Coverage for python/lsst/pipe/tasks/assembleCoadd.py: 13%
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
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
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#
22import os
23import copy
24import numpy
25import warnings
26import logging
27import lsst.pex.config as pexConfig
28import lsst.pex.exceptions as pexExceptions
29import lsst.geom as geom
30import lsst.afw.geom as afwGeom
31import lsst.afw.image as afwImage
32import lsst.afw.math as afwMath
33import lsst.afw.table as afwTable
34import lsst.afw.detection as afwDet
35import lsst.coadd.utils as coaddUtils
36import lsst.pipe.base as pipeBase
37import lsst.meas.algorithms as measAlg
38import lsstDebug
39import lsst.utils as utils
40from lsst.skymap import BaseSkyMap
41from .coaddBase import CoaddBaseTask, SelectDataIdContainer, makeSkyInfo, makeCoaddSuffix, reorderAndPadList
42from .interpImage import InterpImageTask
43from .scaleZeroPoint import ScaleZeroPointTask
44from .coaddHelpers import groupPatchExposures, getGroupDataRef
45from .scaleVariance import ScaleVarianceTask
46from .maskStreaks import MaskStreaksTask
47from .healSparseMapping import HealSparseInputMapTask
48from lsst.meas.algorithms import SourceDetectionTask, AccumulatorMeanStack
49from lsst.daf.butler import DeferredDatasetHandle
51__all__ = ["AssembleCoaddTask", "AssembleCoaddConnections", "AssembleCoaddConfig",
52 "SafeClipAssembleCoaddTask", "SafeClipAssembleCoaddConfig",
53 "CompareWarpAssembleCoaddTask", "CompareWarpAssembleCoaddConfig"]
55log = logging.getLogger(__name__.partition(".")[2])
58class AssembleCoaddConnections(pipeBase.PipelineTaskConnections,
59 dimensions=("tract", "patch", "band", "skymap"),
60 defaultTemplates={"inputCoaddName": "deep",
61 "outputCoaddName": "deep",
62 "warpType": "direct",
63 "warpTypeSuffix": ""}):
65 inputWarps = pipeBase.connectionTypes.Input(
66 doc=("Input list of warps to be assemebled i.e. stacked."
67 "WarpType (e.g. direct, psfMatched) is controlled by the warpType config parameter"),
68 name="{inputCoaddName}Coadd_{warpType}Warp",
69 storageClass="ExposureF",
70 dimensions=("tract", "patch", "skymap", "visit", "instrument"),
71 deferLoad=True,
72 multiple=True
73 )
74 skyMap = pipeBase.connectionTypes.Input(
75 doc="Input definition of geometry/bbox and projection/wcs for coadded exposures",
76 name=BaseSkyMap.SKYMAP_DATASET_TYPE_NAME,
77 storageClass="SkyMap",
78 dimensions=("skymap", ),
79 )
80 selectedVisits = pipeBase.connectionTypes.Input(
81 doc="Selected visits to be coadded.",
82 name="{outputCoaddName}Visits",
83 storageClass="StructuredDataDict",
84 dimensions=("instrument", "tract", "patch", "skymap", "band")
85 )
86 brightObjectMask = pipeBase.connectionTypes.PrerequisiteInput(
87 doc=("Input Bright Object Mask mask produced with external catalogs to be applied to the mask plane"
88 " BRIGHT_OBJECT."),
89 name="brightObjectMask",
90 storageClass="ObjectMaskCatalog",
91 dimensions=("tract", "patch", "skymap", "band"),
92 )
93 coaddExposure = pipeBase.connectionTypes.Output(
94 doc="Output coadded exposure, produced by stacking input warps",
95 name="{outputCoaddName}Coadd{warpTypeSuffix}",
96 storageClass="ExposureF",
97 dimensions=("tract", "patch", "skymap", "band"),
98 )
99 nImage = pipeBase.connectionTypes.Output(
100 doc="Output image of number of input images per pixel",
101 name="{outputCoaddName}Coadd_nImage",
102 storageClass="ImageU",
103 dimensions=("tract", "patch", "skymap", "band"),
104 )
105 inputMap = pipeBase.connectionTypes.Output(
106 doc="Output healsparse map of input images",
107 name="{outputCoaddName}Coadd_inputMap",
108 storageClass="HealSparseMap",
109 dimensions=("tract", "patch", "skymap", "band"),
110 )
112 def __init__(self, *, config=None):
113 super().__init__(config=config)
115 # Override the connection's name template with config to replicate Gen2 behavior
116 # This duplicates some of the logic in the base class, due to wanting Gen2 and
117 # Gen3 configs to stay in sync. This should be removed when gen2 is deprecated
118 templateValues = {name: getattr(config.connections, name) for name in self.defaultTemplates}
119 templateValues['warpType'] = config.warpType
120 templateValues['warpTypeSuffix'] = makeCoaddSuffix(config.warpType)
121 self._nameOverrides = {name: getattr(config.connections, name).format(**templateValues)
122 for name in self.allConnections}
123 self._typeNameToVarName = {v: k for k, v in self._nameOverrides.items()}
124 # End code to remove after deprecation
126 if not config.doMaskBrightObjects:
127 self.prerequisiteInputs.remove("brightObjectMask")
129 if not config.doSelectVisits:
130 self.inputs.remove("selectedVisits")
132 if not config.doNImage:
133 self.outputs.remove("nImage")
135 if not self.config.doInputMap:
136 self.outputs.remove("inputMap")
139class AssembleCoaddConfig(CoaddBaseTask.ConfigClass, pipeBase.PipelineTaskConfig,
140 pipelineConnections=AssembleCoaddConnections):
141 """Configuration parameters for the `AssembleCoaddTask`.
143 Notes
144 -----
145 The `doMaskBrightObjects` and `brightObjectMaskName` configuration options
146 only set the bitplane config.brightObjectMaskName. To make this useful you
147 *must* also configure the flags.pixel algorithm, for example by adding
149 .. code-block:: none
151 config.measurement.plugins["base_PixelFlags"].masksFpCenter.append("BRIGHT_OBJECT")
152 config.measurement.plugins["base_PixelFlags"].masksFpAnywhere.append("BRIGHT_OBJECT")
154 to your measureCoaddSources.py and forcedPhotCoadd.py config overrides.
155 """
156 warpType = pexConfig.Field(
157 doc="Warp name: one of 'direct' or 'psfMatched'",
158 dtype=str,
159 default="direct",
160 )
161 subregionSize = pexConfig.ListField(
162 dtype=int,
163 doc="Width, height of stack subregion size; "
164 "make small enough that a full stack of images will fit into memory at once.",
165 length=2,
166 default=(2000, 2000),
167 )
168 statistic = pexConfig.Field(
169 dtype=str,
170 doc="Main stacking statistic for aggregating over the epochs.",
171 default="MEANCLIP",
172 )
173 doOnlineForMean = pexConfig.Field(
174 dtype=bool,
175 doc="Perform online coaddition when statistic=\"MEAN\" to save memory?",
176 default=False,
177 )
178 doSigmaClip = pexConfig.Field(
179 dtype=bool,
180 doc="Perform sigma clipped outlier rejection with MEANCLIP statistic? (DEPRECATED)",
181 default=False,
182 )
183 sigmaClip = pexConfig.Field(
184 dtype=float,
185 doc="Sigma for outlier rejection; ignored if non-clipping statistic selected.",
186 default=3.0,
187 )
188 clipIter = pexConfig.Field(
189 dtype=int,
190 doc="Number of iterations of outlier rejection; ignored if non-clipping statistic selected.",
191 default=2,
192 )
193 calcErrorFromInputVariance = pexConfig.Field(
194 dtype=bool,
195 doc="Calculate coadd variance from input variance by stacking statistic."
196 "Passed to StatisticsControl.setCalcErrorFromInputVariance()",
197 default=True,
198 )
199 scaleZeroPoint = pexConfig.ConfigurableField(
200 target=ScaleZeroPointTask,
201 doc="Task to adjust the photometric zero point of the coadd temp exposures",
202 )
203 doInterp = pexConfig.Field(
204 doc="Interpolate over NaN pixels? Also extrapolate, if necessary, but the results are ugly.",
205 dtype=bool,
206 default=True,
207 )
208 interpImage = pexConfig.ConfigurableField(
209 target=InterpImageTask,
210 doc="Task to interpolate (and extrapolate) over NaN pixels",
211 )
212 doWrite = pexConfig.Field(
213 doc="Persist coadd?",
214 dtype=bool,
215 default=True,
216 )
217 doNImage = pexConfig.Field(
218 doc="Create image of number of contributing exposures for each pixel",
219 dtype=bool,
220 default=False,
221 )
222 doUsePsfMatchedPolygons = pexConfig.Field(
223 doc="Use ValidPolygons from shrunk Psf-Matched Calexps? Should be set to True by CompareWarp only.",
224 dtype=bool,
225 default=False,
226 )
227 maskPropagationThresholds = pexConfig.DictField(
228 keytype=str,
229 itemtype=float,
230 doc=("Threshold (in fractional weight) of rejection at which we propagate a mask plane to "
231 "the coadd; that is, we set the mask bit on the coadd if the fraction the rejected frames "
232 "would have contributed exceeds this value."),
233 default={"SAT": 0.1},
234 )
235 removeMaskPlanes = pexConfig.ListField(dtype=str, default=["NOT_DEBLENDED"],
236 doc="Mask planes to remove before coadding")
237 doMaskBrightObjects = pexConfig.Field(dtype=bool, default=False,
238 doc="Set mask and flag bits for bright objects?")
239 brightObjectMaskName = pexConfig.Field(dtype=str, default="BRIGHT_OBJECT",
240 doc="Name of mask bit used for bright objects")
241 coaddPsf = pexConfig.ConfigField(
242 doc="Configuration for CoaddPsf",
243 dtype=measAlg.CoaddPsfConfig,
244 )
245 doAttachTransmissionCurve = pexConfig.Field(
246 dtype=bool, default=False, optional=False,
247 doc=("Attach a piecewise TransmissionCurve for the coadd? "
248 "(requires all input Exposures to have TransmissionCurves).")
249 )
250 hasFakes = pexConfig.Field(
251 dtype=bool,
252 default=False,
253 doc="Should be set to True if fake sources have been inserted into the input data."
254 )
255 doSelectVisits = pexConfig.Field(
256 doc="Coadd only visits selected by a SelectVisitsTask",
257 dtype=bool,
258 default=False,
259 )
260 doInputMap = pexConfig.Field(
261 doc="Create a bitwise map of coadd inputs",
262 dtype=bool,
263 default=False,
264 )
265 inputMapper = pexConfig.ConfigurableField(
266 doc="Input map creation subtask.",
267 target=HealSparseInputMapTask,
268 )
270 def setDefaults(self):
271 super().setDefaults()
272 self.badMaskPlanes = ["NO_DATA", "BAD", "SAT", "EDGE"]
274 def validate(self):
275 super().validate()
276 if self.doPsfMatch:
277 # Backwards compatibility.
278 # Configs do not have loggers
279 log.warning("Config doPsfMatch deprecated. Setting warpType='psfMatched'")
280 self.warpType = 'psfMatched'
281 if self.doSigmaClip and self.statistic != "MEANCLIP":
282 log.warning('doSigmaClip deprecated. To replicate behavior, setting statistic to "MEANCLIP"')
283 self.statistic = "MEANCLIP"
284 if self.doInterp and self.statistic not in ['MEAN', 'MEDIAN', 'MEANCLIP', 'VARIANCE', 'VARIANCECLIP']:
285 raise ValueError("Must set doInterp=False for statistic=%s, which does not "
286 "compute and set a non-zero coadd variance estimate." % (self.statistic))
288 unstackableStats = ['NOTHING', 'ERROR', 'ORMASK']
289 if not hasattr(afwMath.Property, self.statistic) or self.statistic in unstackableStats:
290 stackableStats = [str(k) for k in afwMath.Property.__members__.keys()
291 if str(k) not in unstackableStats]
292 raise ValueError("statistic %s is not allowed. Please choose one of %s."
293 % (self.statistic, stackableStats))
296class AssembleCoaddTask(CoaddBaseTask, pipeBase.PipelineTask):
297 """Assemble a coadded image from a set of warps (coadded temporary exposures).
299 We want to assemble a coadded image from a set of Warps (also called
300 coadded temporary exposures or ``coaddTempExps``).
301 Each input Warp covers a patch on the sky and corresponds to a single
302 run/visit/exposure of the covered patch. We provide the task with a list
303 of Warps (``selectDataList``) from which it selects Warps that cover the
304 specified patch (pointed at by ``dataRef``).
305 Each Warp that goes into a coadd will typically have an independent
306 photometric zero-point. Therefore, we must scale each Warp to set it to
307 a common photometric zeropoint. WarpType may be one of 'direct' or
308 'psfMatched', and the boolean configs `config.makeDirect` and
309 `config.makePsfMatched` set which of the warp types will be coadded.
310 The coadd is computed as a mean with optional outlier rejection.
311 Criteria for outlier rejection are set in `AssembleCoaddConfig`.
312 Finally, Warps can have bad 'NaN' pixels which received no input from the
313 source calExps. We interpolate over these bad (NaN) pixels.
315 `AssembleCoaddTask` uses several sub-tasks. These are
317 - `ScaleZeroPointTask`
318 - create and use an ``imageScaler`` object to scale the photometric zeropoint for each Warp
319 - `InterpImageTask`
320 - interpolate across bad pixels (NaN) in the final coadd
322 You can retarget these subtasks if you wish.
324 Notes
325 -----
326 The `lsst.pipe.base.cmdLineTask.CmdLineTask` interface supports a
327 flag ``-d`` to import ``debug.py`` from your ``PYTHONPATH``; see
328 `baseDebug` for more about ``debug.py`` files. `AssembleCoaddTask` has
329 no debug variables of its own. Some of the subtasks may support debug
330 variables. See the documentation for the subtasks for further information.
332 Examples
333 --------
334 `AssembleCoaddTask` assembles a set of warped images into a coadded image.
335 The `AssembleCoaddTask` can be invoked by running ``assembleCoadd.py``
336 with the flag '--legacyCoadd'. Usage of assembleCoadd.py expects two
337 inputs: a data reference to the tract patch and filter to be coadded, and
338 a list of Warps to attempt to coadd. These are specified using ``--id`` and
339 ``--selectId``, respectively:
341 .. code-block:: none
343 --id = [KEY=VALUE1[^VALUE2[^VALUE3...] [KEY=VALUE1[^VALUE2[^VALUE3...] ...]]
344 --selectId [KEY=VALUE1[^VALUE2[^VALUE3...] [KEY=VALUE1[^VALUE2[^VALUE3...] ...]]
