76 dimensions=("tract", "patch", "skymap", "band", "subfilter"), 

77 multiple=True, 

78 ) 

79 dcrNImages = pipeBase.connectionTypes.Output( 

80 doc="Output image of number of input images per pixel", 

81 name="{outputCoaddName}Coadd_nImage", 

82 storageClass="ImageU", 

83 dimensions=("tract", "patch", "skymap", "band", "subfilter"), 

84 multiple=True, 

85 ) 

86 

87 def __init__(self, *, config=None): 

88 super().__init__(config=config) 

89 if not config.doWrite: 

90 self.outputs.remove("dcrCoadds") 

91 if not config.doNImage: 

92 self.outputs.remove("dcrNImages") 

93 # Remove outputs inherited from ``AssembleCoaddConnections`` that are not used 

94 self.outputs.remove("coaddExposure") 

95 self.outputs.remove("nImage") 

96 

97 

98class DcrAssembleCoaddConfig(CompareWarpAssembleCoaddConfig, 

99 pipelineConnections=DcrAssembleCoaddConnections): 

100 dcrNumSubfilters = pexConfig.Field( 

101 dtype=int, 

102 doc="Number of sub-filters to forward model chromatic effects to fit the supplied exposures.", 

103 default=3, 

104 ) 

105 maxNumIter = pexConfig.Field( 

106 dtype=int, 

107 optional=True, 

108 doc="Maximum number of iterations of forward modeling.", 

109 default=None, 

110 ) 

111 minNumIter = pexConfig.Field( 

112 dtype=int, 

113 optional=True, 

114 doc="Minimum number of iterations of forward modeling.", 

115 default=None, 

116 ) 

117 convergenceThreshold = pexConfig.Field( 

118 dtype=float, 

119 doc="Target relative change in convergence between iterations of forward modeling.", 

120 default=0.001, 

121 ) 

122 useConvergence = pexConfig.Field( 

123 dtype=bool, 

124 doc="Use convergence test as a forward modeling end condition?" 

125 "If not set, skips calculating convergence and runs for ``maxNumIter`` iterations", 

126 default=True, 

127 ) 

128 baseGain = pexConfig.Field( 

129 dtype=float, 

130 optional=True, 

131 doc="Relative weight to give the new solution vs. the last solution when updating the model." 

132 "A value of 1.0 gives equal weight to both solutions." 

133 "Small values imply slower convergence of the solution, but can " 

134 "help prevent overshooting and failures in the fit." 

135 "If ``baseGain`` is None, a conservative gain " 

136 "will be calculated from the number of subfilters. ", 

137 default=None, 

138 ) 

139 useProgressiveGain = pexConfig.Field( 

140 dtype=bool, 

141 doc="Use a gain that slowly increases above ``baseGain`` to accelerate convergence? " 

142 "When calculating the next gain, we use up to 5 previous gains and convergence values." 

143 "Can be set to False to force the model to change at the rate of ``baseGain``. ", 

144 default=True, 

145 ) 

146 doAirmassWeight = pexConfig.Field( 

147 dtype=bool, 

148 doc="Weight exposures by airmass? Useful if there are relatively few high-airmass observations.", 

149 default=False, 

150 ) 

151 modelWeightsWidth = pexConfig.Field( 

152 dtype=float, 

153 doc="Width of the region around detected sources to include in the DcrModel.", 

154 default=3, 

155 ) 

156 useModelWeights = pexConfig.Field( 

157 dtype=bool, 

158 doc="Width of the region around detected sources to include in the DcrModel.", 

159 default=True, 

160 ) 

161 splitSubfilters = pexConfig.Field( 

162 dtype=bool, 

163 doc="Calculate DCR for two evenly-spaced wavelengths in each subfilter." 

164 "Instead of at the midpoint", 

165 default=True, 

166 ) 

167 splitThreshold = pexConfig.Field( 

168 dtype=float, 

169 doc="Minimum DCR difference within a subfilter to use ``splitSubfilters``, in pixels." 

170 "Set to 0 to always split the subfilters.", 

171 default=0.1, 

172 ) 

173 regularizeModelIterations = pexConfig.Field( 

174 dtype=float, 

175 doc="Maximum relative change of the model allowed between iterations." 

176 "Set to zero to disable.", 

177 default=2., 

178 ) 

179 regularizeModelFrequency = pexConfig.Field( 

180 dtype=float, 

181 doc="Maximum relative change of the model allowed between subfilters." 

182 "Set to zero to disable.", 

183 default=4., 

184 ) 

185 convergenceMaskPlanes = pexConfig.ListField( 

186 dtype=str, 

187 default=["DETECTED"], 

188 doc="Mask planes to use to calculate convergence." 

189 ) 

190 regularizationWidth = pexConfig.Field( 

191 dtype=int, 

192 default=2, 

193 doc="Minimum radius of a region to include in regularization, in pixels." 

194 ) 

195 imageInterpOrder = pexConfig.Field( 

196 dtype=int, 

197 doc="The order of the spline interpolation used to shift the image plane.", 

198 default=3, 

199 ) 

200 accelerateModel = pexConfig.Field( 

201 dtype=float, 

202 doc="Factor to amplify the differences between model planes by to speed convergence.", 

203 default=3, 

204 ) 

205 doCalculatePsf = pexConfig.Field( 

206 dtype=bool, 

207 doc="Set to detect stars and recalculate the PSF from the final coadd." 

208 "Otherwise the PSF is estimated from a selection of the best input exposures", 

209 default=False, 

210 ) 

211 detectPsfSources = pexConfig.ConfigurableField( 

212 target=measAlg.SourceDetectionTask, 

213 doc="Task to detect sources for PSF measurement, if ``doCalculatePsf`` is set.", 

214 ) 

215 measurePsfSources = pexConfig.ConfigurableField( 

216 target=SingleFrameMeasurementTask, 

217 doc="Task to measure sources for PSF measurement, if ``doCalculatePsf`` is set." 

218 ) 

219 measurePsf = pexConfig.ConfigurableField( 

220 target=MeasurePsfTask, 

221 doc="Task to measure the PSF of the coadd, if ``doCalculatePsf`` is set.", 

222 ) 

223 effectiveWavelength = pexConfig.Field( 

224 doc="Effective wavelength of the filter, in nm." 

225 "Required if transmission curves aren't used." 

226 "Support for using transmission curves is to be added in DM-13668.", 

227 dtype=float, 

228 ) 

229 bandwidth = pexConfig.Field( 

230 doc="Bandwidth of the physical filter, in nm." 

231 "Required if transmission curves aren't used." 

232 "Support for using transmission curves is to be added in DM-13668.", 

233 dtype=float, 

234 ) 

235 

236 def setDefaults(self): 

237 CompareWarpAssembleCoaddConfig.setDefaults(self) 

238 self.assembleStaticSkyModel.retarget(CompareWarpAssembleCoaddTask) 

239 self.doNImage = True 

240 self.assembleStaticSkyModel.warpType = self.warpType 

241 # The deepCoadd and nImage files will be overwritten by this Task, so don't write them the first time 

242 self.assembleStaticSkyModel.doNImage = False 

243 self.assembleStaticSkyModel.doWrite = False 

244 self.detectPsfSources.returnOriginalFootprints = False 

245 self.detectPsfSources.thresholdPolarity = "positive" 

246 # Only use bright sources for PSF measurement 

247 self.detectPsfSources.thresholdValue = 50 

248 self.detectPsfSources.nSigmaToGrow = 2 

249 # A valid star for PSF measurement should at least fill 5x5 pixels 

250 self.detectPsfSources.minPixels = 25 

251 # Use the variance plane to calculate signal to noise 

252 self.detectPsfSources.thresholdType = "pixel_stdev" 

253 # The signal to noise limit is good enough, while the flux limit is set 

254 # in dimensionless units and may not be appropriate for all data sets. 

255 self.measurePsf.starSelector["objectSize"].doFluxLimit = False 

256 

257 

258class DcrAssembleCoaddTask(CompareWarpAssembleCoaddTask): 

259 """Assemble DCR coadded images from a set of warps. 

