def cloudMask(date1, date2):
# dateFns have completed therefore assume that, at least, the
# quadscenes have been created in projectStorage
# (i.e., sceneID1U*, sceneID1L*, sceneID2U*, and sceneID2L* directories are populuated)
# eros_data may or may not be present
# cloudMask dependencies are on files in tiffStorage, /lsfdata/eros_data/extractedTars/
outBasename = date1.sceneID + "_" + date2.sceneID + '_Fmask.tif'
wrs2Name=date1.sceneID[3:9]
if not os.path.exists(date1.folder+"/"+date1.sceneID+"_MTLFmask.TIF"):
rasterAnalysis_GDAL.cloudMask(os.path.join(LSF.tiffsStorage, date1.sceneID[:-2]))
# create quads from the input scene
quadPaths = rasterAnalysis_GDAL.cropToQuad(os.path.join(LSF.tiffsStorage, date1.sceneID[:-2]), LSF.projectStorage, LSF.quadsFolder)
landsatFactTools_GDAL.writeQuadToDB(quadPaths)
# get cloud cover percentage for each quad
quadCCDict = landsatFactTools_GDAL.getQuadCCpercent(quadPaths)
# =========================================================================
# write input scene quads cloud cover percentage to the landsat_metadata table in the database
landsatFactTools_GDAL.writeQuadsCCtoDB(quadCCDict,date1.folder.replace('UR','').replace('UL','').replace('LR','').replace('LL',''))
if not os.path.exists(date2.folder+"/"+date2.sceneID+"_MTLFmask.TIF"):
rasterAnalysis_GDAL.cloudMask(os.path.join(LSF.tiffsStorage, date2.sceneID[:-2]))
# create quads from the input scene
quadPaths = rasterAnalysis_GDAL.cropToQuad(os.path.join(LSF.tiffsStorage, date2.sceneID[:-2]), LSF.projectStorage, LSF.quadsFolder)
landsatFactTools_GDAL.writeQuadToDB(quadPaths)
# get cloud cover percentage for each quad
quadCCDict = landsatFactTools_GDAL.getQuadCCpercent(quadPaths)
# =========================================================================
# write input scene quads cloud cover percentage to the landsat_metadata table in the database
landsatFactTools_GDAL.writeQuadsCCtoDB(quadCCDict,date1.folder.replace('UR','').replace('UL','').replace('LR','').replace('LL',''))
outputTiffName=os.path.join(LSF.outFMASKfolder,outBasename)
if not os.path.exists(outputTiffName):
if os.path.exists(date1.folder+"/"+date1.sceneID+"_MTLFmask.TIF") and os.path.exists(date2.folder+"/"+date2.sceneID+"_MTLFmask.TIF"):
qaTiffName=os.path.join(LSF.tiffsStorage, date1.sceneID[:-2], date1.sceneID[:-2]) + "_BQA.TIF"
if os.path.exists(qaTiffName):
cloud_mask_type='BQA'
else:
cloud_mask_type='FMASK'
FmaskReclassedArray1 = date1.cloudMaskArray()
FmaskReclassedArray2 = date2.cloudMaskArray()
FmaskReclassedArray = FmaskReclassedArray1 * FmaskReclassedArray2
FmaskReclassedArrayPlus1 = FmaskReclassedArray + 1
shpName=os.path.join(LSF.quadsFolder, 'wrs2_'+ wrs2Name + date1.folder[-2:]+'.shp')
LSFGeoTIFF.Unsigned8BitLSFGeoTIFF.fromArray(FmaskReclassedArrayPlus1, date1.geoTiffAtts).write(outputTiffName, shpName)
landsatFactTools_GDAL.writeProductToDB(os.path.basename(outputTiffName),date1.sceneID,date2.sceneID,'CLOUD',date2.sceneID[9:16], 'CR',cloud_mask_type)
# if the product's been created, should be a row in products
# note if the row is missing
else:
resultRowExists = rowExists(outBasename, 'products', 'product_id')
if resultRowExists == False:
print "No row for {} in products".format(outBasename)
def extractProductForCompare(diff_tar,tarStorage,tiffsStorage,fmaskShellCall,quadsFolder,outRasterFolder):
print "call to extractProductForCompare with: "+ diff_tar
try:
# call download_landsat_data_by_sceneid.php using the downloadScene function
# in the event of a DownloadError, retry downloadScene up to 5 times, with a 5 second delay btw calls
retry(1, 5, DownloadError,downloadScene, diff_tar)
# now extract the downloaded file accordingly
extractedPath = checkExisting(os.path.join(LSF.tarStorage, diff_tar+".tar.gz"), tiffsStorage)
#do the other pre-processing stuff
rasterAnalysis_GDAL.runFmask(extractedPath,LSF.fmaskShellCall)
# get DN min number from each band in the scene and write to database
dnminExists = checkForDNminExist(extractedPath) # May not be needed in final design, used during testing
if dnminExists == False:
dnMinDict = rasterAnalysis_GDAL.getDNmin(extractedPath)
# fix bug introduced by me in commit de5df07c4ca3ff71ae4d7da27b6018fe1bc2df04
writeDNminToDB(dnMinDict,extractedPath)
# create quads from the input scene
quadPaths = rasterAnalysis_GDAL.cropToQuad(extractedPath,outRasterFolder,quadsFolder)
writeQuadToDB(quadPaths)
# get cloud cover percentage for each quad
quadCCDict = getQuadCCpercent(quadPaths)
# =========================================================================
# write input scene quads cloud cover percentage to the landsat_metadata table in the database
writeQuadsCCtoDB(quadCCDict,extractedPath)
print os.getcwd()
except Exception as e:
print "Error in extractProductForCompare"
# make sure we've returned to the original directory before raising this exception
raise
def quadScene(quadsceneID):