346 Only the Warps that cover the specified tract and patch will be coadded.
347 A list of the available optional arguments can be obtained by calling
348 ``assembleCoadd.py`` with the ``--help`` command line argument:
350 .. code-block:: none
352 assembleCoadd.py --help
354 To demonstrate usage of the `AssembleCoaddTask` in the larger context of
355 multi-band processing, we will generate the HSC-I & -R band coadds from
356 HSC engineering test data provided in the ``ci_hsc`` package. To begin,
357 assuming that the lsst stack has been already set up, we must set up the
358 obs_subaru and ``ci_hsc`` packages. This defines the environment variable
359 ``$CI_HSC_DIR`` and points at the location of the package. The raw HSC
360 data live in the ``$CI_HSC_DIR/raw directory``. To begin assembling the
361 coadds, we must first
363 - processCcd
364 - process the individual ccds in $CI_HSC_RAW to produce calibrated exposures
365 - makeSkyMap
366 - create a skymap that covers the area of the sky present in the raw exposures
367 - makeCoaddTempExp
368 - warp the individual calibrated exposures to the tangent plane of the coadd
370 We can perform all of these steps by running
372 .. code-block:: none
374 $CI_HSC_DIR scons warp-903986 warp-904014 warp-903990 warp-904010 warp-903988
376 This will produce warped exposures for each visit. To coadd the warped
377 data, we call assembleCoadd.py as follows:
379 .. code-block:: none
381 assembleCoadd.py --legacyCoadd $CI_HSC_DIR/DATA --id patch=5,4 tract=0 filter=HSC-I \
382 --selectId visit=903986 ccd=16 --selectId visit=903986 ccd=22 --selectId visit=903986 ccd=23 \
383 --selectId visit=903986 ccd=100 --selectId visit=904014 ccd=1 --selectId visit=904014 ccd=6 \
384 --selectId visit=904014 ccd=12 --selectId visit=903990 ccd=18 --selectId visit=903990 ccd=25 \
385 --selectId visit=904010 ccd=4 --selectId visit=904010 ccd=10 --selectId visit=904010 ccd=100 \
386 --selectId visit=903988 ccd=16 --selectId visit=903988 ccd=17 --selectId visit=903988 ccd=23 \
387 --selectId visit=903988 ccd=24
389 that will process the HSC-I band data. The results are written in
390 ``$CI_HSC_DIR/DATA/deepCoadd-results/HSC-I``.
392 You may also choose to run:
394 .. code-block:: none
396 scons warp-903334 warp-903336 warp-903338 warp-903342 warp-903344 warp-903346
397 assembleCoadd.py --legacyCoadd $CI_HSC_DIR/DATA --id patch=5,4 tract=0 filter=HSC-R \
398 --selectId visit=903334 ccd=16 --selectId visit=903334 ccd=22 --selectId visit=903334 ccd=23 \
399 --selectId visit=903334 ccd=100 --selectId visit=903336 ccd=17 --selectId visit=903336 ccd=24 \
400 --selectId visit=903338 ccd=18 --selectId visit=903338 ccd=25 --selectId visit=903342 ccd=4 \
401 --selectId visit=903342 ccd=10 --selectId visit=903342 ccd=100 --selectId visit=903344 ccd=0 \
402 --selectId visit=903344 ccd=5 --selectId visit=903344 ccd=11 --selectId visit=903346 ccd=1 \
403 --selectId visit=903346 ccd=6 --selectId visit=903346 ccd=12
405 to generate the coadd for the HSC-R band if you are interested in
406 following multiBand Coadd processing as discussed in `pipeTasks_multiBand`
407 (but note that normally, one would use the `SafeClipAssembleCoaddTask`
408 rather than `AssembleCoaddTask` to make the coadd.
409 """
410 ConfigClass = AssembleCoaddConfig
411 _DefaultName = "assembleCoadd"
413 def __init__(self, *args, **kwargs):
414 # TODO: DM-17415 better way to handle previously allowed passed args e.g.`AssembleCoaddTask(config)`
415 if args:
416 argNames = ["config", "name", "parentTask", "log"]
417 kwargs.update({k: v for k, v in zip(argNames, args)})
418 warnings.warn("AssembleCoadd received positional args, and casting them as kwargs: %s. "
419 "PipelineTask will not take positional args" % argNames, FutureWarning)
421 super().__init__(**kwargs)
422 self.makeSubtask("interpImage")
423 self.makeSubtask("scaleZeroPoint")
425 if self.config.doMaskBrightObjects:
426 mask = afwImage.Mask()
427 try:
428 self.brightObjectBitmask = 1 << mask.addMaskPlane(self.config.brightObjectMaskName)
429 except pexExceptions.LsstCppException:
430 raise RuntimeError("Unable to define mask plane for bright objects; planes used are %s" %
431 mask.getMaskPlaneDict().keys())
432 del mask
434 if self.config.doInputMap:
435 self.makeSubtask("inputMapper")
437 self.warpType = self.config.warpType
439 @utils.inheritDoc(pipeBase.PipelineTask)
440 def runQuantum(self, butlerQC, inputRefs, outputRefs):
441 # Docstring to be formatted with info from PipelineTask.runQuantum
442 """
443 Notes
444 -----
445 Assemble a coadd from a set of Warps.
447 PipelineTask (Gen3) entry point to Coadd a set of Warps.
448 Analogous to `runDataRef`, it prepares all the data products to be
449 passed to `run`, and processes the results before returning a struct
450 of results to be written out. AssembleCoadd cannot fit all Warps in memory.
451 Therefore, its inputs are accessed subregion by subregion
452 by the Gen3 `DeferredDatasetHandle` that is analagous to the Gen2
453 `lsst.daf.persistence.ButlerDataRef`. Any updates to this method should
454 correspond to an update in `runDataRef` while both entry points
455 are used.
456 """
457 inputData = butlerQC.get(inputRefs)
459 # Construct skyInfo expected by run
460 # Do not remove skyMap from inputData in case makeSupplementaryDataGen3 needs it
461 skyMap = inputData["skyMap"]
462 outputDataId = butlerQC.quantum.dataId
464 inputData['skyInfo'] = makeSkyInfo(skyMap,
465 tractId=outputDataId['tract'],
466 patchId=outputDataId['patch'])
468 if self.config.doSelectVisits:
469 warpRefList = self.filterWarps(inputData['inputWarps'], inputData['selectedVisits'])
470 else:
471 warpRefList = inputData['inputWarps']
473 # Perform same middle steps as `runDataRef` does
474 inputs = self.prepareInputs(warpRefList)
475 self.log.info("Found %d %s", len(inputs.tempExpRefList),
476 self.getTempExpDatasetName(self.warpType))
477 if len(inputs.tempExpRefList) == 0:
478 raise pipeBase.NoWorkFound("No coadd temporary exposures found")
480 supplementaryData = self.makeSupplementaryDataGen3(butlerQC, inputRefs, outputRefs)
481 retStruct = self.run(inputData['skyInfo'], inputs.tempExpRefList, inputs.imageScalerList,
482 inputs.weightList, supplementaryData=supplementaryData)
484 inputData.setdefault('brightObjectMask', None)
485 self.processResults(retStruct.coaddExposure, inputData['brightObjectMask'], outputDataId)
487 if self.config.doWrite:
488 butlerQC.put(retStruct, outputRefs)
489 return retStruct
491 @pipeBase.timeMethod
492 def runDataRef(self, dataRef, selectDataList=None, warpRefList=None):
493 """Assemble a coadd from a set of Warps.
495 Pipebase.CmdlineTask entry point to Coadd a set of Warps.
496 Compute weights to be applied to each Warp and
497 find scalings to match the photometric zeropoint to a reference Warp.
498 Assemble the Warps using `run`. Interpolate over NaNs and
499 optionally write the coadd to disk. Return the coadded exposure.
501 Parameters
502 ----------
503 dataRef : `lsst.daf.persistence.butlerSubset.ButlerDataRef`
504 Data reference defining the patch for coaddition and the
505 reference Warp (if ``config.autoReference=False``).
506 Used to access the following data products:
507 - ``self.config.coaddName + "Coadd_skyMap"``
508 - ``self.config.coaddName + "Coadd_ + <warpType> + "Warp"`` (optionally)
509 - ``self.config.coaddName + "Coadd"``
510 selectDataList : `list`
511 List of data references to Calexps. Data to be coadded will be
512 selected from this list based on overlap with the patch defined
513 by dataRef, grouped by visit, and converted to a list of data
514 references to warps.
515 warpRefList : `list`
516 List of data references to Warps to be coadded.
517 Note: `warpRefList` is just the new name for `tempExpRefList`.
519 Returns
520 -------
521 retStruct : `lsst.pipe.base.Struct`
522 Result struct with components:
524 - ``coaddExposure``: coadded exposure (``Exposure``).
525 - ``nImage``: exposure count image (``Image``).
526 """
527 if selectDataList and warpRefList:
528 raise RuntimeError("runDataRef received both a selectDataList and warpRefList, "
529 "and which to use is ambiguous. Please pass only one.")
531 skyInfo = self.getSkyInfo(dataRef)
532 if warpRefList is None:
533 calExpRefList = self.selectExposures(dataRef, skyInfo, selectDataList=selectDataList)
534 if len(calExpRefList) == 0:
535 self.log.warning("No exposures to coadd")
536 return
537 self.log.info("Coadding %d exposures", len(calExpRefList))
539 warpRefList = self.getTempExpRefList(dataRef, calExpRefList)
541 inputData = self.prepareInputs(warpRefList)
542 self.log.info("Found %d %s", len(inputData.tempExpRefList),
543 self.getTempExpDatasetName(self.warpType))
544 if len(inputData.tempExpRefList) == 0:
545 self.log.warning("No coadd temporary exposures found")
546 return
548 supplementaryData = self.makeSupplementaryData(dataRef, warpRefList=inputData.tempExpRefList)
550 retStruct = self.run(skyInfo, inputData.tempExpRefList, inputData.imageScalerList,
551 inputData.weightList, supplementaryData=supplementaryData)
553 brightObjects = self.readBrightObjectMasks(dataRef) if self.config.doMaskBrightObjects else None
554 self.processResults(retStruct.coaddExposure, brightObjectMasks=brightObjects, dataId=dataRef.dataId)
556 if self.config.doWrite:
557 if self.getCoaddDatasetName(self.warpType) == "deepCoadd" and self.config.hasFakes:
558 coaddDatasetName = "fakes_" + self.getCoaddDatasetName(self.warpType)
559 else:
560 coaddDatasetName = self.getCoaddDatasetName(self.warpType)
561 self.log.info("Persisting %s", coaddDatasetName)
562 dataRef.put(retStruct.coaddExposure, coaddDatasetName)
563 if self.config.doNImage and retStruct.nImage is not None:
564 dataRef.put(retStruct.nImage, self.getCoaddDatasetName(self.warpType) + '_nImage')
566 return retStruct
568 def processResults(self, coaddExposure, brightObjectMasks=None, dataId=None):
569 """Interpolate over missing data and mask bright stars.
571 Parameters
572 ----------
573 coaddExposure : `lsst.afw.image.Exposure`
574 The coadded exposure to process.
575 dataRef : `lsst.daf.persistence.ButlerDataRef`
576 Butler data reference for supplementary data.
577 """
578 if self.config.doInterp:
579 self.interpImage.run(coaddExposure.getMaskedImage(), planeName="NO_DATA")
580 # The variance must be positive; work around for DM-3201.
581 varArray = coaddExposure.variance.array
582 with numpy.errstate(invalid="ignore"):
583 varArray[:] = numpy.where(varArray > 0, varArray, numpy.inf)
585 if self.config.doMaskBrightObjects:
586 self.setBrightObjectMasks(coaddExposure, brightObjectMasks, dataId)
588 def makeSupplementaryData(self, dataRef, selectDataList=None, warpRefList=None):
589 """Make additional inputs to run() specific to subclasses (Gen2)
591 Duplicates interface of `runDataRef` method
592 Available to be implemented by subclasses only if they need the
593 coadd dataRef for performing preliminary processing before
594 assembling the coadd.
596 Parameters
597 ----------
598 dataRef : `lsst.daf.persistence.ButlerDataRef`
599 Butler data reference for supplementary data.
600 selectDataList : `list` (optional)
601 Optional List of data references to Calexps.
602 warpRefList : `list` (optional)
603 Optional List of data references to Warps.
604 """
605 return pipeBase.Struct()
607 def makeSupplementaryDataGen3(self, butlerQC, inputRefs, outputRefs):
608 """Make additional inputs to run() specific to subclasses (Gen3)
610 Duplicates interface of `runQuantum` method.
611 Available to be implemented by subclasses only if they need the
612 coadd dataRef for performing preliminary processing before
613 assembling the coadd.
615 Parameters
616 ----------
617 butlerQC : `lsst.pipe.base.ButlerQuantumContext`
618 Gen3 Butler object for fetching additional data products before
619 running the Task specialized for quantum being processed
620 inputRefs : `lsst.pipe.base.InputQuantizedConnection`
621 Attributes are the names of the connections describing input dataset types.
622 Values are DatasetRefs that task consumes for corresponding dataset type.
623 DataIds are guaranteed to match data objects in ``inputData``.
624 outputRefs : `lsst.pipe.base.OutputQuantizedConnection`
625 Attributes are the names of the connections describing output dataset types.
626 Values are DatasetRefs that task is to produce
627 for corresponding dataset type.
628 """
629 return pipeBase.Struct()
631 def getTempExpRefList(self, patchRef, calExpRefList):
632 """Generate list data references corresponding to warped exposures
633 that lie within the patch to be coadded.
635 Parameters
636 ----------
637 patchRef : `dataRef`
638 Data reference for patch.
639 calExpRefList : `list`
640 List of data references for input calexps.
642 Returns
643 -------
644 tempExpRefList : `list`
645 List of Warp/CoaddTempExp data references.
646 """
647 butler = patchRef.getButler()
648 groupData = groupPatchExposures(patchRef, calExpRefList, self.getCoaddDatasetName(self.warpType),
649 self.getTempExpDatasetName(self.warpType))
650 tempExpRefList = [getGroupDataRef(butler, self.getTempExpDatasetName(self.warpType),
651 g, groupData.keys) for
652 g in groupData.groups.keys()]
653 return tempExpRefList
655 def prepareInputs(self, refList):
656 """Prepare the input warps for coaddition by measuring the weight for
657 each warp and the scaling for the photometric zero point.
659 Each Warp has its own photometric zeropoint and background variance.
660 Before coadding these Warps together, compute a scale factor to
661 normalize the photometric zeropoint and compute the weight for each Warp.
663 Parameters
664 ----------
665 refList : `list`
666 List of data references to tempExp
668 Returns
669 -------
670 result : `lsst.pipe.base.Struct`
671 Result struct with components:
673 - ``tempExprefList``: `list` of data references to tempExp.
674 - ``weightList``: `list` of weightings.
675 - ``imageScalerList``: `list` of image scalers.