260 

261 Attributes 

262 ---------- 

263 bufferSize : `int` 

264 The number of pixels to grow each subregion by to allow for DCR. 

265 

266 Notes 

267 ----- 

268 As with AssembleCoaddTask, we want to assemble a coadded image from a set of 

269 Warps (also called coadded temporary exposures), including the effects of 

270 Differential Chromatic Refraction (DCR). 

271 For full details of the mathematics and algorithm, please see 

272 DMTN-037: DCR-matched template generation (https://dmtn-037.lsst.io). 

273 

274 This Task produces a DCR-corrected deepCoadd, as well as a dcrCoadd for 

275 each subfilter used in the iterative calculation. 

276 It begins by dividing the bandpass-defining filter into N equal bandwidth 

277 "subfilters", and divides the flux in each pixel from an initial coadd 

278 equally into each as a "dcrModel". Because the airmass and parallactic 

279 angle of each individual exposure is known, we can calculate the shift 

280 relative to the center of the band in each subfilter due to DCR. For each 

281 exposure we apply this shift as a linear transformation to the dcrModels 

282 and stack the results to produce a DCR-matched exposure. The matched 

283 exposures are subtracted from the input exposures to produce a set of 

284 residual images, and these residuals are reverse shifted for each 

285 exposures' subfilters and stacked. The shifted and stacked residuals are 

286 added to the dcrModels to produce a new estimate of the flux in each pixel 

287 within each subfilter. The dcrModels are solved for iteratively, which 

288 continues until the solution from a new iteration improves by less than 

289 a set percentage, or a maximum number of iterations is reached. 

290 Two forms of regularization are employed to reduce unphysical results. 

291 First, the new solution is averaged with the solution from the previous 

292 iteration, which mitigates oscillating solutions where the model 

293 overshoots with alternating very high and low values. 

294 Second, a common degeneracy when the data have a limited range of airmass or 

295 parallactic angle values is for one subfilter to be fit with very low or 

296 negative values, while another subfilter is fit with very high values. This 

297 typically appears in the form of holes next to sources in one subfilter, 

298 and corresponding extended wings in another. Because each subfilter has 

299 a narrow bandwidth we assume that physical sources that are above the noise 

300 level will not vary in flux by more than a factor of `frequencyClampFactor` 

301 between subfilters, and pixels that have flux deviations larger than that 

302 factor will have the excess flux distributed evenly among all subfilters. 

303 If `splitSubfilters` is set, then each subfilter will be further sub- 

304 divided during the forward modeling step (only). This approximates using 

305 a higher number of subfilters that may be necessary for high airmass 

306 observations, but does not increase the number of free parameters in the 

307 fit. This is needed when there are high airmass observations which would 

308 otherwise have significant DCR even within a subfilter. Because calculating 

309 the shifted images takes most of the time, splitting the subfilters is 

310 turned off by way of the `splitThreshold` option for low-airmass 

311 observations that do not suffer from DCR within a subfilter. 

312 """ 

313 

314 ConfigClass = DcrAssembleCoaddConfig 

315 _DefaultName = "dcrAssembleCoadd" 

316 

317 def __init__(self, *args, **kwargs): 

318 super().__init__(*args, **kwargs) 

319 if self.config.doCalculatePsf: 

320 self.schema = afwTable.SourceTable.makeMinimalSchema() 

321 self.makeSubtask("detectPsfSources", schema=self.schema) 

322 self.makeSubtask("measurePsfSources", schema=self.schema) 

323 self.makeSubtask("measurePsf", schema=self.schema) 

324 

325 @utils.inheritDoc(pipeBase.PipelineTask) 

326 def runQuantum(self, butlerQC, inputRefs, outputRefs): 

327 # Docstring to be formatted with info from PipelineTask.runQuantum 

328 """ 

329 Notes 

330 ----- 

331 Assemble a coadd from a set of Warps. 

332 

333 PipelineTask (Gen3) entry point to Coadd a set of Warps. 

334 Analogous to `runDataRef`, it prepares all the data products to be 

335 passed to `run`, and processes the results before returning a struct 

336 of results to be written out. AssembleCoadd cannot fit all Warps in memory. 

337 Therefore, its inputs are accessed subregion by subregion 

338 by the Gen3 `DeferredDatasetHandle` that is analagous to the Gen2 

339 `lsst.daf.persistence.ButlerDataRef`. Any updates to this method should 

340 correspond to an update in `runDataRef` while both entry points 

341 are used. 

342 """ 

343 inputData = butlerQC.get(inputRefs) 

344 

345 # Construct skyInfo expected by run 

346 # Do not remove skyMap from inputData in case makeSupplementaryDataGen3 needs it 

347 skyMap = inputData["skyMap"] 

348 outputDataId = butlerQC.quantum.dataId 

349 

350 inputData['skyInfo'] = makeSkyInfo(skyMap, 

351 tractId=outputDataId['tract'], 

352 patchId=outputDataId['patch']) 

353 

354 # Construct list of input Deferred Datasets 

355 # These quack a bit like like Gen2 DataRefs 

356 warpRefList = inputData['inputWarps'] 

357 # Perform same middle steps as `runDataRef` does 

358 inputs = self.prepareInputs(warpRefList) 

359 self.log.info("Found %d %s", len(inputs.tempExpRefList), 

360 self.getTempExpDatasetName(self.warpType)) 

361 if len(inputs.tempExpRefList) == 0: 

362 self.log.warning("No coadd temporary exposures found") 

363 return 

364 

365 supplementaryData = self.makeSupplementaryDataGen3(butlerQC, inputRefs, outputRefs) 

366 retStruct = self.run(inputData['skyInfo'], inputs.tempExpRefList, inputs.imageScalerList, 

367 inputs.weightList, supplementaryData=supplementaryData) 

368 

369 inputData.setdefault('brightObjectMask', None) 

370 for subfilter in range(self.config.dcrNumSubfilters): 

371 # Use the PSF of the stacked dcrModel, and do not recalculate the PSF for each subfilter 

372 retStruct.dcrCoadds[subfilter].setPsf(retStruct.coaddExposure.getPsf()) 

373 self.processResults(retStruct.dcrCoadds[subfilter], inputData['brightObjectMask'], outputDataId) 

374 

375 if self.config.doWrite: 

376 butlerQC.put(retStruct, outputRefs) 

377 return retStruct 

378 

379 @timeMethod 

380 def runDataRef(self, dataRef, selectDataList=None, warpRefList=None): 

381 """Assemble a coadd from a set of warps. 

382 

383 Coadd a set of Warps. Compute weights to be applied to each Warp and 

384 find scalings to match the photometric zeropoint to a reference Warp. 

385 Assemble the Warps using run method. 

386 Forward model chromatic effects across multiple subfilters, 

387 and subtract from the input Warps to build sets of residuals. 

388 Use the residuals to construct a new ``DcrModel`` for each subfilter, 

389 and iterate until the model converges. 

390 Interpolate over NaNs and optionally write the coadd to disk. 

391 Return the coadded exposure. 

392 

393 Parameters 

394 ---------- 

395 dataRef : `lsst.daf.persistence.ButlerDataRef` 

396 Data reference defining the patch for coaddition and the 

397 reference Warp 

398 selectDataList : `list` of `lsst.daf.persistence.ButlerDataRef` 

399 List of data references to warps. Data to be coadded will be 

400 selected from this list based on overlap with the patch defined by 

401 the data reference. 

402 

403 Returns 

404 ------- 

405 results : `lsst.pipe.base.Struct` 

406 The Struct contains the following fields: 

407 

408 - ``coaddExposure``: coadded exposure (`lsst.afw.image.Exposure`) 

409 - ``nImage``: exposure count image (`lsst.afw.image.ImageU`) 

410 - ``dcrCoadds``: `list` of coadded exposures for each subfilter 

411 - ``dcrNImages``: `list` of exposure count images for each subfilter 

412 """ 

413 if (selectDataList is None and warpRefList is None) or (selectDataList and warpRefList): 

414 raise RuntimeError("runDataRef must be supplied either a selectDataList or warpRefList") 

415 

416 skyInfo = self.getSkyInfo(dataRef) 

417 if warpRefList is None: 

418 calExpRefList = self.selectExposures(dataRef, skyInfo, selectDataList=selectDataList) 

419 if len(calExpRefList) == 0: 

420 self.log.warning("No exposures to coadd") 

421 return 

422 self.log.info("Coadding %d exposures", len(calExpRefList)) 

423 

424 warpRefList = self.getTempExpRefList(dataRef, calExpRefList) 

425 

426 inputData = self.prepareInputs(warpRefList) 

427 self.log.info("Found %d %s", len(inputData.tempExpRefList), 

428 self.getTempExpDatasetName(self.warpType)) 

429 if len(inputData.tempExpRefList) == 0: 

430 self.log.warning("No coadd temporary exposures found") 

431 return 

432 

433 supplementaryData = self.makeSupplementaryData(dataRef, warpRefList=inputData.tempExpRefList) 

434 

435 results = self.run(skyInfo, inputData.tempExpRefList, inputData.imageScalerList, 

436 inputData.weightList, supplementaryData=supplementaryData) 

437 if results is None: 

438 self.log.warning("Could not construct DcrModel for patch %s: no data to coadd.", 

439 skyInfo.patchInfo.getIndex()) 

440 return 

441 

442 if self.config.doCalculatePsf: 

443 self.measureCoaddPsf(results.coaddExposure) 

444 brightObjects = self.readBrightObjectMasks(dataRef) if self.config.doMaskBrightObjects else None 

445 for subfilter in range(self.config.dcrNumSubfilters): 

446 # Use the PSF of the stacked dcrModel, and do not recalculate the PSF for each subfilter 

447 results.dcrCoadds[subfilter].setPsf(results.coaddExposure.getPsf()) 

448 self.processResults(results.dcrCoadds[subfilter], 

449 brightObjectMasks=brightObjects, dataId=dataRef.dataId) 

450 if self.config.doWrite: 

451 self.log.info("Persisting dcrCoadd") 

452 dataRef.put(results.dcrCoadds[subfilter], "dcrCoadd", subfilter=subfilter, 

453 numSubfilters=self.config.dcrNumSubfilters) 

454 if self.config.doNImage and results.dcrNImages is not None: 

455 dataRef.put(results.dcrNImages[subfilter], "dcrCoadd_nImage", subfilter=subfilter, 

456 numSubfilters=self.config.dcrNumSubfilters) 

457 

458 return results 

459 

460 @utils.inheritDoc(AssembleCoaddTask) 

461 def makeSupplementaryDataGen3(self, butlerQC, inputRefs, outputRefs): 

462 """Load the previously-generated template coadd. 

463 

464 This can be removed entirely once we no longer support the Gen 2 butler. 

465 

466 Returns 

467 ------- 

468 templateCoadd : `lsst.pipe.base.Struct` 

469 Result struct with components: 

470 

471 - ``templateCoadd``: coadded exposure (`lsst.afw.image.ExposureF`) 

472 """ 

473 templateCoadd = butlerQC.get(inputRefs.templateExposure) 

474 

475 return pipeBase.Struct(templateCoadd=templateCoadd) 

476 

477 def measureCoaddPsf(self, coaddExposure): 

478 """Detect sources on the coadd exposure and measure the final PSF. 

479 

480 Parameters 

481 ---------- 

482 coaddExposure : `lsst.afw.image.Exposure` 

483 The final coadded exposure. 

484 """ 

485 table = afwTable.SourceTable.make(self.schema) 

486 detResults = self.detectPsfSources.run(table, coaddExposure, clearMask=False) 

487 coaddSources = detResults.sources 

488 self.measurePsfSources.run( 

489 measCat=coaddSources, 

490 exposure=coaddExposure 

491 ) 

492 # Measure the PSF on the stacked subfilter coadds if possible. 

493 # We should already have a decent estimate of the coadd PSF, however, 

494 # so in case of any errors simply log them as a warning and use the 

495 # default PSF. 

496 try: 

497 psfResults = self.measurePsf.run(coaddExposure, coaddSources) 

498 except Exception as e: 

499 self.log.warning("Unable to calculate PSF, using default coadd PSF: %s", e) 

500 else: 

501 coaddExposure.setPsf(psfResults.psf) 

502 

503 def prepareDcrInputs(self, templateCoadd, warpRefList, weightList): 

504 """Prepare the DCR coadd by iterating through the visitInfo of the input warps. 

505 

506 Sets the property ``bufferSize``. 

507 

508 Parameters 

509 ---------- 

510 templateCoadd : `lsst.afw.image.ExposureF` 

511 The initial coadd exposure before accounting for DCR. 

512 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

513 `lsst.daf.persistence.ButlerDataRef` 

514 The data references to the input warped exposures. 

515 weightList : `list` of `float` 

516 The weight to give each input exposure in the coadd 

517 Will be modified in place if ``doAirmassWeight`` is set. 

518 

519 Returns 

520 ------- 

521 dcrModels : `lsst.pipe.tasks.DcrModel` 

522 Best fit model of the true sky after correcting chromatic effects. 

523 

524 Raises 

525 ------ 

526 NotImplementedError 

527 If ``lambdaMin`` is missing from the Mapper class of the obs package being used. 

528 """ 

529 sigma2fwhm = 2.*np.sqrt(2.*np.log(2.)) 

530 filterLabel = templateCoadd.getFilterLabel() 

531 tempExpName = self.getTempExpDatasetName(self.warpType) 

532 dcrShifts = [] 

533 airmassDict = {} 

534 angleDict = {} 

535 psfSizeDict = {} 

536 for visitNum, warpExpRef in enumerate(warpRefList): 

537 if isinstance(warpExpRef, DeferredDatasetHandle): 

538 # Gen 3 API 

539 visitInfo = warpExpRef.get(component="visitInfo") 

540 psf = warpExpRef.get(component="psf") 

541 else: 

542 # Gen 2 API. Delete this when Gen 2 retired 

543 visitInfo = warpExpRef.get(tempExpName + "_visitInfo") 

544 psf = warpExpRef.get(tempExpName).getPsf() 

545 visit = warpExpRef.dataId["visit"] 

546 # Just need a rough estimate; average positions are fine 

547 psfAvgPos = psf.getAveragePosition() 

548 psfSize = psf.computeShape(psfAvgPos).getDeterminantRadius()*sigma2fwhm 

549 airmass = visitInfo.getBoresightAirmass() 

550 parallacticAngle = visitInfo.getBoresightParAngle().asDegrees() 

551 airmassDict[visit] = airmass 

552 angleDict[visit] = parallacticAngle 

553 psfSizeDict[visit] = psfSize 

554 if self.config.doAirmassWeight: 

555 weightList[visitNum] *= airmass 

556 dcrShifts.append(np.max(np.abs(calculateDcr(visitInfo, templateCoadd.getWcs(), 

557 self.config.effectiveWavelength, 

558 self.config.bandwidth, 

559 self.config.dcrNumSubfilters)))) 

560 self.log.info("Selected airmasses:\n%s", airmassDict) 

561 self.log.info("Selected parallactic angles:\n%s", angleDict) 

562 self.log.info("Selected PSF sizes:\n%s", psfSizeDict) 

563 self.bufferSize = int(np.ceil(np.max(dcrShifts)) + 1) 

564 try: 

565 psf = self.selectCoaddPsf(templateCoadd, warpRefList) 

566 except Exception as e: 

567 self.log.warning("Unable to calculate restricted PSF, using default coadd PSF: %s", e) 

568 else: 

569 psf = templateCoadd.getPsf() 

570 dcrModels = DcrModel.fromImage(templateCoadd.maskedImage, 

571 self.config.dcrNumSubfilters, 

572 effectiveWavelength=self.config.effectiveWavelength, 

573 bandwidth=self.config.bandwidth, 

574 filterLabel=filterLabel, 

575 psf=psf) 

576 return dcrModels 

577 

578 @timeMethod 

579 def run(self, skyInfo, warpRefList, imageScalerList, weightList, 

580 supplementaryData=None): 

581 """Assemble the coadd. 