676 """
677 statsCtrl = afwMath.StatisticsControl()
678 statsCtrl.setNumSigmaClip(self.config.sigmaClip)
679 statsCtrl.setNumIter(self.config.clipIter)
680 statsCtrl.setAndMask(self.getBadPixelMask())
681 statsCtrl.setNanSafe(True)
682 # compute tempExpRefList: a list of tempExpRef that actually exist
683 # and weightList: a list of the weight of the associated coadd tempExp
684 # and imageScalerList: a list of scale factors for the associated coadd tempExp
685 tempExpRefList = []
686 weightList = []
687 imageScalerList = []
688 tempExpName = self.getTempExpDatasetName(self.warpType)
689 for tempExpRef in refList:
690 # Gen3's DeferredDatasetHandles are guaranteed to exist and
691 # therefore have no datasetExists() method
692 if not isinstance(tempExpRef, DeferredDatasetHandle):
693 if not tempExpRef.datasetExists(tempExpName):
694 self.log.warning("Could not find %s %s; skipping it", tempExpName, tempExpRef.dataId)
695 continue
697 tempExp = tempExpRef.get(datasetType=tempExpName, immediate=True)
698 # Ignore any input warp that is empty of data
699 if numpy.isnan(tempExp.image.array).all():
700 continue
701 maskedImage = tempExp.getMaskedImage()
702 imageScaler = self.scaleZeroPoint.computeImageScaler(
703 exposure=tempExp,
704 dataRef=tempExpRef,
705 )
706 try:
707 imageScaler.scaleMaskedImage(maskedImage)
708 except Exception as e:
709 self.log.warning("Scaling failed for %s (skipping it): %s", tempExpRef.dataId, e)
710 continue
711 statObj = afwMath.makeStatistics(maskedImage.getVariance(), maskedImage.getMask(),
712 afwMath.MEANCLIP, statsCtrl)
713 meanVar, meanVarErr = statObj.getResult(afwMath.MEANCLIP)
714 weight = 1.0 / float(meanVar)
715 if not numpy.isfinite(weight):
716 self.log.warning("Non-finite weight for %s: skipping", tempExpRef.dataId)
717 continue
718 self.log.info("Weight of %s %s = %0.3f", tempExpName, tempExpRef.dataId, weight)
720 del maskedImage
721 del tempExp
723 tempExpRefList.append(tempExpRef)
724 weightList.append(weight)
725 imageScalerList.append(imageScaler)
727 return pipeBase.Struct(tempExpRefList=tempExpRefList, weightList=weightList,
728 imageScalerList=imageScalerList)
730 def prepareStats(self, mask=None):
731 """Prepare the statistics for coadding images.
733 Parameters
734 ----------
735 mask : `int`, optional
736 Bit mask value to exclude from coaddition.
738 Returns
739 -------
740 stats : `lsst.pipe.base.Struct`
741 Statistics structure with the following fields:
743 - ``statsCtrl``: Statistics control object for coadd
744 (`lsst.afw.math.StatisticsControl`)
745 - ``statsFlags``: Statistic for coadd (`lsst.afw.math.Property`)
746 """
747 if mask is None:
748 mask = self.getBadPixelMask()
749 statsCtrl = afwMath.StatisticsControl()
750 statsCtrl.setNumSigmaClip(self.config.sigmaClip)
751 statsCtrl.setNumIter(self.config.clipIter)
752 statsCtrl.setAndMask(mask)
753 statsCtrl.setNanSafe(True)
754 statsCtrl.setWeighted(True)
755 statsCtrl.setCalcErrorFromInputVariance(self.config.calcErrorFromInputVariance)
756 for plane, threshold in self.config.maskPropagationThresholds.items():
757 bit = afwImage.Mask.getMaskPlane(plane)
758 statsCtrl.setMaskPropagationThreshold(bit, threshold)
759 statsFlags = afwMath.stringToStatisticsProperty(self.config.statistic)
760 return pipeBase.Struct(ctrl=statsCtrl, flags=statsFlags)
762 @pipeBase.timeMethod
763 def run(self, skyInfo, tempExpRefList, imageScalerList, weightList,
764 altMaskList=None, mask=None, supplementaryData=None):
765 """Assemble a coadd from input warps
767 Assemble the coadd using the provided list of coaddTempExps. Since
768 the full coadd covers a patch (a large area), the assembly is
769 performed over small areas on the image at a time in order to
770 conserve memory usage. Iterate over subregions within the outer
771 bbox of the patch using `assembleSubregion` to stack the corresponding
772 subregions from the coaddTempExps with the statistic specified.
773 Set the edge bits the coadd mask based on the weight map.
775 Parameters
776 ----------
777 skyInfo : `lsst.pipe.base.Struct`
778 Struct with geometric information about the patch.
779 tempExpRefList : `list`
780 List of data references to Warps (previously called CoaddTempExps).
781 imageScalerList : `list`
782 List of image scalers.
783 weightList : `list`
784 List of weights
785 altMaskList : `list`, optional
786 List of alternate masks to use rather than those stored with
787 tempExp.
788 mask : `int`, optional
789 Bit mask value to exclude from coaddition.
790 supplementaryData : lsst.pipe.base.Struct, optional
791 Struct with additional data products needed to assemble coadd.
792 Only used by subclasses that implement `makeSupplementaryData`
793 and override `run`.
795 Returns
796 -------
797 result : `lsst.pipe.base.Struct`
798 Result struct with components:
800 - ``coaddExposure``: coadded exposure (``lsst.afw.image.Exposure``).
801 - ``nImage``: exposure count image (``lsst.afw.image.Image``), if requested.
802 - ``inputMap``: bit-wise map of inputs, if requested.
803 - ``warpRefList``: input list of refs to the warps (
804 ``lsst.daf.butler.DeferredDatasetHandle`` or
805 ``lsst.daf.persistence.ButlerDataRef``)
806 (unmodified)
807 - ``imageScalerList``: input list of image scalers (unmodified)
808 - ``weightList``: input list of weights (unmodified)
809 """
810 tempExpName = self.getTempExpDatasetName(self.warpType)
811 self.log.info("Assembling %s %s", len(tempExpRefList), tempExpName)
812 stats = self.prepareStats(mask=mask)
814 if altMaskList is None:
815 altMaskList = [None]*len(tempExpRefList)
817 coaddExposure = afwImage.ExposureF(skyInfo.bbox, skyInfo.wcs)
818 coaddExposure.setPhotoCalib(self.scaleZeroPoint.getPhotoCalib())
819 coaddExposure.getInfo().setCoaddInputs(self.inputRecorder.makeCoaddInputs())
820 self.assembleMetadata(coaddExposure, tempExpRefList, weightList)
821 coaddMaskedImage = coaddExposure.getMaskedImage()
822 subregionSizeArr = self.config.subregionSize
823 subregionSize = geom.Extent2I(subregionSizeArr[0], subregionSizeArr[1])
824 # if nImage is requested, create a zero one which can be passed to assembleSubregion
825 if self.config.doNImage:
826 nImage = afwImage.ImageU(skyInfo.bbox)
827 else:
828 nImage = None
829 # If inputMap is requested, create the initial version that can be masked in
830 # assembleSubregion.
831 if self.config.doInputMap:
832 self.inputMapper.build_ccd_input_map(skyInfo.bbox,
833 skyInfo.wcs,
834 coaddExposure.getInfo().getCoaddInputs().ccds)
836 if self.config.doOnlineForMean and self.config.statistic == "MEAN":
837 try:
838 self.assembleOnlineMeanCoadd(coaddExposure, tempExpRefList, imageScalerList,
839 weightList, altMaskList, stats.ctrl,
840 nImage=nImage)
841 except Exception as e:
842 self.log.fatal("Cannot compute online coadd %s", e)
843 else:
844 for subBBox in self._subBBoxIter(skyInfo.bbox, subregionSize):
845 try:
846 self.assembleSubregion(coaddExposure, subBBox, tempExpRefList, imageScalerList,
847 weightList, altMaskList, stats.flags, stats.ctrl,
848 nImage=nImage)
849 except Exception as e:
850 self.log.fatal("Cannot compute coadd %s: %s", subBBox, e)
852 # If inputMap is requested, we must finalize the map after the accumulation.
853 if self.config.doInputMap:
854 self.inputMapper.finalize_ccd_input_map_mask()
855 inputMap = self.inputMapper.ccd_input_map
856 else:
857 inputMap = None
859 self.setInexactPsf(coaddMaskedImage.getMask())
860 # Despite the name, the following doesn't really deal with "EDGE" pixels: it identifies
861 # pixels that didn't receive any unmasked inputs (as occurs around the edge of the field).
862 coaddUtils.setCoaddEdgeBits(coaddMaskedImage.getMask(), coaddMaskedImage.getVariance())
863 return pipeBase.Struct(coaddExposure=coaddExposure, nImage=nImage,
864 warpRefList=tempExpRefList, imageScalerList=imageScalerList,
865 weightList=weightList, inputMap=inputMap)
867 def assembleMetadata(self, coaddExposure, tempExpRefList, weightList):
868 """Set the metadata for the coadd.
870 This basic implementation sets the filter from the first input.
872 Parameters
873 ----------
874 coaddExposure : `lsst.afw.image.Exposure`
875 The target exposure for the coadd.
876 tempExpRefList : `list`
877 List of data references to tempExp.
878 weightList : `list`
879 List of weights.
880 """
881 assert len(tempExpRefList) == len(weightList), "Length mismatch"
882 tempExpName = self.getTempExpDatasetName(self.warpType)
883 # We load a single pixel of each coaddTempExp, because we just want to get at the metadata
884 # (and we need more than just the PropertySet that contains the header), which is not possible
885 # with the current butler (see #2777).
886 bbox = geom.Box2I(coaddExposure.getBBox().getMin(), geom.Extent2I(1, 1))
888 if isinstance(tempExpRefList[0], DeferredDatasetHandle):
889 # Gen 3 API
890 tempExpList = [tempExpRef.get(parameters={'bbox': bbox}) for tempExpRef in tempExpRefList]
891 else:
892 # Gen 2 API. Delete this when Gen 2 retired
893 tempExpList = [tempExpRef.get(tempExpName + "_sub", bbox=bbox, immediate=True)
894 for tempExpRef in tempExpRefList]
895 numCcds = sum(len(tempExp.getInfo().getCoaddInputs().ccds) for tempExp in tempExpList)
897 # Set the coadd FilterLabel to the band of the first input exposure:
898 # Coadds are calibrated, so the physical label is now meaningless.
899 coaddExposure.setFilterLabel(afwImage.FilterLabel(tempExpList[0].getFilterLabel().bandLabel))
900 coaddInputs = coaddExposure.getInfo().getCoaddInputs()
901 coaddInputs.ccds.reserve(numCcds)
902 coaddInputs.visits.reserve(len(tempExpList))
904 for tempExp, weight in zip(tempExpList, weightList):
905 self.inputRecorder.addVisitToCoadd(coaddInputs, tempExp, weight)
907 if self.config.doUsePsfMatchedPolygons:
908 self.shrinkValidPolygons(coaddInputs)
910 coaddInputs.visits.sort()
911 if self.warpType == "psfMatched":
912 # The modelPsf BBox for a psfMatchedWarp/coaddTempExp was dynamically defined by
913 # ModelPsfMatchTask as the square box bounding its spatially-variable, pre-matched WarpedPsf.
914 # Likewise, set the PSF of a PSF-Matched Coadd to the modelPsf
915 # having the maximum width (sufficient because square)
916 modelPsfList = [tempExp.getPsf() for tempExp in tempExpList]
917 modelPsfWidthList = [modelPsf.computeBBox().getWidth() for modelPsf in modelPsfList]
918 psf = modelPsfList[modelPsfWidthList.index(max(modelPsfWidthList))]
919 else:
920 psf = measAlg.CoaddPsf(coaddInputs.ccds, coaddExposure.getWcs(),
921 self.config.coaddPsf.makeControl())
922 coaddExposure.setPsf(psf)
923 apCorrMap = measAlg.makeCoaddApCorrMap(coaddInputs.ccds, coaddExposure.getBBox(afwImage.PARENT),
924 coaddExposure.getWcs())
925 coaddExposure.getInfo().setApCorrMap(apCorrMap)
926 if self.config.doAttachTransmissionCurve:
927 transmissionCurve = measAlg.makeCoaddTransmissionCurve(coaddExposure.getWcs(), coaddInputs.ccds)
928 coaddExposure.getInfo().setTransmissionCurve(transmissionCurve)
930 def assembleSubregion(self, coaddExposure, bbox, tempExpRefList, imageScalerList, weightList,
931 altMaskList, statsFlags, statsCtrl, nImage=None):
932 """Assemble the coadd for a sub-region.
934 For each coaddTempExp, check for (and swap in) an alternative mask
935 if one is passed. Remove mask planes listed in
936 `config.removeMaskPlanes`. Finally, stack the actual exposures using
937 `lsst.afw.math.statisticsStack` with the statistic specified by
938 statsFlags. Typically, the statsFlag will be one of lsst.afw.math.MEAN for
939 a mean-stack or `lsst.afw.math.MEANCLIP` for outlier rejection using
940 an N-sigma clipped mean where N and iterations are specified by
941 statsCtrl. Assign the stacked subregion back to the coadd.
943 Parameters
944 ----------
945 coaddExposure : `lsst.afw.image.Exposure`
946 The target exposure for the coadd.
947 bbox : `lsst.geom.Box`
948 Sub-region to coadd.
949 tempExpRefList : `list`
950 List of data reference to tempExp.
951 imageScalerList : `list`
952 List of image scalers.
953 weightList : `list`
954 List of weights.
955 altMaskList : `list`
956 List of alternate masks to use rather than those stored with
957 tempExp, or None. Each element is dict with keys = mask plane
958 name to which to add the spans.
959 statsFlags : `lsst.afw.math.Property`
960 Property object for statistic for coadd.
961 statsCtrl : `lsst.afw.math.StatisticsControl`
962 Statistics control object for coadd.
963 nImage : `lsst.afw.image.ImageU`, optional
964 Keeps track of exposure count for each pixel.
965 """
966 self.log.debug("Computing coadd over %s", bbox)
967 tempExpName = self.getTempExpDatasetName(self.warpType)
968 coaddExposure.mask.addMaskPlane("REJECTED")
969 coaddExposure.mask.addMaskPlane("CLIPPED")
970 coaddExposure.mask.addMaskPlane("SENSOR_EDGE")
971 maskMap = self.setRejectedMaskMapping(statsCtrl)
972 clipped = afwImage.Mask.getPlaneBitMask("CLIPPED")
973 maskedImageList = []
974 if nImage is not None:
975 subNImage = afwImage.ImageU(bbox.getWidth(), bbox.getHeight())
976 for tempExpRef, imageScaler, altMask in zip(tempExpRefList, imageScalerList, altMaskList):
978 if isinstance(tempExpRef, DeferredDatasetHandle):
979 # Gen 3 API
980 exposure = tempExpRef.get(parameters={'bbox': bbox})
981 else:
982 # Gen 2 API. Delete this when Gen 2 retired
983 exposure = tempExpRef.get(tempExpName + "_sub", bbox=bbox)
985 maskedImage = exposure.getMaskedImage()
986 mask = maskedImage.getMask()
987 if altMask is not None:
988 self.applyAltMaskPlanes(mask, altMask)
989 imageScaler.scaleMaskedImage(maskedImage)
991 # Add 1 for each pixel which is not excluded by the exclude mask.
992 # In legacyCoadd, pixels may also be excluded by afwMath.statisticsStack.
993 if nImage is not None:
994 subNImage.getArray()[maskedImage.getMask().getArray() & statsCtrl.getAndMask() == 0] += 1
995 if self.config.removeMaskPlanes:
996 self.removeMaskPlanes(maskedImage)
997 maskedImageList.append(maskedImage)
999 if self.config.doInputMap:
1000 visit = exposure.getInfo().getCoaddInputs().visits[0].getId()
1001 self.inputMapper.mask_warp_bbox(bbox, visit, mask, statsCtrl.getAndMask())
1003 with self.timer("stack"):
1004 coaddSubregion = afwMath.statisticsStack(maskedImageList, statsFlags, statsCtrl, weightList,
1005 clipped, # also set output to CLIPPED if sigma-clipped
1006 maskMap)
1007 coaddExposure.maskedImage.assign(coaddSubregion, bbox)
1008 if nImage is not None:
1009 nImage.assign(subNImage, bbox)
1011 def assembleOnlineMeanCoadd(self, coaddExposure, tempExpRefList, imageScalerList, weightList,
1012 altMaskList, statsCtrl, nImage=None):
1013 """Assemble the coadd using the "online" method.
1015 This method takes a running sum of images and weights to save memory.
1016 It only works for MEAN statistics.
1018 Parameters
1019 ----------
1020 coaddExposure : `lsst.afw.image.Exposure`
1021 The target exposure for the coadd.
1022 tempExpRefList : `list`
1023 List of data reference to tempExp.
1024 imageScalerList : `list`
1025 List of image scalers.
1026 weightList : `list`
1027 List of weights.
1028 altMaskList : `list`
1029 List of alternate masks to use rather than those stored with
1030 tempExp, or None. Each element is dict with keys = mask plane
1031 name to which to add the spans.
1032 statsCtrl : `lsst.afw.math.StatisticsControl`
1033 Statistics control object for coadd
1034 nImage : `lsst.afw.image.ImageU`, optional
1035 Keeps track of exposure count for each pixel.
1036 """
1037 self.log.debug("Computing online coadd.")
1038 tempExpName = self.getTempExpDatasetName(self.warpType)
1039 coaddExposure.mask.addMaskPlane("REJECTED")
1040 coaddExposure.mask.addMaskPlane("CLIPPED")
1041 coaddExposure.mask.addMaskPlane("SENSOR_EDGE")
1042 maskMap = self.setRejectedMaskMapping(statsCtrl)
1043 thresholdDict = AccumulatorMeanStack.stats_ctrl_to_threshold_dict(statsCtrl)
1045 bbox = coaddExposure.maskedImage.getBBox()
1047 stacker = AccumulatorMeanStack(
1048 coaddExposure.image.array.shape,
1049 statsCtrl.getAndMask(),
1050 mask_threshold_dict=thresholdDict,
1051 mask_map=maskMap,
1052 no_good_pixels_mask=statsCtrl.getNoGoodPixelsMask(),
1053 calc_error_from_input_variance=self.config.calcErrorFromInputVariance,
1054 compute_n_image=(nImage is not None)
1055 )
1057 for tempExpRef, imageScaler, altMask, weight in zip(tempExpRefList,
1058 imageScalerList,
1059 altMaskList,
1060 weightList):
1061 if isinstance(tempExpRef, DeferredDatasetHandle):
1062 # Gen 3 API
1063 exposure = tempExpRef.get()
1064 else:
1065 # Gen 2 API. Delete this when Gen 2 retired
1066 exposure = tempExpRef.get(tempExpName)
1068 maskedImage = exposure.getMaskedImage()
1069 mask = maskedImage.getMask()
1070 if altMask is not None:
1071 self.applyAltMaskPlanes(mask, altMask)
1072 imageScaler.scaleMaskedImage(maskedImage)
1073 if self.config.removeMaskPlanes:
1074 self.removeMaskPlanes(maskedImage)
1076 stacker.add_masked_image(maskedImage, weight=weight)
1078 if self.config.doInputMap:
1079 visit = exposure.getInfo().getCoaddInputs().visits[0].getId()
1080 self.inputMapper.mask_warp_bbox(bbox, visit, mask, statsCtrl.getAndMask())
1082 stacker.fill_stacked_masked_image(coaddExposure.maskedImage)
1084 if nImage is not None:
1085 nImage.array[:, :] = stacker.n_image
1087 def removeMaskPlanes(self, maskedImage):
1088 """Unset the mask of an image for mask planes specified in the config.
1090 Parameters
1091 ----------
1092 maskedImage : `lsst.afw.image.MaskedImage`
1093 The masked image to be modified.
1094 """
1095 mask = maskedImage.getMask()
1096 for maskPlane in self.config.removeMaskPlanes:
1097 try:
1098 mask &= ~mask.getPlaneBitMask(maskPlane)
1099 except pexExceptions.InvalidParameterError:
1100 self.log.debug("Unable to remove mask plane %s: no mask plane with that name was found.",
1101 maskPlane)
1103 @staticmethod
1104 def setRejectedMaskMapping(statsCtrl):
1105 """Map certain mask planes of the warps to new planes for the coadd.
1107 If a pixel is rejected due to a mask value other than EDGE, NO_DATA,
1108 or CLIPPED, set it to REJECTED on the coadd.
1109 If a pixel is rejected due to EDGE, set the coadd pixel to SENSOR_EDGE.
1110 If a pixel is rejected due to CLIPPED, set the coadd pixel to CLIPPED.
1112 Parameters
1113 ----------
1114 statsCtrl : `lsst.afw.math.StatisticsControl`
1115 Statistics control object for coadd
1117 Returns
1118 -------
1119 maskMap : `list` of `tuple` of `int`
1120 A list of mappings of mask planes of the warped exposures to
1121 mask planes of the coadd.
1122 """
1123 edge = afwImage.Mask.getPlaneBitMask("EDGE")
1124 noData = afwImage.Mask.getPlaneBitMask("NO_DATA")
1125 clipped = afwImage.Mask.getPlaneBitMask("CLIPPED")
1126 toReject = statsCtrl.getAndMask() & (~noData) & (~edge) & (~clipped)
1127 maskMap = [(toReject, afwImage.Mask.getPlaneBitMask("REJECTED")),
1128 (edge, afwImage.Mask.getPlaneBitMask("SENSOR_EDGE")),
1129 (clipped, clipped)]
1130 return maskMap
1132 def applyAltMaskPlanes(self, mask, altMaskSpans):
1133 """Apply in place alt mask formatted as SpanSets to a mask.
1135 Parameters
1136 ----------
1137 mask : `lsst.afw.image.Mask`
1138 Original mask.
1139 altMaskSpans : `dict`
1140 SpanSet lists to apply. Each element contains the new mask
1141 plane name (e.g. "CLIPPED and/or "NO_DATA") as the key,
1142 and list of SpanSets to apply to the mask.
1144 Returns
1145 -------
1146 mask : `lsst.afw.image.Mask`
1147 Updated mask.
1148 """
1149 if self.config.doUsePsfMatchedPolygons:
1150 if ("NO_DATA" in altMaskSpans) and ("NO_DATA" in self.config.badMaskPlanes):
1151 # Clear away any other masks outside the validPolygons. These pixels are no longer
1152 # contributing to inexact PSFs, and will still be rejected because of NO_DATA
1153 # self.config.doUsePsfMatchedPolygons should be True only in CompareWarpAssemble
1154 # This mask-clearing step must only occur *before* applying the new masks below
1155 for spanSet in altMaskSpans['NO_DATA']:
1156 spanSet.clippedTo(mask.getBBox()).clearMask(mask, self.getBadPixelMask())
1158 for plane, spanSetList in altMaskSpans.items():
1159 maskClipValue = mask.addMaskPlane(plane)
1160 for spanSet in spanSetList:
1161 spanSet.clippedTo(mask.getBBox()).setMask(mask, 2**maskClipValue)
1162 return mask
1164 def shrinkValidPolygons(self, coaddInputs):
1165 """Shrink coaddInputs' ccds' ValidPolygons in place.
1167 Either modify each ccd's validPolygon in place, or if CoaddInputs
1168 does not have a validPolygon, create one from its bbox.
1170 Parameters
1171 ----------
1172 coaddInputs : `lsst.afw.image.coaddInputs`
1173 Original mask.
1175 """
1176 for ccd in coaddInputs.ccds:
1177 polyOrig = ccd.getValidPolygon()
1178 validPolyBBox = polyOrig.getBBox() if polyOrig else ccd.getBBox()
1179 validPolyBBox.grow(-self.config.matchingKernelSize//2)
1180 if polyOrig:
1181 validPolygon = polyOrig.intersectionSingle(validPolyBBox)
1182 else:
1183 validPolygon = afwGeom.polygon.Polygon(geom.Box2D(validPolyBBox))
1184 ccd.setValidPolygon(validPolygon)
1186 def readBrightObjectMasks(self, dataRef):
1187 """Retrieve the bright object masks.
1189 Returns None on failure.
1191 Parameters
1192 ----------
1193 dataRef : `lsst.daf.persistence.butlerSubset.ButlerDataRef`
1194 A Butler dataRef.
1196 Returns
1197 -------
1198 result : `lsst.daf.persistence.butlerSubset.ButlerDataRef`
1199 Bright object mask from the Butler object, or None if it cannot
1200 be retrieved.
1201 """
1202 try:
1203 return dataRef.get(datasetType="brightObjectMask", immediate=True)
1204 except Exception as e:
1205 self.log.warning("Unable to read brightObjectMask for %s: %s", dataRef.dataId, e)
1206 return None
1208 def setBrightObjectMasks(self, exposure, brightObjectMasks, dataId=None):
1209 """Set the bright object masks.
1211 Parameters
1212 ----------
1213 exposure : `lsst.afw.image.Exposure`
1214 Exposure under consideration.
1215 dataId : `lsst.daf.persistence.dataId`
1216 Data identifier dict for patch.
1217 brightObjectMasks : `lsst.afw.table`
1218 Table of bright objects to mask.
1219 """
1221 if brightObjectMasks is None:
1222 self.log.warning("Unable to apply bright object mask: none supplied")
1223 return
1224 self.log.info("Applying %d bright object masks to %s", len(brightObjectMasks), dataId)
1225 mask = exposure.getMaskedImage().getMask()
1226 wcs = exposure.getWcs()
1227 plateScale = wcs.getPixelScale().asArcseconds()
1229 for rec in brightObjectMasks:
1230 center = geom.PointI(wcs.skyToPixel(rec.getCoord()))
1231 if rec["type"] == "box":
1232 assert rec["angle"] == 0.0, ("Angle != 0 for mask object %s" % rec["id"])
1233 width = rec["width"].asArcseconds()/plateScale # convert to pixels
1234 height = rec["height"].asArcseconds()/plateScale # convert to pixels
1236 halfSize = geom.ExtentI(0.5*width, 0.5*height)
1237 bbox = geom.Box2I(center - halfSize, center + halfSize)
1239 bbox = geom.BoxI(geom.PointI(int(center[0] - 0.5*width), int(center[1] - 0.5*height)),
1240 geom.PointI(int(center[0] + 0.5*width), int(center[1] + 0.5*height)))
1241 spans = afwGeom.SpanSet(bbox)
1242 elif rec["type"] == "circle":
1243 radius = int(rec["radius"].asArcseconds()/plateScale) # convert to pixels
1244 spans = afwGeom.SpanSet.fromShape(radius, offset=center)
1245 else:
1246 self.log.warning("Unexpected region type %s at %s", rec["type"], center)
1247 continue
1248 spans.clippedTo(mask.getBBox()).setMask(mask, self.brightObjectBitmask)
1250 def setInexactPsf(self, mask):
1251 """Set INEXACT_PSF mask plane.
1253 If any of the input images isn't represented in the coadd (due to
1254 clipped pixels or chip gaps), the `CoaddPsf` will be inexact. Flag
1255 these pixels.
1257 Parameters
1258 ----------
1259 mask : `lsst.afw.image.Mask`
1260 Coadded exposure's mask, modified in-place.
1261 """
1262 mask.addMaskPlane("INEXACT_PSF")
1263 inexactPsf = mask.getPlaneBitMask("INEXACT_PSF")
1264 sensorEdge = mask.getPlaneBitMask("SENSOR_EDGE") # chip edges (so PSF is discontinuous)
1265 clipped = mask.getPlaneBitMask("CLIPPED") # pixels clipped from coadd
1266 rejected = mask.getPlaneBitMask("REJECTED") # pixels rejected from coadd due to masks
1267 array = mask.getArray()
1268 selected = array & (sensorEdge | clipped | rejected) > 0
1269 array[selected] |= inexactPsf
1271 @classmethod
1272 def _makeArgumentParser(cls):
1273 """Create an argument parser.
1274 """
1275 parser = pipeBase.ArgumentParser(name=cls._DefaultName)
1276 parser.add_id_argument("--id", cls.ConfigClass().coaddName + "Coadd_"
1277 + cls.ConfigClass().warpType + "Warp",
1278 help="data ID, e.g. --id tract=12345 patch=1,2",
1279 ContainerClass=AssembleCoaddDataIdContainer)
1280 parser.add_id_argument("--selectId", "calexp", help="data ID, e.g. --selectId visit=6789 ccd=0..9",
1281 ContainerClass=SelectDataIdContainer)
1282 return parser
1284 @staticmethod
1285 def _subBBoxIter(bbox, subregionSize):
1286 """Iterate over subregions of a bbox.
1288 Parameters
1289 ----------
1290 bbox : `lsst.geom.Box2I`
1291 Bounding box over which to iterate.
1292 subregionSize: `lsst.geom.Extent2I`
1293 Size of sub-bboxes.
1295 Yields
1296 ------
1297 subBBox : `lsst.geom.Box2I`
1298 Next sub-bounding box of size ``subregionSize`` or smaller; each ``subBBox``
1299 is contained within ``bbox``, so it may be smaller than ``subregionSize`` at
1300 the edges of ``bbox``, but it will never be empty.
1301 """
1302 if bbox.isEmpty():
1303 raise RuntimeError("bbox %s is empty" % (bbox,))
1304 if subregionSize[0] < 1 or subregionSize[1] < 1:
1305 raise RuntimeError("subregionSize %s must be nonzero" % (subregionSize,))
1307 for rowShift in range(0, bbox.getHeight(), subregionSize[1]):
1308 for colShift in range(0, bbox.getWidth(), subregionSize[0]):
1309 subBBox = geom.Box2I(bbox.getMin() + geom.Extent2I(colShift, rowShift), subregionSize)
1310 subBBox.clip(bbox)
1311 if subBBox.isEmpty():
1312 raise RuntimeError("Bug: empty bbox! bbox=%s, subregionSize=%s, "
1313 "colShift=%s, rowShift=%s" %
1314 (bbox, subregionSize, colShift, rowShift))
1315 yield subBBox
1317 def filterWarps(self, inputs, goodVisits):
1318 """Return list of only inputRefs with visitId in goodVisits ordered by goodVisit
1320 Parameters
1321 ----------
1322 inputs : list
1323 List of `lsst.pipe.base.connections.DeferredDatasetRef` with dataId containing visit
1324 goodVisit : `dict`
1325 Dictionary with good visitIds as the keys. Value ignored.
1327 Returns:
1328 --------
1329 filteredInputs : `list`
1330 Filtered and sorted list of `lsst.pipe.base.connections.DeferredDatasetRef`
1331 """
1332 inputWarpDict = {inputRef.ref.dataId['visit']: inputRef for inputRef in inputs}
1333 filteredInputs = []
1334 for visit in goodVisits.keys():
1335 if visit in inputWarpDict:
1336 filteredInputs.append(inputWarpDict[visit])
1337 return filteredInputs
1340class AssembleCoaddDataIdContainer(pipeBase.DataIdContainer):
1341 """A version of `lsst.pipe.base.DataIdContainer` specialized for assembleCoadd.
1342 """
1344 def makeDataRefList(self, namespace):
1345 """Make self.refList from self.idList.