582 

583 Requires additional inputs Struct ``supplementaryData`` to contain a 

584 ``templateCoadd`` that serves as the model of the static sky. 

585 

586 Find artifacts and apply them to the warps' masks creating a list of 

587 alternative masks with a new "CLIPPED" plane and updated "NO_DATA" plane 

588 Then pass these alternative masks to the base class's assemble method. 

589 

590 Divide the ``templateCoadd`` evenly between each subfilter of a 

591 ``DcrModel`` as the starting best estimate of the true wavelength- 

592 dependent sky. Forward model the ``DcrModel`` using the known 

593 chromatic effects in each subfilter and calculate a convergence metric 

594 based on how well the modeled template matches the input warps. If 

595 the convergence has not yet reached the desired threshold, then shift 

596 and stack the residual images to build a new ``DcrModel``. Apply 

597 conditioning to prevent oscillating solutions between iterations or 

598 between subfilters. 

599 

600 Once the ``DcrModel`` reaches convergence or the maximum number of 

601 iterations has been reached, fill the metadata for each subfilter 

602 image and make them proper ``coaddExposure``s. 

603 

604 Parameters 

605 ---------- 

606 skyInfo : `lsst.pipe.base.Struct` 

607 Patch geometry information, from getSkyInfo 

608 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

609 `lsst.daf.persistence.ButlerDataRef` 

610 The data references to the input warped exposures. 

611 imageScalerList : `list` of `lsst.pipe.task.ImageScaler` 

612 The image scalars correct for the zero point of the exposures. 

613 weightList : `list` of `float` 

614 The weight to give each input exposure in the coadd 

615 supplementaryData : `lsst.pipe.base.Struct` 

616 Result struct returned by ``makeSupplementaryData`` with components: 

617 

618 - ``templateCoadd``: coadded exposure (`lsst.afw.image.Exposure`) 

619 

620 Returns 

621 ------- 

622 result : `lsst.pipe.base.Struct` 

623 Result struct with components: 

624 

625 - ``coaddExposure``: coadded exposure (`lsst.afw.image.Exposure`) 

626 - ``nImage``: exposure count image (`lsst.afw.image.ImageU`) 

627 - ``dcrCoadds``: `list` of coadded exposures for each subfilter 

628 - ``dcrNImages``: `list` of exposure count images for each subfilter 

629 """ 

630 minNumIter = self.config.minNumIter or self.config.dcrNumSubfilters 

631 maxNumIter = self.config.maxNumIter or self.config.dcrNumSubfilters*3 

632 templateCoadd = supplementaryData.templateCoadd 

633 baseMask = templateCoadd.mask.clone() 

634 # The variance plane is for each subfilter 

635 # and should be proportionately lower than the full-band image 

636 baseVariance = templateCoadd.variance.clone() 

637 baseVariance /= self.config.dcrNumSubfilters 

638 spanSetMaskList = self.findArtifacts(templateCoadd, warpRefList, imageScalerList) 

639 # Note that the mask gets cleared in ``findArtifacts``, but we want to preserve the mask. 

640 templateCoadd.setMask(baseMask) 

641 badMaskPlanes = self.config.badMaskPlanes[:] 

642 # Note that is important that we do not add "CLIPPED" to ``badMaskPlanes`` 

643 # This is because pixels in observations that are significantly affect by DCR 

644 # are likely to have many pixels that are both "DETECTED" and "CLIPPED", 

645 # but those are necessary to constrain the DCR model. 

646 badPixelMask = templateCoadd.mask.getPlaneBitMask(badMaskPlanes) 

647 

648 stats = self.prepareStats(mask=badPixelMask) 

649 dcrModels = self.prepareDcrInputs(templateCoadd, warpRefList, weightList) 

650 if self.config.doNImage: 

651 dcrNImages, dcrWeights = self.calculateNImage(dcrModels, skyInfo.bbox, warpRefList, 

652 spanSetMaskList, stats.ctrl) 

653 nImage = afwImage.ImageU(skyInfo.bbox) 

654 # Note that this nImage will be a factor of dcrNumSubfilters higher than 

655 # the nImage returned by assembleCoadd for most pixels. This is because each 

656 # subfilter may have a different nImage, and fractional values are not allowed. 

657 for dcrNImage in dcrNImages: 

658 nImage += dcrNImage 

659 else: 

660 dcrNImages = None 

661 

662 subregionSize = geom.Extent2I(*self.config.subregionSize) 

663 nSubregions = (ceil(skyInfo.bbox.getHeight()/subregionSize[1]) 

664 * ceil(skyInfo.bbox.getWidth()/subregionSize[0])) 

665 subIter = 0 

666 for subBBox in self._subBBoxIter(skyInfo.bbox, subregionSize): 

667 modelIter = 0 

668 subIter += 1 

669 self.log.info("Computing coadd over patch %s subregion %s of %s: %s", 

670 skyInfo.patchInfo.getIndex(), subIter, nSubregions, subBBox) 

671 dcrBBox = geom.Box2I(subBBox) 

672 dcrBBox.grow(self.bufferSize) 

673 dcrBBox.clip(dcrModels.bbox) 

674 modelWeights = self.calculateModelWeights(dcrModels, dcrBBox) 

675 subExposures = self.loadSubExposures(dcrBBox, stats.ctrl, warpRefList, 

676 imageScalerList, spanSetMaskList) 

677 convergenceMetric = self.calculateConvergence(dcrModels, subExposures, subBBox, 

678 warpRefList, weightList, stats.ctrl) 

679 self.log.info("Initial convergence : %s", convergenceMetric) 

680 convergenceList = [convergenceMetric] 

681 gainList = [] 

682 convergenceCheck = 1. 

683 refImage = templateCoadd.image 

684 while (convergenceCheck > self.config.convergenceThreshold or modelIter <= minNumIter): 

685 gain = self.calculateGain(convergenceList, gainList) 

686 self.dcrAssembleSubregion(dcrModels, subExposures, subBBox, dcrBBox, warpRefList, 

687 stats.ctrl, convergenceMetric, gain, 

688 modelWeights, refImage, dcrWeights) 

689 if self.config.useConvergence: 

690 convergenceMetric = self.calculateConvergence(dcrModels, subExposures, subBBox, 

691 warpRefList, weightList, stats.ctrl) 

692 if convergenceMetric == 0: 

693 self.log.warning("Coadd patch %s subregion %s had convergence metric of 0.0 which is " 

694 "most likely due to there being no valid data in the region.", 

695 skyInfo.patchInfo.getIndex(), subIter) 

696 break 

697 convergenceCheck = (convergenceList[-1] - convergenceMetric)/convergenceMetric 

698 if (convergenceCheck < 0) & (modelIter > minNumIter): 

699 self.log.warning("Coadd patch %s subregion %s diverged before reaching maximum " 

700 "iterations or desired convergence improvement of %s." 

701 " Divergence: %s", 

702 skyInfo.patchInfo.getIndex(), subIter, 

703 self.config.convergenceThreshold, convergenceCheck) 

704 break 

705 convergenceList.append(convergenceMetric) 

706 if modelIter > maxNumIter: 

707 if self.config.useConvergence: 

708 self.log.warning("Coadd patch %s subregion %s reached maximum iterations " 

709 "before reaching desired convergence improvement of %s." 

710 " Final convergence improvement: %s", 

711 skyInfo.patchInfo.getIndex(), subIter, 

712 self.config.convergenceThreshold, convergenceCheck) 

713 break 

714 

715 if self.config.useConvergence: 

716 self.log.info("Iteration %s with convergence metric %s, %.4f%% improvement (gain: %.2f)", 

717 modelIter, convergenceMetric, 100.*convergenceCheck, gain) 

718 modelIter += 1 

719 else: 

720 if self.config.useConvergence: 

721 self.log.info("Coadd patch %s subregion %s finished with " 

722 "convergence metric %s after %s iterations", 

723 skyInfo.patchInfo.getIndex(), subIter, convergenceMetric, modelIter) 

724 else: 

725 self.log.info("Coadd patch %s subregion %s finished after %s iterations", 

726 skyInfo.patchInfo.getIndex(), subIter, modelIter) 

727 if self.config.useConvergence and convergenceMetric > 0: 

728 self.log.info("Final convergence improvement was %.4f%% overall", 

729 100*(convergenceList[0] - convergenceMetric)/convergenceMetric) 

730 

731 dcrCoadds = self.fillCoadd(dcrModels, skyInfo, warpRefList, weightList, 

732 calibration=self.scaleZeroPoint.getPhotoCalib(), 

733 coaddInputs=templateCoadd.getInfo().getCoaddInputs(), 

734 mask=baseMask, 

735 variance=baseVariance) 

736 coaddExposure = self.stackCoadd(dcrCoadds) 

737 return pipeBase.Struct(coaddExposure=coaddExposure, nImage=nImage, 

738 dcrCoadds=dcrCoadds, dcrNImages=dcrNImages) 

739 

740 def calculateNImage(self, dcrModels, bbox, warpRefList, spanSetMaskList, statsCtrl): 

741 """Calculate the number of exposures contributing to each subfilter. 

742 

743 Parameters 

744 ---------- 

745 dcrModels : `lsst.pipe.tasks.DcrModel` 

746 Best fit model of the true sky after correcting chromatic effects. 

747 bbox : `lsst.geom.box.Box2I` 

748 Bounding box of the patch to coadd. 

749 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

750 `lsst.daf.persistence.ButlerDataRef` 

751 The data references to the input warped exposures. 

752 spanSetMaskList : `list` of `dict` containing spanSet lists, or None 

753 Each element of the `dict` contains the new mask plane name 

754 (e.g. "CLIPPED and/or "NO_DATA") as the key, 

755 and the list of SpanSets to apply to the mask. 

756 statsCtrl : `lsst.afw.math.StatisticsControl` 

757 Statistics control object for coadd 

758 

759 Returns 

760 ------- 

761 dcrNImages : `list` of `lsst.afw.image.ImageU` 

762 List of exposure count images for each subfilter 

763 dcrWeights : `list` of `lsst.afw.image.ImageF` 

764 Per-pixel weights for each subfilter. 

765 Equal to 1/(number of unmasked images contributing to each pixel). 

766 """ 

767 dcrNImages = [afwImage.ImageU(bbox) for subfilter in range(self.config.dcrNumSubfilters)] 

768 dcrWeights = [afwImage.ImageF(bbox) for subfilter in range(self.config.dcrNumSubfilters)] 

769 tempExpName = self.getTempExpDatasetName(self.warpType) 

770 for warpExpRef, altMaskSpans in zip(warpRefList, spanSetMaskList): 

771 if isinstance(warpExpRef, DeferredDatasetHandle): 

772 # Gen 3 API 

773 exposure = warpExpRef.get(parameters={'bbox': bbox}) 

774 else: 

775 # Gen 2 API. Delete this when Gen 2 retired 

776 exposure = warpExpRef.get(tempExpName + "_sub", bbox=bbox) 

777 visitInfo = exposure.getInfo().getVisitInfo() 

778 wcs = exposure.getInfo().getWcs() 

779 mask = exposure.mask 

780 if altMaskSpans is not None: 

781 self.applyAltMaskPlanes(mask, altMaskSpans) 

782 weightImage = np.zeros_like(exposure.image.array) 

783 weightImage[(mask.array & statsCtrl.getAndMask()) == 0] = 1. 

784 # The weights must be shifted in exactly the same way as the residuals, 

785 # because they will be used as the denominator in the weighted average of residuals. 

786 weightsGenerator = self.dcrResiduals(weightImage, visitInfo, wcs, 

787 dcrModels.effectiveWavelength, dcrModels.bandwidth) 

788 for shiftedWeights, dcrNImage, dcrWeight in zip(weightsGenerator, dcrNImages, dcrWeights): 

789 dcrNImage.array += np.rint(shiftedWeights).astype(dcrNImage.array.dtype) 

790 dcrWeight.array += shiftedWeights 

791 # Exclude any pixels that don't have at least one exposure contributing in all subfilters 

792 weightsThreshold = 1. 

793 goodPix = dcrWeights[0].array > weightsThreshold 

794 for weights in dcrWeights[1:]: 

795 goodPix = (weights.array > weightsThreshold) & goodPix 

796 for subfilter in range(self.config.dcrNumSubfilters): 

797 dcrWeights[subfilter].array[goodPix] = 1./dcrWeights[subfilter].array[goodPix] 

798 dcrWeights[subfilter].array[~goodPix] = 0. 

799 dcrNImages[subfilter].array[~goodPix] = 0 

800 return (dcrNImages, dcrWeights) 

801 

802 def dcrAssembleSubregion(self, dcrModels, subExposures, bbox, dcrBBox, warpRefList, 

803 statsCtrl, convergenceMetric, 

804 gain, modelWeights, refImage, dcrWeights): 

805 """Assemble the DCR coadd for a sub-region. 

806 

807 Build a DCR-matched template for each input exposure, then shift the 

808 residuals according to the DCR in each subfilter. 

809 Stack the shifted residuals and apply them as a correction to the 

810 solution from the previous iteration. 

811 Restrict the new model solutions from varying by more than a factor of 

812 `modelClampFactor` from the last solution, and additionally restrict the 

813 individual subfilter models from varying by more than a factor of 

814 `frequencyClampFactor` from their average. 

815 Finally, mitigate potentially oscillating solutions by averaging the new 

816 solution with the solution from the previous iteration, weighted by 

817 their convergence metric. 

818 

819 Parameters 

820 ---------- 

821 dcrModels : `lsst.pipe.tasks.DcrModel` 

822 Best fit model of the true sky after correcting chromatic effects. 

823 subExposures : `dict` of `lsst.afw.image.ExposureF` 

824 The pre-loaded exposures for the current subregion. 

825 bbox : `lsst.geom.box.Box2I` 

826 Bounding box of the subregion to coadd. 

827 dcrBBox : `lsst.geom.box.Box2I` 

828 Sub-region of the coadd which includes a buffer to allow for DCR. 

829 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

830 `lsst.daf.persistence.ButlerDataRef` 

831 The data references to the input warped exposures. 

832 statsCtrl : `lsst.afw.math.StatisticsControl` 

833 Statistics control object for coadd 

834 convergenceMetric : `float` 

835 Quality of fit metric for the matched templates of the input images. 

836 gain : `float`, optional 

837 Relative weight to give the new solution when updating the model. 

838 modelWeights : `numpy.ndarray` or `float` 

839 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

840 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

841 refImage : `lsst.afw.image.Image` 

842 A reference image used to supply the default pixel values. 

843 dcrWeights : `list` of `lsst.afw.image.Image` 

844 Per-pixel weights for each subfilter. 

845 Equal to 1/(number of unmasked images contributing to each pixel). 

846 """ 

847 residualGeneratorList = [] 

848 

849 for warpExpRef in warpRefList: 

850 visit = warpExpRef.dataId["visit"] 

851 exposure = subExposures[visit] 

852 visitInfo = exposure.getInfo().getVisitInfo() 

853 wcs = exposure.getInfo().getWcs() 

854 templateImage = dcrModels.buildMatchedTemplate(exposure=exposure, 

855 order=self.config.imageInterpOrder, 

856 splitSubfilters=self.config.splitSubfilters, 

857 splitThreshold=self.config.splitThreshold, 

858 amplifyModel=self.config.accelerateModel) 

859 residual = exposure.image.array - templateImage.array 

860 # Note that the variance plane here is used to store weights, not the actual variance 

861 residual *= exposure.variance.array 

862 # The residuals are stored as a list of generators. 

863 # This allows the residual for a given subfilter and exposure to be created 

864 # on the fly, instead of needing to store them all in memory. 

865 residualGeneratorList.append(self.dcrResiduals(residual, visitInfo, wcs, 

866 dcrModels.effectiveWavelength, 

867 dcrModels.bandwidth)) 

868 

869 dcrSubModelOut = self.newModelFromResidual(dcrModels, residualGeneratorList, dcrBBox, statsCtrl, 

870 gain=gain, 

871 modelWeights=modelWeights, 

872 refImage=refImage, 

873 dcrWeights=dcrWeights) 

874 dcrModels.assign(dcrSubModelOut, bbox) 

875 

876 def dcrResiduals(self, residual, visitInfo, wcs, effectiveWavelength, bandwidth): 

877 """Prepare a residual image for stacking in each subfilter by applying the reverse DCR shifts. 