1347 Parameters
1348 ----------
1349 namespace
1350 Results of parsing command-line (with ``butler`` and ``log`` elements).
1351 """
1352 datasetType = namespace.config.coaddName + "Coadd"
1353 keysCoadd = namespace.butler.getKeys(datasetType=datasetType, level=self.level)
1355 for dataId in self.idList:
1356 # tract and patch are required
1357 for key in keysCoadd:
1358 if key not in dataId:
1359 raise RuntimeError("--id must include " + key)
1361 dataRef = namespace.butler.dataRef(
1362 datasetType=datasetType,
1363 dataId=dataId,
1364 )
1365 self.refList.append(dataRef)
1368def countMaskFromFootprint(mask, footprint, bitmask, ignoreMask):
1369 """Function to count the number of pixels with a specific mask in a
1370 footprint.
1372 Find the intersection of mask & footprint. Count all pixels in the mask
1373 that are in the intersection that have bitmask set but do not have
1374 ignoreMask set. Return the count.
1376 Parameters
1377 ----------
1378 mask : `lsst.afw.image.Mask`
1379 Mask to define intersection region by.
1380 footprint : `lsst.afw.detection.Footprint`
1381 Footprint to define the intersection region by.
1382 bitmask
1383 Specific mask that we wish to count the number of occurances of.
1384 ignoreMask
1385 Pixels to not consider.
1387 Returns
1388 -------
1389 result : `int`
1390 Count of number of pixels in footprint with specified mask.
1391 """
1392 bbox = footprint.getBBox()
1393 bbox.clip(mask.getBBox(afwImage.PARENT))
1394 fp = afwImage.Mask(bbox)
1395 subMask = mask.Factory(mask, bbox, afwImage.PARENT)
1396 footprint.spans.setMask(fp, bitmask)
1397 return numpy.logical_and((subMask.getArray() & fp.getArray()) > 0,
1398 (subMask.getArray() & ignoreMask) == 0).sum()
1401class SafeClipAssembleCoaddConfig(AssembleCoaddConfig, pipelineConnections=AssembleCoaddConnections):
1402 """Configuration parameters for the SafeClipAssembleCoaddTask.
1403 """
1404 clipDetection = pexConfig.ConfigurableField(
1405 target=SourceDetectionTask,
1406 doc="Detect sources on difference between unclipped and clipped coadd")
1407 minClipFootOverlap = pexConfig.Field(
1408 doc="Minimum fractional overlap of clipped footprint with visit DETECTED to be clipped",
1409 dtype=float,
1410 default=0.6
1411 )
1412 minClipFootOverlapSingle = pexConfig.Field(
1413 doc="Minimum fractional overlap of clipped footprint with visit DETECTED to be "
1414 "clipped when only one visit overlaps",
1415 dtype=float,
1416 default=0.5
1417 )
1418 minClipFootOverlapDouble = pexConfig.Field(
1419 doc="Minimum fractional overlap of clipped footprints with visit DETECTED to be "
1420 "clipped when two visits overlap",
1421 dtype=float,
1422 default=0.45
1423 )
1424 maxClipFootOverlapDouble = pexConfig.Field(
1425 doc="Maximum fractional overlap of clipped footprints with visit DETECTED when "
1426 "considering two visits",
1427 dtype=float,
1428 default=0.15
1429 )
1430 minBigOverlap = pexConfig.Field(
1431 doc="Minimum number of pixels in footprint to use DETECTED mask from the single visits "
1432 "when labeling clipped footprints",
1433 dtype=int,
1434 default=100
1435 )
1437 def setDefaults(self):
1438 """Set default values for clipDetection.
1440 Notes
1441 -----
1442 The numeric values for these configuration parameters were
1443 empirically determined, future work may further refine them.
1444 """
1445 AssembleCoaddConfig.setDefaults(self)
1446 self.clipDetection.doTempLocalBackground = False
1447 self.clipDetection.reEstimateBackground = False
1448 self.clipDetection.returnOriginalFootprints = False
1449 self.clipDetection.thresholdPolarity = "both"
1450 self.clipDetection.thresholdValue = 2
1451 self.clipDetection.nSigmaToGrow = 2
1452 self.clipDetection.minPixels = 4
1453 self.clipDetection.isotropicGrow = True
1454 self.clipDetection.thresholdType = "pixel_stdev"
1455 self.sigmaClip = 1.5
1456 self.clipIter = 3
1457 self.statistic = "MEAN"
1459 def validate(self):
1460 if self.doSigmaClip:
1461 log.warning("Additional Sigma-clipping not allowed in Safe-clipped Coadds. "
1462 "Ignoring doSigmaClip.")
1463 self.doSigmaClip = False
1464 if self.statistic != "MEAN":
1465 raise ValueError("Only MEAN statistic allowed for final stacking in SafeClipAssembleCoadd "
1466 "(%s chosen). Please set statistic to MEAN."
1467 % (self.statistic))
1468 AssembleCoaddTask.ConfigClass.validate(self)
1471class SafeClipAssembleCoaddTask(AssembleCoaddTask):
1472 """Assemble a coadded image from a set of coadded temporary exposures,
1473 being careful to clip & flag areas with potential artifacts.
1475 In ``AssembleCoaddTask``, we compute the coadd as an clipped mean (i.e.,
1476 we clip outliers). The problem with doing this is that when computing the
1477 coadd PSF at a given location, individual visit PSFs from visits with
1478 outlier pixels contribute to the coadd PSF and cannot be treated correctly.
1479 In this task, we correct for this behavior by creating a new
1480 ``badMaskPlane`` 'CLIPPED'. We populate this plane on the input
1481 coaddTempExps and the final coadd where
1483 i. difference imaging suggests that there is an outlier and
1484 ii. this outlier appears on only one or two images.
1486 Such regions will not contribute to the final coadd. Furthermore, any
1487 routine to determine the coadd PSF can now be cognizant of clipped regions.
1488 Note that the algorithm implemented by this task is preliminary and works
1489 correctly for HSC data. Parameter modifications and or considerable
1490 redesigning of the algorithm is likley required for other surveys.
1492 ``SafeClipAssembleCoaddTask`` uses a ``SourceDetectionTask``
1493 "clipDetection" subtask and also sub-classes ``AssembleCoaddTask``.
1494 You can retarget the ``SourceDetectionTask`` "clipDetection" subtask
1495 if you wish.
1497 Notes
1498 -----
1499 The `lsst.pipe.base.cmdLineTask.CmdLineTask` interface supports a
1500 flag ``-d`` to import ``debug.py`` from your ``PYTHONPATH``;
1501 see `baseDebug` for more about ``debug.py`` files.
1502 `SafeClipAssembleCoaddTask` has no debug variables of its own.
1503 The ``SourceDetectionTask`` "clipDetection" subtasks may support debug
1504 variables. See the documetation for `SourceDetectionTask` "clipDetection"
1505 for further information.
1507 Examples
1508 --------
1509 `SafeClipAssembleCoaddTask` assembles a set of warped ``coaddTempExp``
1510 images into a coadded image. The `SafeClipAssembleCoaddTask` is invoked by
1511 running assembleCoadd.py *without* the flag '--legacyCoadd'.
1513 Usage of ``assembleCoadd.py`` expects a data reference to the tract patch
1514 and filter to be coadded (specified using
1515 '--id = [KEY=VALUE1[^VALUE2[^VALUE3...] [KEY=VALUE1[^VALUE2[^VALUE3...] ...]]')
1516 along with a list of coaddTempExps to attempt to coadd (specified using
1517 '--selectId [KEY=VALUE1[^VALUE2[^VALUE3...] [KEY=VALUE1[^VALUE2[^VALUE3...] ...]]').
1518 Only the coaddTempExps that cover the specified tract and patch will be
1519 coadded. A list of the available optional arguments can be obtained by
1520 calling assembleCoadd.py with the --help command line argument:
1522 .. code-block:: none
1524 assembleCoadd.py --help
1526 To demonstrate usage of the `SafeClipAssembleCoaddTask` in the larger
1527 context of multi-band processing, we will generate the HSC-I & -R band
1528 coadds from HSC engineering test data provided in the ci_hsc package.
1529 To begin, assuming that the lsst stack has been already set up, we must
1530 set up the obs_subaru and ci_hsc packages. This defines the environment
1531 variable $CI_HSC_DIR and points at the location of the package. The raw
1532 HSC data live in the ``$CI_HSC_DIR/raw`` directory. To begin assembling
1533 the coadds, we must first
1535 - ``processCcd``
1536 process the individual ccds in $CI_HSC_RAW to produce calibrated exposures
1537 - ``makeSkyMap``
1538 create a skymap that covers the area of the sky present in the raw exposures
1539 - ``makeCoaddTempExp``
1540 warp the individual calibrated exposures to the tangent plane of the coadd</DD>
1542 We can perform all of these steps by running
1544 .. code-block:: none
1546 $CI_HSC_DIR scons warp-903986 warp-904014 warp-903990 warp-904010 warp-903988
1548 This will produce warped coaddTempExps for each visit. To coadd the
1549 warped data, we call ``assembleCoadd.py`` as follows:
1551 .. code-block:: none
1553 assembleCoadd.py $CI_HSC_DIR/DATA --id patch=5,4 tract=0 filter=HSC-I \
1554 --selectId visit=903986 ccd=16 --selectId visit=903986 ccd=22 --selectId visit=903986 ccd=23 \
1555 --selectId visit=903986 ccd=100--selectId visit=904014 ccd=1 --selectId visit=904014 ccd=6 \
1556 --selectId visit=904014 ccd=12 --selectId visit=903990 ccd=18 --selectId visit=903990 ccd=25 \
1557 --selectId visit=904010 ccd=4 --selectId visit=904010 ccd=10 --selectId visit=904010 ccd=100 \
1558 --selectId visit=903988 ccd=16 --selectId visit=903988 ccd=17 --selectId visit=903988 ccd=23 \
1559 --selectId visit=903988 ccd=24
1561 This will process the HSC-I band data. The results are written in
1562 ``$CI_HSC_DIR/DATA/deepCoadd-results/HSC-I``.
1564 You may also choose to run:
1566 .. code-block:: none
1568 scons warp-903334 warp-903336 warp-903338 warp-903342 warp-903344 warp-903346 nnn
1569 assembleCoadd.py $CI_HSC_DIR/DATA --id patch=5,4 tract=0 filter=HSC-R --selectId visit=903334 ccd=16 \
1570 --selectId visit=903334 ccd=22 --selectId visit=903334 ccd=23 --selectId visit=903334 ccd=100 \
1571 --selectId visit=903336 ccd=17 --selectId visit=903336 ccd=24 --selectId visit=903338 ccd=18 \
1572 --selectId visit=903338 ccd=25 --selectId visit=903342 ccd=4 --selectId visit=903342 ccd=10 \
1573 --selectId visit=903342 ccd=100 --selectId visit=903344 ccd=0 --selectId visit=903344 ccd=5 \
1574 --selectId visit=903344 ccd=11 --selectId visit=903346 ccd=1 --selectId visit=903346 ccd=6 \
1575 --selectId visit=903346 ccd=12
1577 to generate the coadd for the HSC-R band if you are interested in following
1578 multiBand Coadd processing as discussed in ``pipeTasks_multiBand``.
1579 """
1580 ConfigClass = SafeClipAssembleCoaddConfig
1581 _DefaultName = "safeClipAssembleCoadd"
1583 def __init__(self, *args, **kwargs):
1584 AssembleCoaddTask.__init__(self, *args, **kwargs)
1585 schema = afwTable.SourceTable.makeMinimalSchema()
1586 self.makeSubtask("clipDetection", schema=schema)
1588 @utils.inheritDoc(AssembleCoaddTask)
1589 @pipeBase.timeMethod
1590 def run(self, skyInfo, tempExpRefList, imageScalerList, weightList, *args, **kwargs):
1591 """Assemble the coadd for a region.
1593 Compute the difference of coadds created with and without outlier
1594 rejection to identify coadd pixels that have outlier values in some
1595 individual visits.
1596 Detect clipped regions on the difference image and mark these regions
1597 on the one or two individual coaddTempExps where they occur if there
1598 is significant overlap between the clipped region and a source. This
1599 leaves us with a set of footprints from the difference image that have
1600 been identified as having occured on just one or two individual visits.
1601 However, these footprints were generated from a difference image. It
1602 is conceivable for a large diffuse source to have become broken up
1603 into multiple footprints acrosss the coadd difference in this process.
1604 Determine the clipped region from all overlapping footprints from the
1605 detected sources in each visit - these are big footprints.
1606 Combine the small and big clipped footprints and mark them on a new
1607 bad mask plane.
1608 Generate the coadd using `AssembleCoaddTask.run` without outlier
1609 removal. Clipped footprints will no longer make it into the coadd
1610 because they are marked in the new bad mask plane.
1612 Notes
1613 -----
1614 args and kwargs are passed but ignored in order to match the call
1615 signature expected by the parent task.
1616 """
1617 exp = self.buildDifferenceImage(skyInfo, tempExpRefList, imageScalerList, weightList)
1618 mask = exp.getMaskedImage().getMask()
1619 mask.addMaskPlane("CLIPPED")
1621 result = self.detectClip(exp, tempExpRefList)
1623 self.log.info('Found %d clipped objects', len(result.clipFootprints))
1625 maskClipValue = mask.getPlaneBitMask("CLIPPED")
1626 maskDetValue = mask.getPlaneBitMask("DETECTED") | mask.getPlaneBitMask("DETECTED_NEGATIVE")
1627 # Append big footprints from individual Warps to result.clipSpans
1628 bigFootprints = self.detectClipBig(result.clipSpans, result.clipFootprints, result.clipIndices,
1629 result.detectionFootprints, maskClipValue, maskDetValue,
1630 exp.getBBox())
1631 # Create mask of the current clipped footprints
1632 maskClip = mask.Factory(mask.getBBox(afwImage.PARENT))
1633 afwDet.setMaskFromFootprintList(maskClip, result.clipFootprints, maskClipValue)
1635 maskClipBig = maskClip.Factory(mask.getBBox(afwImage.PARENT))
1636 afwDet.setMaskFromFootprintList(maskClipBig, bigFootprints, maskClipValue)
1637 maskClip |= maskClipBig
1639 # Assemble coadd from base class, but ignoring CLIPPED pixels
1640 badMaskPlanes = self.config.badMaskPlanes[:]
1641 badMaskPlanes.append("CLIPPED")
1642 badPixelMask = afwImage.Mask.getPlaneBitMask(badMaskPlanes)
1643 return AssembleCoaddTask.run(self, skyInfo, tempExpRefList, imageScalerList, weightList,
1644 result.clipSpans, mask=badPixelMask)
1646 def buildDifferenceImage(self, skyInfo, tempExpRefList, imageScalerList, weightList):
1647 """Return an exposure that contains the difference between unclipped
1648 and clipped coadds.
1650 Generate a difference image between clipped and unclipped coadds.
1651 Compute the difference image by subtracting an outlier-clipped coadd
1652 from an outlier-unclipped coadd. Return the difference image.
1654 Parameters
1655 ----------
1656 skyInfo : `lsst.pipe.base.Struct`
1657 Patch geometry information, from getSkyInfo
1658 tempExpRefList : `list`
1659 List of data reference to tempExp
1660 imageScalerList : `list`
1661 List of image scalers
1662 weightList : `list`
1663 List of weights
1665 Returns
1666 -------
1667 exp : `lsst.afw.image.Exposure`
1668 Difference image of unclipped and clipped coadd wrapped in an Exposure
1669 """
1670 config = AssembleCoaddConfig()
1671 # getattr necessary because subtasks do not survive Config.toDict()
1672 # exclude connections because the class of self.config.connections is not
1673 # the same as AssembleCoaddConfig.connections, and the connections are not
1674 # needed to run this task anyway.