878 

879 Parameters 

880 ---------- 

881 residual : `numpy.ndarray` 

882 The residual masked image for one exposure, 

883 after subtracting the matched template 

884 visitInfo : `lsst.afw.image.VisitInfo` 

885 Metadata for the exposure. 

886 wcs : `lsst.afw.geom.SkyWcs` 

887 Coordinate system definition (wcs) for the exposure. 

888 

889 Yields 

890 ------ 

891 residualImage : `numpy.ndarray` 

892 The residual image for the next subfilter, shifted for DCR. 

893 """ 

894 # Pre-calculate the spline-filtered residual image, so that step can be 

895 # skipped in the shift calculation in `applyDcr`. 

896 filteredResidual = ndimage.spline_filter(residual, order=self.config.imageInterpOrder) 

897 # Note that `splitSubfilters` is always turned off in the reverse direction. 

898 # This option introduces additional blurring if applied to the residuals. 

899 dcrShift = calculateDcr(visitInfo, wcs, effectiveWavelength, bandwidth, self.config.dcrNumSubfilters, 

900 splitSubfilters=False) 

901 for dcr in dcrShift: 

902 yield applyDcr(filteredResidual, dcr, useInverse=True, splitSubfilters=False, 

903 doPrefilter=False, order=self.config.imageInterpOrder) 

904 

905 def newModelFromResidual(self, dcrModels, residualGeneratorList, dcrBBox, statsCtrl, 

906 gain, modelWeights, refImage, dcrWeights): 

907 """Calculate a new DcrModel from a set of image residuals. 

908 

909 Parameters 

910 ---------- 

911 dcrModels : `lsst.pipe.tasks.DcrModel` 

912 Current model of the true sky after correcting chromatic effects. 

913 residualGeneratorList : `generator` of `numpy.ndarray` 

914 The residual image for the next subfilter, shifted for DCR. 

915 dcrBBox : `lsst.geom.box.Box2I` 

916 Sub-region of the coadd which includes a buffer to allow for DCR. 

917 statsCtrl : `lsst.afw.math.StatisticsControl` 

918 Statistics control object for coadd 

919 gain : `float` 

920 Relative weight to give the new solution when updating the model. 

921 modelWeights : `numpy.ndarray` or `float` 

922 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

923 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

924 refImage : `lsst.afw.image.Image` 

925 A reference image used to supply the default pixel values. 

926 dcrWeights : `list` of `lsst.afw.image.Image` 

927 Per-pixel weights for each subfilter. 

928 Equal to 1/(number of unmasked images contributing to each pixel). 

929 

930 Returns 

931 ------- 

932 dcrModel : `lsst.pipe.tasks.DcrModel` 

933 New model of the true sky after correcting chromatic effects. 

934 """ 

935 newModelImages = [] 

936 for subfilter, model in enumerate(dcrModels): 

937 residualsList = [next(residualGenerator) for residualGenerator in residualGeneratorList] 

938 residual = np.sum(residualsList, axis=0) 

939 residual *= dcrWeights[subfilter][dcrBBox].array 

940 # `MaskedImage`s only support in-place addition, so rename for readability 

941 newModel = model[dcrBBox].clone() 

942 newModel.array += residual 

943 # Catch any invalid values 

944 badPixels = ~np.isfinite(newModel.array) 

945 newModel.array[badPixels] = model[dcrBBox].array[badPixels] 

946 if self.config.regularizeModelIterations > 0: 

947 dcrModels.regularizeModelIter(subfilter, newModel, dcrBBox, 

948 self.config.regularizeModelIterations, 

949 self.config.regularizationWidth) 

950 newModelImages.append(newModel) 

951 if self.config.regularizeModelFrequency > 0: 

952 dcrModels.regularizeModelFreq(newModelImages, dcrBBox, statsCtrl, 

953 self.config.regularizeModelFrequency, 

954 self.config.regularizationWidth) 

955 dcrModels.conditionDcrModel(newModelImages, dcrBBox, gain=gain) 

956 self.applyModelWeights(newModelImages, refImage[dcrBBox], modelWeights) 

957 return DcrModel(newModelImages, dcrModels.filter, dcrModels.effectiveWavelength, 

958 dcrModels.bandwidth, dcrModels.psf, 

959 dcrModels.mask, dcrModels.variance) 

960 

961 def calculateConvergence(self, dcrModels, subExposures, bbox, warpRefList, weightList, statsCtrl): 

962 """Calculate a quality of fit metric for the matched templates. 

963 

964 Parameters 

965 ---------- 

966 dcrModels : `lsst.pipe.tasks.DcrModel` 

967 Best fit model of the true sky after correcting chromatic effects. 

968 subExposures : `dict` of `lsst.afw.image.ExposureF` 

969 The pre-loaded exposures for the current subregion. 

970 bbox : `lsst.geom.box.Box2I` 

971 Sub-region to coadd 

972 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

973 `lsst.daf.persistence.ButlerDataRef` 

974 The data references to the input warped exposures. 

975 weightList : `list` of `float` 

976 The weight to give each input exposure in the coadd 

977 statsCtrl : `lsst.afw.math.StatisticsControl` 

978 Statistics control object for coadd 

979 

980 Returns 

981 ------- 

982 convergenceMetric : `float` 

983 Quality of fit metric for all input exposures, within the sub-region 

984 """ 

985 significanceImage = np.abs(dcrModels.getReferenceImage(bbox)) 

986 nSigma = 3. 

987 significanceImage += nSigma*dcrModels.calculateNoiseCutoff(dcrModels[1], statsCtrl, 

988 bufferSize=self.bufferSize) 

989 if np.max(significanceImage) == 0: 

990 significanceImage += 1. 

991 weight = 0 

992 metric = 0. 

993 metricList = {} 

994 for warpExpRef, expWeight in zip(warpRefList, weightList): 

995 visit = warpExpRef.dataId["visit"] 

996 exposure = subExposures[visit][bbox] 

997 singleMetric = self.calculateSingleConvergence(dcrModels, exposure, significanceImage, statsCtrl) 

998 metric += singleMetric 

999 metricList[visit] = singleMetric 

1000 weight += 1. 

1001 self.log.info("Individual metrics:\n%s", metricList) 

1002 return 1.0 if weight == 0.0 else metric/weight 

1003 

1004 def calculateSingleConvergence(self, dcrModels, exposure, significanceImage, statsCtrl): 

1005 """Calculate a quality of fit metric for a single matched template. 

1006 

1007 Parameters 

1008 ---------- 

1009 dcrModels : `lsst.pipe.tasks.DcrModel` 

1010 Best fit model of the true sky after correcting chromatic effects. 

1011 exposure : `lsst.afw.image.ExposureF` 

1012 The input warped exposure to evaluate. 

1013 significanceImage : `numpy.ndarray` 

1014 Array of weights for each pixel corresponding to its significance 

1015 for the convergence calculation. 

1016 statsCtrl : `lsst.afw.math.StatisticsControl` 

1017 Statistics control object for coadd 

1018 

1019 Returns 

1020 ------- 

1021 convergenceMetric : `float` 

1022 Quality of fit metric for one exposure, within the sub-region. 

1023 """ 

1024 convergeMask = exposure.mask.getPlaneBitMask(self.config.convergenceMaskPlanes) 

1025 templateImage = dcrModels.buildMatchedTemplate(exposure=exposure, 

1026 order=self.config.imageInterpOrder, 

1027 splitSubfilters=self.config.splitSubfilters, 

1028 splitThreshold=self.config.splitThreshold, 

1029 amplifyModel=self.config.accelerateModel) 

1030 diffVals = np.abs(exposure.image.array - templateImage.array)*significanceImage 

1031 refVals = np.abs(exposure.image.array + templateImage.array)*significanceImage/2. 

1032 

1033 finitePixels = np.isfinite(diffVals) 

1034 goodMaskPixels = (exposure.mask.array & statsCtrl.getAndMask()) == 0 

1035 convergeMaskPixels = exposure.mask.array & convergeMask > 0 

1036 usePixels = finitePixels & goodMaskPixels & convergeMaskPixels 

1037 if np.sum(usePixels) == 0: 

1038 metric = 0. 

1039 else: 

1040 diffUse = diffVals[usePixels] 

1041 refUse = refVals[usePixels] 

1042 metric = np.sum(diffUse/np.median(diffUse))/np.sum(refUse/np.median(diffUse)) 

1043 return metric 

1044 

1045 def stackCoadd(self, dcrCoadds): 

1046 """Add a list of sub-band coadds together. 

1047 

1048 Parameters 

1049 ---------- 

1050 dcrCoadds : `list` of `lsst.afw.image.ExposureF` 

1051 A list of coadd exposures, each exposure containing 

1052 the model for one subfilter. 

1053 

1054 Returns 

1055 ------- 

1056 coaddExposure : `lsst.afw.image.ExposureF` 

1057 A single coadd exposure that is the sum of the sub-bands. 

1058 """ 

1059 coaddExposure = dcrCoadds[0].clone() 

1060 for coadd in dcrCoadds[1:]: 

1061 coaddExposure.maskedImage += coadd.maskedImage 

1062 return coaddExposure 

1063 

1064 def fillCoadd(self, dcrModels, skyInfo, warpRefList, weightList, calibration=None, coaddInputs=None, 

1065 mask=None, variance=None): 

1066 """Create a list of coadd exposures from a list of masked images. 

1067 

1068 Parameters 

1069 ---------- 

1070 dcrModels : `lsst.pipe.tasks.DcrModel` 

1071 Best fit model of the true sky after correcting chromatic effects. 

1072 skyInfo : `lsst.pipe.base.Struct` 

1073 Patch geometry information, from getSkyInfo 

1074 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

1075 `lsst.daf.persistence.ButlerDataRef` 

1076 The data references to the input warped exposures. 

1077 weightList : `list` of `float` 

1078 The weight to give each input exposure in the coadd 

1079 calibration : `lsst.afw.Image.PhotoCalib`, optional 

1080 Scale factor to set the photometric calibration of an exposure. 

1081 coaddInputs : `lsst.afw.Image.CoaddInputs`, optional 

1082 A record of the observations that are included in the coadd. 

1083 mask : `lsst.afw.image.Mask`, optional 

1084 Optional mask to override the values in the final coadd. 

1085 variance : `lsst.afw.image.Image`, optional 

1086 Optional variance plane to override the values in the final coadd. 

1087 

1088 Returns 

1089 ------- 

1090 dcrCoadds : `list` of `lsst.afw.image.ExposureF` 

1091 A list of coadd exposures, each exposure containing 

1092 the model for one subfilter. 

1093 """ 

1094 dcrCoadds = [] 

1095 refModel = dcrModels.getReferenceImage() 

1096 for model in dcrModels: 

1097 if self.config.accelerateModel > 1: 

1098 model.array = (model.array - refModel)*self.config.accelerateModel + refModel 

1099 coaddExposure = afwImage.ExposureF(skyInfo.bbox, skyInfo.wcs) 

1100 if calibration is not None: 

1101 coaddExposure.setPhotoCalib(calibration) 

1102 if coaddInputs is not None: 

1103 coaddExposure.getInfo().setCoaddInputs(coaddInputs) 

1104 # Set the metadata for the coadd, including PSF and aperture corrections. 

1105 self.assembleMetadata(coaddExposure, warpRefList, weightList) 

1106 # Overwrite the PSF 

1107 coaddExposure.setPsf(dcrModels.psf) 

1108 coaddUtils.setCoaddEdgeBits(dcrModels.mask[skyInfo.bbox], dcrModels.variance[skyInfo.bbox]) 

1109 maskedImage = afwImage.MaskedImageF(dcrModels.bbox) 

1110 maskedImage.image = model 

1111 maskedImage.mask = dcrModels.mask 

1112 maskedImage.variance = dcrModels.variance 

1113 coaddExposure.setMaskedImage(maskedImage[skyInfo.bbox]) 

1114 coaddExposure.setPhotoCalib(self.scaleZeroPoint.getPhotoCalib()) 

1115 if mask is not None: 

1116 coaddExposure.setMask(mask) 

1117 if variance is not None: 

1118 coaddExposure.setVariance(variance) 

1119 dcrCoadds.append(coaddExposure) 

1120 return dcrCoadds 

1121 

1122 def calculateGain(self, convergenceList, gainList): 

1123 """Calculate the gain to use for the current iteration. 

1124 

1125 After calculating a new DcrModel, each value is averaged with the 

1126 value in the corresponding pixel from the previous iteration. This 

1127 reduces oscillating solutions that iterative techniques are plagued by, 

1128 and speeds convergence. By far the biggest changes to the model 

1129 happen in the first couple iterations, so we can also use a more 

1130 aggressive gain later when the model is changing slowly. 

1131 

1132 Parameters 

1133 ---------- 

1134 convergenceList : `list` of `float` 

1135 The quality of fit metric from each previous iteration. 

1136 gainList : `list` of `float` 

1137 The gains used in each previous iteration: appended with the new 

1138 gain value. 

1139 Gains are numbers between ``self.config.baseGain`` and 1. 

1140 

1141 Returns 

1142 ------- 

1143 gain : `float` 

1144 Relative weight to give the new solution when updating the model. 

1145 A value of 1.0 gives equal weight to both solutions. 

1146 

1147 Raises 

1148 ------ 

1149 ValueError 

1150 If ``len(convergenceList) != len(gainList)+1``. 

1151 """ 

1152 nIter = len(convergenceList) 

1153 if nIter != len(gainList) + 1: 

1154 raise ValueError("convergenceList (%d) must be one element longer than gainList (%d)." 

1155 % (len(convergenceList), len(gainList))) 

1156 

1157 if self.config.baseGain is None: 

1158 # If ``baseGain`` is not set, calculate it from the number of DCR subfilters 

1159 # The more subfilters being modeled, the lower the gain should be. 

1160 baseGain = 1./(self.config.dcrNumSubfilters - 1) 

1161 else: 

1162 baseGain = self.config.baseGain 

1163 

1164 if self.config.useProgressiveGain and nIter > 2: 

1165 # To calculate the best gain to use, compare the past gains that have been used 

1166 # with the resulting convergences to estimate the best gain to use. 

1167 # Algorithmically, this is a Kalman filter. 

1168 # If forward modeling proceeds perfectly, the convergence metric should 

1169 # asymptotically approach a final value. 

1170 # We can estimate that value from the measured changes in convergence 

1171 # weighted by the gains used in each previous iteration. 

1172 estFinalConv = [((1 + gainList[i])*convergenceList[i + 1] - convergenceList[i])/gainList[i] 

1173 for i in range(nIter - 1)] 

1174 # The convergence metric is strictly positive, so if the estimated final convergence is 

1175 # less than zero, force it to zero. 

1176 estFinalConv = np.array(estFinalConv) 

1177 estFinalConv[estFinalConv < 0] = 0 

1178 # Because the estimate may slowly change over time, only use the most recent measurements. 

1179 estFinalConv = np.median(estFinalConv[max(nIter - 5, 0):]) 

1180 lastGain = gainList[-1] 

1181 lastConv = convergenceList[-2] 

1182 newConv = convergenceList[-1] 

1183 # The predicted convergence is the value we would get if the new model calculated 

1184 # in the previous iteration was perfect. Recall that the updated model that is 

1185 # actually used is the gain-weighted average of the new and old model, 

1186 # so the convergence would be similarly weighted. 

1187 predictedConv = (estFinalConv*lastGain + lastConv)/(1. + lastGain) 

1188 # If the measured and predicted convergence are very close, that indicates 

1189 # that our forward model is accurate and we can use a more aggressive gain 

1190 # If the measured convergence is significantly worse (or better!) than predicted, 

1191 # that indicates that the model is not converging as expected and 

1192 # we should use a more conservative gain. 