1675 configIntersection = {k: getattr(self.config, k)
1676 for k, v in self.config.toDict().items()
1677 if (k in config.keys() and k != "connections")}
1678 configIntersection['doInputMap'] = False
1679 configIntersection['doNImage'] = False
1680 config.update(**configIntersection)
1682 # statistic MEAN copied from self.config.statistic, but for clarity explicitly assign
1683 config.statistic = 'MEAN'
1684 task = AssembleCoaddTask(config=config)
1685 coaddMean = task.run(skyInfo, tempExpRefList, imageScalerList, weightList).coaddExposure
1687 config.statistic = 'MEANCLIP'
1688 task = AssembleCoaddTask(config=config)
1689 coaddClip = task.run(skyInfo, tempExpRefList, imageScalerList, weightList).coaddExposure
1691 coaddDiff = coaddMean.getMaskedImage().Factory(coaddMean.getMaskedImage())
1692 coaddDiff -= coaddClip.getMaskedImage()
1693 exp = afwImage.ExposureF(coaddDiff)
1694 exp.setPsf(coaddMean.getPsf())
1695 return exp
1697 def detectClip(self, exp, tempExpRefList):
1698 """Detect clipped regions on an exposure and set the mask on the
1699 individual tempExp masks.
1701 Detect footprints in the difference image after smoothing the
1702 difference image with a Gaussian kernal. Identify footprints that
1703 overlap with one or two input ``coaddTempExps`` by comparing the
1704 computed overlap fraction to thresholds set in the config. A different
1705 threshold is applied depending on the number of overlapping visits
1706 (restricted to one or two). If the overlap exceeds the thresholds,
1707 the footprint is considered "CLIPPED" and is marked as such on the
1708 coaddTempExp. Return a struct with the clipped footprints, the indices
1709 of the ``coaddTempExps`` that end up overlapping with the clipped
1710 footprints, and a list of new masks for the ``coaddTempExps``.
1712 Parameters
1713 ----------
1714 exp : `lsst.afw.image.Exposure`
1715 Exposure to run detection on.
1716 tempExpRefList : `list`
1717 List of data reference to tempExp.
1719 Returns
1720 -------
1721 result : `lsst.pipe.base.Struct`
1722 Result struct with components:
1724 - ``clipFootprints``: list of clipped footprints.
1725 - ``clipIndices``: indices for each ``clippedFootprint`` in
1726 ``tempExpRefList``.
1727 - ``clipSpans``: List of dictionaries containing spanSet lists
1728 to clip. Each element contains the new maskplane name
1729 ("CLIPPED") as the key and list of ``SpanSets`` as the value.
1730 - ``detectionFootprints``: List of DETECTED/DETECTED_NEGATIVE plane
1731 compressed into footprints.
1732 """
1733 mask = exp.getMaskedImage().getMask()
1734 maskDetValue = mask.getPlaneBitMask("DETECTED") | mask.getPlaneBitMask("DETECTED_NEGATIVE")
1735 fpSet = self.clipDetection.detectFootprints(exp, doSmooth=True, clearMask=True)
1736 # Merge positive and negative together footprints together
1737 fpSet.positive.merge(fpSet.negative)
1738 footprints = fpSet.positive
1739 self.log.info('Found %d potential clipped objects', len(footprints.getFootprints()))
1740 ignoreMask = self.getBadPixelMask()
1742 clipFootprints = []
1743 clipIndices = []
1744 artifactSpanSets = [{'CLIPPED': list()} for _ in tempExpRefList]
1746 # for use by detectClipBig
1747 visitDetectionFootprints = []
1749 dims = [len(tempExpRefList), len(footprints.getFootprints())]
1750 overlapDetArr = numpy.zeros(dims, dtype=numpy.uint16)
1751 ignoreArr = numpy.zeros(dims, dtype=numpy.uint16)
1753 # Loop over masks once and extract/store only relevant overlap metrics and detection footprints
1754 for i, warpRef in enumerate(tempExpRefList):
1755 tmpExpMask = warpRef.get(datasetType=self.getTempExpDatasetName(self.warpType),
1756 immediate=True).getMaskedImage().getMask()
1757 maskVisitDet = tmpExpMask.Factory(tmpExpMask, tmpExpMask.getBBox(afwImage.PARENT),
1758 afwImage.PARENT, True)
1759 maskVisitDet &= maskDetValue
1760 visitFootprints = afwDet.FootprintSet(maskVisitDet, afwDet.Threshold(1))
1761 visitDetectionFootprints.append(visitFootprints)
1763 for j, footprint in enumerate(footprints.getFootprints()):
1764 ignoreArr[i, j] = countMaskFromFootprint(tmpExpMask, footprint, ignoreMask, 0x0)
1765 overlapDetArr[i, j] = countMaskFromFootprint(tmpExpMask, footprint, maskDetValue, ignoreMask)
1767 # build a list of clipped spans for each visit
1768 for j, footprint in enumerate(footprints.getFootprints()):
1769 nPixel = footprint.getArea()
1770 overlap = [] # hold the overlap with each visit
1771 indexList = [] # index of visit in global list
1772 for i in range(len(tempExpRefList)):
1773 ignore = ignoreArr[i, j]
1774 overlapDet = overlapDetArr[i, j]
1775 totPixel = nPixel - ignore
1777 # If we have more bad pixels than detection skip
1778 if ignore > overlapDet or totPixel <= 0.5*nPixel or overlapDet == 0:
1779 continue
1780 overlap.append(overlapDet/float(totPixel))
1781 indexList.append(i)
1783 overlap = numpy.array(overlap)
1784 if not len(overlap):
1785 continue
1787 keep = False # Should this footprint be marked as clipped?
1788 keepIndex = [] # Which tempExps does the clipped footprint belong to
1790 # If footprint only has one overlap use a lower threshold
1791 if len(overlap) == 1:
1792 if overlap[0] > self.config.minClipFootOverlapSingle:
1793 keep = True
1794 keepIndex = [0]
1795 else:
1796 # This is the general case where only visit should be clipped
1797 clipIndex = numpy.where(overlap > self.config.minClipFootOverlap)[0]
1798 if len(clipIndex) == 1:
1799 keep = True
1800 keepIndex = [clipIndex[0]]
1802 # Test if there are clipped objects that overlap two different visits
1803 clipIndex = numpy.where(overlap > self.config.minClipFootOverlapDouble)[0]
1804 if len(clipIndex) == 2 and len(overlap) > 3:
1805 clipIndexComp = numpy.where(overlap <= self.config.minClipFootOverlapDouble)[0]
1806 if numpy.max(overlap[clipIndexComp]) <= self.config.maxClipFootOverlapDouble:
1807 keep = True
1808 keepIndex = clipIndex
1810 if not keep:
1811 continue
1813 for index in keepIndex:
1814 globalIndex = indexList[index]
1815 artifactSpanSets[globalIndex]['CLIPPED'].append(footprint.spans)
1817 clipIndices.append(numpy.array(indexList)[keepIndex])
1818 clipFootprints.append(footprint)
1820 return pipeBase.Struct(clipFootprints=clipFootprints, clipIndices=clipIndices,
1821 clipSpans=artifactSpanSets, detectionFootprints=visitDetectionFootprints)
1823 def detectClipBig(self, clipList, clipFootprints, clipIndices, detectionFootprints,
1824 maskClipValue, maskDetValue, coaddBBox):
1825 """Return individual warp footprints for large artifacts and append
1826 them to ``clipList`` in place.
1828 Identify big footprints composed of many sources in the coadd
1829 difference that may have originated in a large diffuse source in the
1830 coadd. We do this by indentifying all clipped footprints that overlap
1831 significantly with each source in all the coaddTempExps.
1833 Parameters
1834 ----------
1835 clipList : `list`
1836 List of alt mask SpanSets with clipping information. Modified.
1837 clipFootprints : `list`
1838 List of clipped footprints.
1839 clipIndices : `list`
1840 List of which entries in tempExpClipList each footprint belongs to.
1841 maskClipValue
1842 Mask value of clipped pixels.
1843 maskDetValue
1844 Mask value of detected pixels.
1845 coaddBBox : `lsst.geom.Box`
1846 BBox of the coadd and warps.
1848 Returns
1849 -------
1850 bigFootprintsCoadd : `list`
1851 List of big footprints
1852 """
1853 bigFootprintsCoadd = []
1854 ignoreMask = self.getBadPixelMask()
1855 for index, (clippedSpans, visitFootprints) in enumerate(zip(clipList, detectionFootprints)):
1856 maskVisitDet = afwImage.MaskX(coaddBBox, 0x0)
1857 for footprint in visitFootprints.getFootprints():
1858 footprint.spans.setMask(maskVisitDet, maskDetValue)
1860 # build a mask of clipped footprints that are in this visit
1861 clippedFootprintsVisit = []
1862 for foot, clipIndex in zip(clipFootprints, clipIndices):
1863 if index not in clipIndex:
1864 continue
1865 clippedFootprintsVisit.append(foot)
1866 maskVisitClip = maskVisitDet.Factory(maskVisitDet.getBBox(afwImage.PARENT))
1867 afwDet.setMaskFromFootprintList(maskVisitClip, clippedFootprintsVisit, maskClipValue)
1869 bigFootprintsVisit = []
1870 for foot in visitFootprints.getFootprints():
1871 if foot.getArea() < self.config.minBigOverlap:
1872 continue
1873 nCount = countMaskFromFootprint(maskVisitClip, foot, maskClipValue, ignoreMask)
1874 if nCount > self.config.minBigOverlap:
1875 bigFootprintsVisit.append(foot)
1876 bigFootprintsCoadd.append(foot)
1878 for footprint in bigFootprintsVisit:
1879 clippedSpans["CLIPPED"].append(footprint.spans)
1881 return bigFootprintsCoadd
1884class CompareWarpAssembleCoaddConnections(AssembleCoaddConnections):
1885 psfMatchedWarps = pipeBase.connectionTypes.Input(
1886 doc=("PSF-Matched Warps are required by CompareWarp regardless of the coadd type requested. "
1887 "Only PSF-Matched Warps make sense for image subtraction. "
1888 "Therefore, they must be an additional declared input."),
1889 name="{inputCoaddName}Coadd_psfMatchedWarp",
1890 storageClass="ExposureF",
1891 dimensions=("tract", "patch", "skymap", "visit"),
1892 deferLoad=True,
1893 multiple=True
1894 )
1895 templateCoadd = pipeBase.connectionTypes.Output(
1896 doc=("Model of the static sky, used to find temporal artifacts. Typically a PSF-Matched, "
1897 "sigma-clipped coadd. Written if and only if assembleStaticSkyModel.doWrite=True"),
1898 name="{outputCoaddName}CoaddPsfMatched",
1899 storageClass="ExposureF",
1900 dimensions=("tract", "patch", "skymap", "band"),
1901 )
1903 def __init__(self, *, config=None):
1904 super().__init__(config=config)
1905 if not config.assembleStaticSkyModel.doWrite:
1906 self.outputs.remove("templateCoadd")
1907 config.validate()
1910class CompareWarpAssembleCoaddConfig(AssembleCoaddConfig,
1911 pipelineConnections=CompareWarpAssembleCoaddConnections):
1912 assembleStaticSkyModel = pexConfig.ConfigurableField(
1913 target=AssembleCoaddTask,
1914 doc="Task to assemble an artifact-free, PSF-matched Coadd to serve as a"
1915 " naive/first-iteration model of the static sky.",
1916 )
1917 detect = pexConfig.ConfigurableField(
1918 target=SourceDetectionTask,
1919 doc="Detect outlier sources on difference between each psfMatched warp and static sky model"
1920 )
1921 detectTemplate = pexConfig.ConfigurableField(
1922 target=SourceDetectionTask,
1923 doc="Detect sources on static sky model. Only used if doPreserveContainedBySource is True"
1924 )
1925 maskStreaks = pexConfig.ConfigurableField(
1926 target=MaskStreaksTask,
1927 doc="Detect streaks on difference between each psfMatched warp and static sky model. Only used if "
1928 "doFilterMorphological is True. Adds a mask plane to an exposure, with the mask plane name set by"
1929 "streakMaskName"
1930 )
1931 streakMaskName = pexConfig.Field(
1932 dtype=str,
1933 default="STREAK",
1934 doc="Name of mask bit used for streaks"
1935 )
1936 maxNumEpochs = pexConfig.Field(
1937 doc="Charactistic maximum local number of epochs/visits in which an artifact candidate can appear "
1938 "and still be masked. The effective maxNumEpochs is a broken linear function of local "
1939 "number of epochs (N): min(maxFractionEpochsLow*N, maxNumEpochs + maxFractionEpochsHigh*N). "
1940 "For each footprint detected on the image difference between the psfMatched warp and static sky "
1941 "model, if a significant fraction of pixels (defined by spatialThreshold) are residuals in more "
1942 "than the computed effective maxNumEpochs, the artifact candidate is deemed persistant rather "
1943 "than transient and not masked.",
1944 dtype=int,
1945 default=2
1946 )
1947 maxFractionEpochsLow = pexConfig.RangeField(
1948 doc="Fraction of local number of epochs (N) to use as effective maxNumEpochs for low N. "
1949 "Effective maxNumEpochs = "
1950 "min(maxFractionEpochsLow * N, maxNumEpochs + maxFractionEpochsHigh * N)",
1951 dtype=float,
1952 default=0.4,
1953 min=0., max=1.,
1954 )
1955 maxFractionEpochsHigh = pexConfig.RangeField(
1956 doc="Fraction of local number of epochs (N) to use as effective maxNumEpochs for high N. "
1957 "Effective maxNumEpochs = "
1958 "min(maxFractionEpochsLow * N, maxNumEpochs + maxFractionEpochsHigh * N)",
1959 dtype=float,
1960 default=0.03,
1961 min=0., max=1.,
1962 )
1963 spatialThreshold = pexConfig.RangeField(
1964 doc="Unitless fraction of pixels defining how much of the outlier region has to meet the "
1965 "temporal criteria. If 0, clip all. If 1, clip none.",
1966 dtype=float,
1967 default=0.5,
1968 min=0., max=1.,
1969 inclusiveMin=True, inclusiveMax=True
1970 )
1971 doScaleWarpVariance = pexConfig.Field(
1972 doc="Rescale Warp variance plane using empirical noise?",
1973 dtype=bool,
1974 default=True,
1975 )
1976 scaleWarpVariance = pexConfig.ConfigurableField(
1977 target=ScaleVarianceTask,
1978 doc="Rescale variance on warps",
1979 )
1980 doPreserveContainedBySource = pexConfig.Field(
1981 doc="Rescue artifacts from clipping that completely lie within a footprint detected"
1982 "on the PsfMatched Template Coadd. Replicates a behavior of SafeClip.",
1983 dtype=bool,
1984 default=True,
1985 )
1986 doPrefilterArtifacts = pexConfig.Field(
1987 doc="Ignore artifact candidates that are mostly covered by the bad pixel mask, "
1988 "because they will be excluded anyway. This prevents them from contributing "
1989 "to the outlier epoch count image and potentially being labeled as persistant."