1193 delta = (predictedConv - newConv)/((lastConv - estFinalConv)/(1 + lastGain)) 

1194 newGain = 1 - abs(delta) 

1195 # Average the gains to prevent oscillating solutions. 

1196 newGain = (newGain + lastGain)/2. 

1197 gain = max(baseGain, newGain) 

1198 else: 

1199 gain = baseGain 

1200 gainList.append(gain) 

1201 return gain 

1202 

1203 def calculateModelWeights(self, dcrModels, dcrBBox): 

1204 """Build an array that smoothly tapers to 0 away from detected sources. 

1205 

1206 Parameters 

1207 ---------- 

1208 dcrModels : `lsst.pipe.tasks.DcrModel` 

1209 Best fit model of the true sky after correcting chromatic effects. 

1210 dcrBBox : `lsst.geom.box.Box2I` 

1211 Sub-region of the coadd which includes a buffer to allow for DCR. 

1212 

1213 Returns 

1214 ------- 

1215 weights : `numpy.ndarray` or `float` 

1216 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

1217 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

1218 

1219 Raises 

1220 ------ 

1221 ValueError 

1222 If ``useModelWeights`` is set and ``modelWeightsWidth`` is negative. 

1223 """ 

1224 if not self.config.useModelWeights: 

1225 return 1.0 

1226 if self.config.modelWeightsWidth < 0: 

1227 raise ValueError("modelWeightsWidth must not be negative if useModelWeights is set") 

1228 convergeMask = dcrModels.mask.getPlaneBitMask(self.config.convergenceMaskPlanes) 

1229 convergeMaskPixels = dcrModels.mask[dcrBBox].array & convergeMask > 0 

1230 weights = np.zeros_like(dcrModels[0][dcrBBox].array) 

1231 weights[convergeMaskPixels] = 1. 

1232 weights = ndimage.filters.gaussian_filter(weights, self.config.modelWeightsWidth) 

1233 weights /= np.max(weights) 

1234 return weights 

1235 

1236 def applyModelWeights(self, modelImages, refImage, modelWeights): 

1237 """Smoothly replace model pixel values with those from a 

1238 reference at locations away from detected sources. 

1239 

1240 Parameters 

1241 ---------- 

1242 modelImages : `list` of `lsst.afw.image.Image` 

1243 The new DCR model images from the current iteration. 

1244 The values will be modified in place. 

1245 refImage : `lsst.afw.image.MaskedImage` 

1246 A reference image used to supply the default pixel values. 

1247 modelWeights : `numpy.ndarray` or `float` 

1248 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

1249 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

1250 """ 

1251 if self.config.useModelWeights: 

1252 for model in modelImages: 

1253 model.array *= modelWeights 

1254 model.array += refImage.array*(1. - modelWeights)/self.config.dcrNumSubfilters 

1255 

1256 def loadSubExposures(self, bbox, statsCtrl, warpRefList, imageScalerList, spanSetMaskList): 

1257 """Pre-load sub-regions of a list of exposures. 

1258 

1259 Parameters 

1260 ---------- 

1261 bbox : `lsst.geom.box.Box2I` 

1262 Sub-region to coadd 

1263 statsCtrl : `lsst.afw.math.StatisticsControl` 

1264 Statistics control object for coadd 

1265 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

1266 `lsst.daf.persistence.ButlerDataRef` 

1267 The data references to the input warped exposures. 

1268 imageScalerList : `list` of `lsst.pipe.task.ImageScaler` 

1269 The image scalars correct for the zero point of the exposures. 

1270 spanSetMaskList : `list` of `dict` containing spanSet lists, or None 

1271 Each element is dict with keys = mask plane name to add the spans to 

1272 

1273 Returns 

1274 ------- 

1275 subExposures : `dict` 

1276 The `dict` keys are the visit IDs, 

1277 and the values are `lsst.afw.image.ExposureF` 

1278 The pre-loaded exposures for the current subregion. 

1279 The variance plane contains weights, and not the variance 

1280 """ 

1281 tempExpName = self.getTempExpDatasetName(self.warpType) 

1282 zipIterables = zip(warpRefList, imageScalerList, spanSetMaskList) 

1283 subExposures = {} 

1284 for warpExpRef, imageScaler, altMaskSpans in zipIterables: 

1285 if isinstance(warpExpRef, DeferredDatasetHandle): 

1286 exposure = warpExpRef.get(parameters={'bbox': bbox}) 

1287 else: 

1288 exposure = warpExpRef.get(tempExpName + "_sub", bbox=bbox) 

1289 visit = warpExpRef.dataId["visit"] 

1290 if altMaskSpans is not None: 

1291 self.applyAltMaskPlanes(exposure.mask, altMaskSpans) 

1292 imageScaler.scaleMaskedImage(exposure.maskedImage) 

1293 # Note that the variance plane here is used to store weights, not the actual variance 

1294 exposure.variance.array[:, :] = 0. 

1295 # Set the weight of unmasked pixels to 1. 

1296 exposure.variance.array[(exposure.mask.array & statsCtrl.getAndMask()) == 0] = 1. 

1297 # Set the image value of masked pixels to zero. 

1298 # This eliminates needing the mask plane when stacking images in ``newModelFromResidual`` 

1299 exposure.image.array[(exposure.mask.array & statsCtrl.getAndMask()) > 0] = 0. 

1300 subExposures[visit] = exposure 

1301 return subExposures 

1302 

1303 def selectCoaddPsf(self, templateCoadd, warpRefList): 

1304 """Compute the PSF of the coadd from the exposures with the best seeing. 

1305 

1306 Parameters 

1307 ---------- 

1308 templateCoadd : `lsst.afw.image.ExposureF` 

1309 The initial coadd exposure before accounting for DCR. 

1310 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

1311 `lsst.daf.persistence.ButlerDataRef` 

1312 The data references to the input warped exposures. 

1313 

1314 Returns 

1315 ------- 

1316 psf : `lsst.meas.algorithms.CoaddPsf` 

1317 The average PSF of the input exposures with the best seeing. 

1318 """ 

1319 sigma2fwhm = 2.*np.sqrt(2.*np.log(2.)) 

1320 tempExpName = self.getTempExpDatasetName(self.warpType) 

1321 # Note: ``ccds`` is a `lsst.afw.table.ExposureCatalog` with one entry per ccd and per visit 

1322 # If there are multiple ccds, it will have that many times more elements than ``warpExpRef`` 

1323 ccds = templateCoadd.getInfo().getCoaddInputs().ccds 

1324 templatePsf = templateCoadd.getPsf() 

1325 # Just need a rough estimate; average positions are fine 

1326 templateAvgPos = templatePsf.getAveragePosition() 

1327 psfRefSize = templatePsf.computeShape(templateAvgPos).getDeterminantRadius()*sigma2fwhm 

1328 psfSizes = np.zeros(len(ccds)) 

1329 ccdVisits = np.array(ccds["visit"]) 

1330 for warpExpRef in warpRefList: 

1331 if isinstance(warpExpRef, DeferredDatasetHandle): 

1332 # Gen 3 API 

1333 psf = warpExpRef.get(component="psf") 

1334 else: 

1335 # Gen 2 API. Delete this when Gen 2 retired 

1336 psf = warpExpRef.get(tempExpName).getPsf() 

1337 visit = warpExpRef.dataId["visit"] 

1338 psfAvgPos = psf.getAveragePosition() 

1339 psfSize = psf.computeShape(psfAvgPos).getDeterminantRadius()*sigma2fwhm 

1340 psfSizes[ccdVisits == visit] = psfSize 

1341 # Note that the input PSFs include DCR, which should be absent from the DcrCoadd 

1342 # The selected PSFs are those that have a FWHM less than or equal to the smaller 

1343 # of the mean or median FWHM of the input exposures. 

1344 sizeThreshold = min(np.median(psfSizes), psfRefSize) 

1345 goodPsfs = psfSizes <= sizeThreshold 

1346 psf = measAlg.CoaddPsf(ccds[goodPsfs], templateCoadd.getWcs(), 

1347 self.config.coaddPsf.makeControl()) 

1348 return psf