1990 "'Mostly' is defined by the config 'prefilterArtifactsRatio'.",
1991 dtype=bool,
1992 default=True
1993 )
1994 prefilterArtifactsMaskPlanes = pexConfig.ListField(
1995 doc="Prefilter artifact candidates that are mostly covered by these bad mask planes.",
1996 dtype=str,
1997 default=('NO_DATA', 'BAD', 'SAT', 'SUSPECT'),
1998 )
1999 prefilterArtifactsRatio = pexConfig.Field(
2000 doc="Prefilter artifact candidates with less than this fraction overlapping good pixels",
2001 dtype=float,
2002 default=0.05
2003 )
2004 doFilterMorphological = pexConfig.Field(
2005 doc="Filter artifact candidates based on morphological criteria, i.g. those that appear to "
2006 "be streaks.",
2007 dtype=bool,
2008 default=False
2009 )
2011 def setDefaults(self):
2012 AssembleCoaddConfig.setDefaults(self)
2013 self.statistic = 'MEAN'
2014 self.doUsePsfMatchedPolygons = True
2016 # Real EDGE removed by psfMatched NO_DATA border half the width of the matching kernel
2017 # CompareWarp applies psfMatched EDGE pixels to directWarps before assembling
2018 if "EDGE" in self.badMaskPlanes:
2019 self.badMaskPlanes.remove('EDGE')
2020 self.removeMaskPlanes.append('EDGE')
2021 self.assembleStaticSkyModel.badMaskPlanes = ["NO_DATA", ]
2022 self.assembleStaticSkyModel.warpType = 'psfMatched'
2023 self.assembleStaticSkyModel.connections.warpType = 'psfMatched'
2024 self.assembleStaticSkyModel.statistic = 'MEANCLIP'
2025 self.assembleStaticSkyModel.sigmaClip = 2.5
2026 self.assembleStaticSkyModel.clipIter = 3
2027 self.assembleStaticSkyModel.calcErrorFromInputVariance = False
2028 self.assembleStaticSkyModel.doWrite = False
2029 self.detect.doTempLocalBackground = False
2030 self.detect.reEstimateBackground = False
2031 self.detect.returnOriginalFootprints = False
2032 self.detect.thresholdPolarity = "both"
2033 self.detect.thresholdValue = 5
2034 self.detect.minPixels = 4
2035 self.detect.isotropicGrow = True
2036 self.detect.thresholdType = "pixel_stdev"
2037 self.detect.nSigmaToGrow = 0.4
2038 # The default nSigmaToGrow for SourceDetectionTask is already 2.4,
2039 # Explicitly restating because ratio with detect.nSigmaToGrow matters
2040 self.detectTemplate.nSigmaToGrow = 2.4
2041 self.detectTemplate.doTempLocalBackground = False
2042 self.detectTemplate.reEstimateBackground = False
2043 self.detectTemplate.returnOriginalFootprints = False
2045 def validate(self):
2046 super().validate()
2047 if self.assembleStaticSkyModel.doNImage:
2048 raise ValueError("No dataset type exists for a PSF-Matched Template N Image."
2049 "Please set assembleStaticSkyModel.doNImage=False")
2051 if self.assembleStaticSkyModel.doWrite and (self.warpType == self.assembleStaticSkyModel.warpType):
2052 raise ValueError("warpType (%s) == assembleStaticSkyModel.warpType (%s) and will compete for "
2053 "the same dataset name. Please set assembleStaticSkyModel.doWrite to False "
2054 "or warpType to 'direct'. assembleStaticSkyModel.warpType should ways be "
2055 "'PsfMatched'" % (self.warpType, self.assembleStaticSkyModel.warpType))
2058class CompareWarpAssembleCoaddTask(AssembleCoaddTask):
2059 """Assemble a compareWarp coadded image from a set of warps
2060 by masking artifacts detected by comparing PSF-matched warps.
2062 In ``AssembleCoaddTask``, we compute the coadd as an clipped mean (i.e.,
2063 we clip outliers). The problem with doing this is that when computing the
2064 coadd PSF at a given location, individual visit PSFs from visits with
2065 outlier pixels contribute to the coadd PSF and cannot be treated correctly.
2066 In this task, we correct for this behavior by creating a new badMaskPlane
2067 'CLIPPED' which marks pixels in the individual warps suspected to contain
2068 an artifact. We populate this plane on the input warps by comparing
2069 PSF-matched warps with a PSF-matched median coadd which serves as a
2070 model of the static sky. Any group of pixels that deviates from the
2071 PSF-matched template coadd by more than config.detect.threshold sigma,
2072 is an artifact candidate. The candidates are then filtered to remove
2073 variable sources and sources that are difficult to subtract such as
2074 bright stars. This filter is configured using the config parameters
2075 ``temporalThreshold`` and ``spatialThreshold``. The temporalThreshold is
2076 the maximum fraction of epochs that the deviation can appear in and still
2077 be considered an artifact. The spatialThreshold is the maximum fraction of
2078 pixels in the footprint of the deviation that appear in other epochs
2079 (where other epochs is defined by the temporalThreshold). If the deviant
2080 region meets this criteria of having a significant percentage of pixels
2081 that deviate in only a few epochs, these pixels have the 'CLIPPED' bit
2082 set in the mask. These regions will not contribute to the final coadd.
2083 Furthermore, any routine to determine the coadd PSF can now be cognizant
2084 of clipped regions. Note that the algorithm implemented by this task is
2085 preliminary and works correctly for HSC data. Parameter modifications and
2086 or considerable redesigning of the algorithm is likley required for other
2087 surveys.
2089 ``CompareWarpAssembleCoaddTask`` sub-classes
2090 ``AssembleCoaddTask`` and instantiates ``AssembleCoaddTask``
2091 as a subtask to generate the TemplateCoadd (the model of the static sky).
2093 Notes
2094 -----
2095 The `lsst.pipe.base.cmdLineTask.CmdLineTask` interface supports a
2096 flag ``-d`` to import ``debug.py`` from your ``PYTHONPATH``; see
2097 ``baseDebug`` for more about ``debug.py`` files.
2099 This task supports the following debug variables:
2101 - ``saveCountIm``
2102 If True then save the Epoch Count Image as a fits file in the `figPath`
2103 - ``figPath``
2104 Path to save the debug fits images and figures
2106 For example, put something like:
2108 .. code-block:: python
2110 import lsstDebug
2111 def DebugInfo(name):
2112 di = lsstDebug.getInfo(name)
2113 if name == "lsst.pipe.tasks.assembleCoadd":
2114 di.saveCountIm = True
2115 di.figPath = "/desired/path/to/debugging/output/images"
2116 return di
2117 lsstDebug.Info = DebugInfo
2119 into your ``debug.py`` file and run ``assemebleCoadd.py`` with the
2120 ``--debug`` flag. Some subtasks may have their own debug variables;
2121 see individual Task documentation.
2123 Examples
2124 --------
2125 ``CompareWarpAssembleCoaddTask`` assembles a set of warped images into a
2126 coadded image. The ``CompareWarpAssembleCoaddTask`` is invoked by running
2127 ``assembleCoadd.py`` with the flag ``--compareWarpCoadd``.
2128 Usage of ``assembleCoadd.py`` expects a data reference to the tract patch
2129 and filter to be coadded (specified using
2130 '--id = [KEY=VALUE1[^VALUE2[^VALUE3...] [KEY=VALUE1[^VALUE2[^VALUE3...] ...]]')
2131 along with a list of coaddTempExps to attempt to coadd (specified using
2132 '--selectId [KEY=VALUE1[^VALUE2[^VALUE3...] [KEY=VALUE1[^VALUE2[^VALUE3...] ...]]').
2133 Only the warps that cover the specified tract and patch will be coadded.
2134 A list of the available optional arguments can be obtained by calling
2135 ``assembleCoadd.py`` with the ``--help`` command line argument:
2137 .. code-block:: none
2139 assembleCoadd.py --help
2141 To demonstrate usage of the ``CompareWarpAssembleCoaddTask`` in the larger
2142 context of multi-band processing, we will generate the HSC-I & -R band
2143 oadds from HSC engineering test data provided in the ``ci_hsc`` package.
2144 To begin, assuming that the lsst stack has been already set up, we must
2145 set up the ``obs_subaru`` and ``ci_hsc`` packages.
2146 This defines the environment variable ``$CI_HSC_DIR`` and points at the
2147 location of the package. The raw HSC data live in the ``$CI_HSC_DIR/raw``
2148 directory. To begin assembling the coadds, we must first
2150 - processCcd
2151 process the individual ccds in $CI_HSC_RAW to produce calibrated exposures
2152 - makeSkyMap
2153 create a skymap that covers the area of the sky present in the raw exposures
2154 - makeCoaddTempExp
2155 warp the individual calibrated exposures to the tangent plane of the coadd
2157 We can perform all of these steps by running
2159 .. code-block:: none
2161 $CI_HSC_DIR scons warp-903986 warp-904014 warp-903990 warp-904010 warp-903988
2163 This will produce warped ``coaddTempExps`` for each visit. To coadd the
2164 warped data, we call ``assembleCoadd.py`` as follows:
2166 .. code-block:: none
2168 assembleCoadd.py --compareWarpCoadd $CI_HSC_DIR/DATA --id patch=5,4 tract=0 filter=HSC-I \
2169 --selectId visit=903986 ccd=16 --selectId visit=903986 ccd=22 --selectId visit=903986 ccd=23 \
2170 --selectId visit=903986 ccd=100 --selectId visit=904014 ccd=1 --selectId visit=904014 ccd=6 \
2171 --selectId visit=904014 ccd=12 --selectId visit=903990 ccd=18 --selectId visit=903990 ccd=25 \
2172 --selectId visit=904010 ccd=4 --selectId visit=904010 ccd=10 --selectId visit=904010 ccd=100 \
2173 --selectId visit=903988 ccd=16 --selectId visit=903988 ccd=17 --selectId visit=903988 ccd=23 \
2174 --selectId visit=903988 ccd=24
2176 This will process the HSC-I band data. The results are written in
2177 ``$CI_HSC_DIR/DATA/deepCoadd-results/HSC-I``.
2178 """
2179 ConfigClass = CompareWarpAssembleCoaddConfig
2180 _DefaultName = "compareWarpAssembleCoadd"
2182 def __init__(self, *args, **kwargs):
2183 AssembleCoaddTask.__init__(self, *args, **kwargs)
2184 self.makeSubtask("assembleStaticSkyModel")
2185 detectionSchema = afwTable.SourceTable.makeMinimalSchema()
2186 self.makeSubtask("detect", schema=detectionSchema)
2187 if self.config.doPreserveContainedBySource:
2188 self.makeSubtask("detectTemplate", schema=afwTable.SourceTable.makeMinimalSchema())
2189 if self.config.doScaleWarpVariance:
2190 self.makeSubtask("scaleWarpVariance")
2191 if self.config.doFilterMorphological:
2192 self.makeSubtask("maskStreaks")
2194 @utils.inheritDoc(AssembleCoaddTask)
2195 def makeSupplementaryDataGen3(self, butlerQC, inputRefs, outputRefs):
2196 """
2197 Generate a templateCoadd to use as a naive model of static sky to
2198 subtract from PSF-Matched warps.
2200 Returns
2201 -------
2202 result : `lsst.pipe.base.Struct`
2203 Result struct with components:
2205 - ``templateCoadd`` : coadded exposure (``lsst.afw.image.Exposure``)
2206 - ``nImage`` : N Image (``lsst.afw.image.Image``)
2207 """
2208 # Ensure that psfMatchedWarps are used as input warps for template generation
2209 staticSkyModelInputRefs = copy.deepcopy(inputRefs)
2210 staticSkyModelInputRefs.inputWarps = inputRefs.psfMatchedWarps
2212 # Because subtasks don't have connections we have to make one.
2213 # The main task's `templateCoadd` is the subtask's `coaddExposure`
2214 staticSkyModelOutputRefs = copy.deepcopy(outputRefs)
2215 if self.config.assembleStaticSkyModel.doWrite:
2216 staticSkyModelOutputRefs.coaddExposure = staticSkyModelOutputRefs.templateCoadd
2217 # Remove template coadd from both subtask's and main tasks outputs,
2218 # because it is handled by the subtask as `coaddExposure`
2219 del outputRefs.templateCoadd
2220 del staticSkyModelOutputRefs.templateCoadd
2222 # A PSF-Matched nImage does not exist as a dataset type
2223 if 'nImage' in staticSkyModelOutputRefs.keys():
2224 del staticSkyModelOutputRefs.nImage
2226 templateCoadd = self.assembleStaticSkyModel.runQuantum(butlerQC, staticSkyModelInputRefs,
2227 staticSkyModelOutputRefs)
2228 if templateCoadd is None:
2229 raise RuntimeError(self._noTemplateMessage(self.assembleStaticSkyModel.warpType))
2231 return pipeBase.Struct(templateCoadd=templateCoadd.coaddExposure,
2232 nImage=templateCoadd.nImage,
2233 warpRefList=templateCoadd.warpRefList,
2234 imageScalerList=templateCoadd.imageScalerList,
2235 weightList=templateCoadd.weightList)
2237 @utils.inheritDoc(AssembleCoaddTask)
2238 def makeSupplementaryData(self, dataRef, selectDataList=None, warpRefList=None):
2239 """
2240 Generate a templateCoadd to use as a naive model of static sky to
2241 subtract from PSF-Matched warps.
2243 Returns
2244 -------
2245 result : `lsst.pipe.base.Struct`
2246 Result struct with components:
2248 - ``templateCoadd``: coadded exposure (``lsst.afw.image.Exposure``)
2249 - ``nImage``: N Image (``lsst.afw.image.Image``)
2250 """
2251 templateCoadd = self.assembleStaticSkyModel.runDataRef(dataRef, selectDataList, warpRefList)
2252 if templateCoadd is None:
2253 raise RuntimeError(self._noTemplateMessage(self.assembleStaticSkyModel.warpType))
2255 return pipeBase.Struct(templateCoadd=templateCoadd.coaddExposure,
2256 nImage=templateCoadd.nImage,
2257 warpRefList=templateCoadd.warpRefList,
2258 imageScalerList=templateCoadd.imageScalerList,
2259 weightList=templateCoadd.weightList)
2261 def _noTemplateMessage(self, warpType):
2262 warpName = (warpType[0].upper() + warpType[1:])
2263 message = """No %(warpName)s warps were found to build the template coadd which is
2264 required to run CompareWarpAssembleCoaddTask. To continue assembling this type of coadd,
2265 first either rerun makeCoaddTempExp with config.make%(warpName)s=True or
2266 coaddDriver with config.makeCoadTempExp.make%(warpName)s=True, before assembleCoadd.
2268 Alternatively, to use another algorithm with existing warps, retarget the CoaddDriverConfig to
2269 another algorithm like:
2271 from lsst.pipe.tasks.assembleCoadd import SafeClipAssembleCoaddTask
2272 config.assemble.retarget(SafeClipAssembleCoaddTask)
2273 """ % {"warpName": warpName}
2274 return message
2276 @utils.inheritDoc(AssembleCoaddTask)
2277 @pipeBase.timeMethod
2278 def run(self, skyInfo, tempExpRefList, imageScalerList, weightList,
2279 supplementaryData, *args, **kwargs):
2280 """Assemble the coadd.
2282 Find artifacts and apply them to the warps' masks creating a list of
2283 alternative masks with a new "CLIPPED" plane and updated "NO_DATA"
2284 plane. Then pass these alternative masks to the base class's `run`
2285 method.
2287 The input parameters ``supplementaryData`` is a `lsst.pipe.base.Struct`
2288 that must contain a ``templateCoadd`` that serves as the
2289 model of the static sky.
2290 """
2292 # Check and match the order of the supplementaryData
2293 # (PSF-matched) inputs to the order of the direct inputs,
2294 # so that the artifact mask is applied to the right warp
2295 dataIds = [ref.dataId for ref in tempExpRefList]
2296 psfMatchedDataIds = [ref.dataId for ref in supplementaryData.warpRefList]
2298 if dataIds != psfMatchedDataIds:
2299 self.log.info("Reordering and or/padding PSF-matched visit input list")
2300 supplementaryData.warpRefList = reorderAndPadList(supplementaryData.warpRefList,
2301 psfMatchedDataIds, dataIds)
2302 supplementaryData.imageScalerList = reorderAndPadList(supplementaryData.imageScalerList,
2303 psfMatchedDataIds, dataIds)
2305 # Use PSF-Matched Warps (and corresponding scalers) and coadd to find artifacts
2306 spanSetMaskList = self.findArtifacts(supplementaryData.templateCoadd,
2307 supplementaryData.warpRefList,
2308 supplementaryData.imageScalerList)
2310 badMaskPlanes = self.config.badMaskPlanes[:]
2311 badMaskPlanes.append("CLIPPED")
2312 badPixelMask = afwImage.Mask.getPlaneBitMask(badMaskPlanes)
2314 result = AssembleCoaddTask.run(self, skyInfo, tempExpRefList, imageScalerList, weightList,
2315 spanSetMaskList, mask=badPixelMask)
2317 # Propagate PSF-matched EDGE pixels to coadd SENSOR_EDGE and INEXACT_PSF
2318 # Psf-Matching moves the real edge inwards
2319 self.applyAltEdgeMask(result.coaddExposure.maskedImage.mask, spanSetMaskList)
2320 return result
2322 def applyAltEdgeMask(self, mask, altMaskList):
2323 """Propagate alt EDGE mask to SENSOR_EDGE AND INEXACT_PSF planes.
2325 Parameters
2326 ----------
2327 mask : `lsst.afw.image.Mask`
2328 Original mask.
2329 altMaskList : `list`
2330 List of Dicts containing ``spanSet`` lists.
2331 Each element contains the new mask plane name (e.g. "CLIPPED
2332 and/or "NO_DATA") as the key, and list of ``SpanSets`` to apply to
2333 the mask.
2334 """
2335 maskValue = mask.getPlaneBitMask(["SENSOR_EDGE", "INEXACT_PSF"])
2336 for visitMask in altMaskList:
2337 if "EDGE" in visitMask:
2338 for spanSet in visitMask['EDGE']:
2339 spanSet.clippedTo(mask.getBBox()).setMask(mask, maskValue)
2341 def findArtifacts(self, templateCoadd, tempExpRefList, imageScalerList):
2342 """Find artifacts.
2344 Loop through warps twice. The first loop builds a map with the count
2345 of how many epochs each pixel deviates from the templateCoadd by more
2346 than ``config.chiThreshold`` sigma. The second loop takes each
2347 difference image and filters the artifacts detected in each using
2348 count map to filter out variable sources and sources that are
2349 difficult to subtract cleanly.
2351 Parameters
2352 ----------
2353 templateCoadd : `lsst.afw.image.Exposure`
2354 Exposure to serve as model of static sky.
2355 tempExpRefList : `list`
2356 List of data references to warps.
2357 imageScalerList : `list`
2358 List of image scalers.
2360 Returns
2361 -------
2362 altMasks : `list`
2363 List of dicts containing information about CLIPPED
2364 (i.e., artifacts), NO_DATA, and EDGE pixels.
2365 """
2367 self.log.debug("Generating Count Image, and mask lists.")
2368 coaddBBox = templateCoadd.getBBox()
2369 slateIm = afwImage.ImageU(coaddBBox)
2370 epochCountImage = afwImage.ImageU(coaddBBox)
2371 nImage = afwImage.ImageU(coaddBBox)
2372 spanSetArtifactList = []
2373 spanSetNoDataMaskList = []
2374 spanSetEdgeList = []
2375 spanSetBadMorphoList = []
2376 badPixelMask = self.getBadPixelMask()
2378 # mask of the warp diffs should = that of only the warp
2379 templateCoadd.mask.clearAllMaskPlanes()
2381 if self.config.doPreserveContainedBySource:
2382 templateFootprints = self.detectTemplate.detectFootprints(templateCoadd)
2383 else:
2384 templateFootprints = None
2386 for warpRef, imageScaler in zip(tempExpRefList, imageScalerList):
2387 warpDiffExp = self._readAndComputeWarpDiff(warpRef, imageScaler, templateCoadd)
2388 if warpDiffExp is not None:
2389 # This nImage only approximates the final nImage because it uses the PSF-matched mask
2390 nImage.array += (numpy.isfinite(warpDiffExp.image.array)
2391 * ((warpDiffExp.mask.array & badPixelMask) == 0)).astype(numpy.uint16)
2392 fpSet = self.detect.detectFootprints(warpDiffExp, doSmooth=False, clearMask=True)
2393 fpSet.positive.merge(fpSet.negative)
2394 footprints = fpSet.positive
2395 slateIm.set(0)
2396 spanSetList = [footprint.spans for footprint in footprints.getFootprints()]
2398 # Remove artifacts due to defects before they contribute to the epochCountImage
2399 if self.config.doPrefilterArtifacts:
2400 spanSetList = self.prefilterArtifacts(spanSetList, warpDiffExp)
2402 # Clear mask before adding prefiltered spanSets
2403 self.detect.clearMask(warpDiffExp.mask)
2404 for spans in spanSetList:
2405 spans.setImage(slateIm, 1, doClip=True)
2406 spans.setMask(warpDiffExp.mask, warpDiffExp.mask.getPlaneBitMask("DETECTED"))
2407 epochCountImage += slateIm
2409 if self.config.doFilterMorphological:
2410 maskName = self.config.streakMaskName
2411 _ = self.maskStreaks.run(warpDiffExp)
2412 streakMask = warpDiffExp.mask
2413 spanSetStreak = afwGeom.SpanSet.fromMask(streakMask,
2414 streakMask.getPlaneBitMask(maskName)).split()
2416 # PSF-Matched warps have less available area (~the matching kernel) because the calexps
2417 # undergo a second convolution. Pixels with data in the direct warp
2418 # but not in the PSF-matched warp will not have their artifacts detected.
2419 # NaNs from the PSF-matched warp therefore must be masked in the direct warp
2420 nans = numpy.where(numpy.isnan(warpDiffExp.maskedImage.image.array), 1, 0)
2421 nansMask = afwImage.makeMaskFromArray(nans.astype(afwImage.MaskPixel))
2422 nansMask.setXY0(warpDiffExp.getXY0())
2423 edgeMask = warpDiffExp.mask
2424 spanSetEdgeMask = afwGeom.SpanSet.fromMask(edgeMask,
2425 edgeMask.getPlaneBitMask("EDGE")).split()
2426 else:
2427 # If the directWarp has <1% coverage, the psfMatchedWarp can have 0% and not exist
2428 # In this case, mask the whole epoch
2429 nansMask = afwImage.MaskX(coaddBBox, 1)
2430 spanSetList = []
2431 spanSetEdgeMask = []
2432 spanSetStreak = []
2434 spanSetNoDataMask = afwGeom.SpanSet.fromMask(nansMask).split()
2436 spanSetNoDataMaskList.append(spanSetNoDataMask)
2437 spanSetArtifactList.append(spanSetList)
2438 spanSetEdgeList.append(spanSetEdgeMask)
2439 if self.config.doFilterMorphological:
2440 spanSetBadMorphoList.append(spanSetStreak)
2442 if lsstDebug.Info(__name__).saveCountIm:
2443 path = self._dataRef2DebugPath("epochCountIm", tempExpRefList[0], coaddLevel=True)
2444 epochCountImage.writeFits(path)
2446 for i, spanSetList in enumerate(spanSetArtifactList):
2447 if spanSetList:
2448 filteredSpanSetList = self.filterArtifacts(spanSetList, epochCountImage, nImage,
2449 templateFootprints)
2450 spanSetArtifactList[i] = filteredSpanSetList
2451 if self.config.doFilterMorphological:
2452 spanSetArtifactList[i] += spanSetBadMorphoList[i]
2454 altMasks = []
2455 for artifacts, noData, edge in zip(spanSetArtifactList, spanSetNoDataMaskList, spanSetEdgeList):
2456 altMasks.append({'CLIPPED': artifacts,
2457 'NO_DATA': noData,
2458 'EDGE': edge})
2459 return altMasks
2461 def prefilterArtifacts(self, spanSetList, exp):
2462 """Remove artifact candidates covered by bad mask plane.
2464 Any future editing of the candidate list that does not depend on
2465 temporal information should go in this method.
2467 Parameters
2468 ----------
2469 spanSetList : `list`
2470 List of SpanSets representing artifact candidates.
2471 exp : `lsst.afw.image.Exposure`
2472 Exposure containing mask planes used to prefilter.
2474 Returns
2475 -------
2476 returnSpanSetList : `list`
2477 List of SpanSets with artifacts.
2478 """
2479 badPixelMask = exp.mask.getPlaneBitMask(self.config.prefilterArtifactsMaskPlanes)
2480 goodArr = (exp.mask.array & badPixelMask) == 0
2481 returnSpanSetList = []
2482 bbox = exp.getBBox()
2483 x0, y0 = exp.getXY0()
2484 for i, span in enumerate(spanSetList):
2485 y, x = span.clippedTo(bbox).indices()
2486 yIndexLocal = numpy.array(y) - y0
2487 xIndexLocal = numpy.array(x) - x0
2488 goodRatio = numpy.count_nonzero(goodArr[yIndexLocal, xIndexLocal])/span.getArea()
2489 if goodRatio > self.config.prefilterArtifactsRatio:
2490 returnSpanSetList.append(span)
2491 return returnSpanSetList
2493 def filterArtifacts(self, spanSetList, epochCountImage, nImage, footprintsToExclude=None):
2494 """Filter artifact candidates.
2496 Parameters
2497 ----------
2498 spanSetList : `list`
2499 List of SpanSets representing artifact candidates.
2500 epochCountImage : `lsst.afw.image.Image`
2501 Image of accumulated number of warpDiff detections.
2502 nImage : `lsst.afw.image.Image`
2503 Image of the accumulated number of total epochs contributing.
2505 Returns
2506 -------
2507 maskSpanSetList : `list`
2508 List of SpanSets with artifacts.
2509 """
2511 maskSpanSetList = []
2512 x0, y0 = epochCountImage.getXY0()
2513 for i, span in enumerate(spanSetList):
2514 y, x = span.indices()
2515 yIdxLocal = [y1 - y0 for y1 in y]
2516 xIdxLocal = [x1 - x0 for x1 in x]
2517 outlierN = epochCountImage.array[yIdxLocal, xIdxLocal]
2518 totalN = nImage.array[yIdxLocal, xIdxLocal]
2520 # effectiveMaxNumEpochs is broken line (fraction of N) with characteristic config.maxNumEpochs
2521 effMaxNumEpochsHighN = (self.config.maxNumEpochs
2522 + self.config.maxFractionEpochsHigh*numpy.mean(totalN))
2523 effMaxNumEpochsLowN = self.config.maxFractionEpochsLow * numpy.mean(totalN)
2524 effectiveMaxNumEpochs = int(min(effMaxNumEpochsLowN, effMaxNumEpochsHighN))
2525 nPixelsBelowThreshold = numpy.count_nonzero((outlierN > 0)
2526 & (outlierN <= effectiveMaxNumEpochs))
2527 percentBelowThreshold = nPixelsBelowThreshold / len(outlierN)
2528 if percentBelowThreshold > self.config.spatialThreshold:
2529 maskSpanSetList.append(span)
2531 if self.config.doPreserveContainedBySource and footprintsToExclude is not None:
2532 # If a candidate is contained by a footprint on the template coadd, do not clip
2533 filteredMaskSpanSetList = []
2534 for span in maskSpanSetList:
2535 doKeep = True
2536 for footprint in footprintsToExclude.positive.getFootprints():
2537 if footprint.spans.contains(span):
2538 doKeep = False
2539 break
2540 if doKeep:
2541 filteredMaskSpanSetList.append(span)
2542 maskSpanSetList = filteredMaskSpanSetList
2544 return maskSpanSetList
2546 def _readAndComputeWarpDiff(self, warpRef, imageScaler, templateCoadd):
2547 """Fetch a warp from the butler and return a warpDiff.
2549 Parameters
2550 ----------
2551 warpRef : `lsst.daf.persistence.butlerSubset.ButlerDataRef`
2552 Butler dataRef for the warp.
2553 imageScaler : `lsst.pipe.tasks.scaleZeroPoint.ImageScaler`
2554 An image scaler object.
2555 templateCoadd : `lsst.afw.image.Exposure`
2556 Exposure to be substracted from the scaled warp.
2558 Returns
2559 -------
2560 warp : `lsst.afw.image.Exposure`
2561 Exposure of the image difference between the warp and template.
2562 """
2564 # If the PSF-Matched warp did not exist for this direct warp
2565 # None is holding its place to maintain order in Gen 3
2566 if warpRef is None:
2567 return None
2568 # Warp comparison must use PSF-Matched Warps regardless of requested coadd warp type
2569 warpName = self.getTempExpDatasetName('psfMatched')
2570 if not isinstance(warpRef, DeferredDatasetHandle):
2571 if not warpRef.datasetExists(warpName):
2572 self.log.warning("Could not find %s %s; skipping it", warpName, warpRef.dataId)
2573 return None
2574 warp = warpRef.get(datasetType=warpName, immediate=True)
2575 # direct image scaler OK for PSF-matched Warp
2576 imageScaler.scaleMaskedImage(warp.getMaskedImage())
2577 mi = warp.getMaskedImage()
2578 if self.config.doScaleWarpVariance:
2579 try:
2580 self.scaleWarpVariance.run(mi)
2581 except Exception as exc:
2582 self.log.warning("Unable to rescale variance of warp (%s); leaving it as-is", exc)
2583 mi -= templateCoadd.getMaskedImage()
2584 return warp
2586 def _dataRef2DebugPath(self, prefix, warpRef, coaddLevel=False):
2587 """Return a path to which to write debugging output.
2589 Creates a hyphen-delimited string of dataId values for simple filenames.
2591 Parameters
2592 ----------
2593 prefix : `str`
2594 Prefix for filename.
2595 warpRef : `lsst.daf.persistence.butlerSubset.ButlerDataRef`
2596 Butler dataRef to make the path from.
2597 coaddLevel : `bool`, optional.
2598 If True, include only coadd-level keys (e.g., 'tract', 'patch',
2599 'filter', but no 'visit').
2601 Returns
2602 -------
2603 result : `str`
2604 Path for debugging output.
2605 """
2606 if coaddLevel:
2607 keys = warpRef.getButler().getKeys(self.getCoaddDatasetName(self.warpType))
2608 else:
2609 keys = warpRef.dataId.keys()
2610 keyList = sorted(keys, reverse=True)
2611 directory = lsstDebug.Info(__name__).figPath if lsstDebug.Info(__name__).figPath else "."
2612 filename = "%s-%s.fits" % (prefix, '-'.join([str(warpRef.dataId[k]) for k in keyList]))
2613 return os.path.join(directory, filename)