# this function is a little more complicated because it tries to take one of two possible shortcuts.
# if there are clipped TIFFs, the previous work artifacts can safely no longer be there.
# if the clipped TIFFs aren't there, look for unclipped TIFFs in tiffsStorage first.
# if the unclipped TIFFs aren't there, the extractedTar "target" function will cause the tar to be downloaded
sceneID=quadsceneID[:-2]
if not(re.search(quadsceneID, ' '.join(glob.glob(os.path.join(LSF.projectStorage, '*'))))):
# if the band files for quadsceneID in are not in projectStorage, get them from tiffsStorage, if possible, and clip them
if not (re.search(sceneID, ' '.join(glob.glob(os.path.join(LSF.tiffsStorage, '*'))))):
extractedTar(quadsceneID)
quadPaths = rasterAnalysis_GDAL.cropToQuad(os.path.join(LSF.tiffsStorage, sceneID), LSF.projectStorage, LSF.quadsFolder)
landsatFactTools_GDAL.writeQuadToDB(quadPaths)
# get cloud cover percentage for each quad
quadCCDict = landsatFactTools_GDAL.getQuadCCpercent(quadPaths)
# =========================================================================
# write input scene quads cloud cover percentage to the landsat_metadata table in the database
landsatFactTools_GDAL.writeQuadsCCtoDB(quadCCDict,os.path.join(LSF.tiffsStorage, sceneID))
# if the quad's been processed, should at least be a row in extracted_imagery
# note if the row is missing
else:
resultRowExists = rowExists(quadsceneID, 'extracted_imagery', 'quad_scene')
if resultRowExists == False:
print "No row for {} in extracted_imagery".format(quadsceneID)
# run Fmask
# jdm 4/22/15: after spending a couple of days trying to get FMASK installed on cloud4
# I have not been able to get it to work. Therefore, for now I am commenting this out
# rasterAnalysis_GDAL.runFmask(extractedPath,Fmaskexe) #BM's original
print extractedPath
runFmaskBool = rasterAnalysis_GDAL.runFmask(extractedPath,fmaskShellCall)
#print "Fmask Boolean: "+runFmaskBool
if (runFmaskBool == True):
# get DN min number from each band in the scene and write to database
wrs2Name=tar[3:9]
dnminExists = landsatFactTools_GDAL.checkForDNminExist(extractedPath) # May not be needed in final design, used during testing
if dnminExists == False:
dnMinDict = rasterAnalysis_GDAL.getDNmin(extractedPath)
landsatFactTools_GDAL.writeDNminToDB(dnMinDict,extractedPath)
# create quads from the input scene
quadPaths = rasterAnalysis_GDAL.cropToQuad(extractedPath,projectStorage,quadsFolder)
landsatFactTools_GDAL.writeQuadToDB(quadPaths)
# get cloud cover percentage for each quad
# write input scene quads cloud cover percentage to the landsat_metadata table in the database
quadCCDict=landsatFactTools_GDAL.readAndWriteQuadCC(quadPaths, extractedPath)
# for each quad this finds the closest scene that passes the cloud cover threshold for processing
quadTiffList2Process = landsatFactTools_GDAL.getNextBestQuad(quadCCDict,cloudCoverThreshold)
# =========================================================================
#jdm: Need to make sure the next best quads we are going to be comparing to are actually extracted
#and pre-processed to a level appropriate for differencing.
print "quadTiffList2Process: ", quadTiffList2Process
#TO-DO: create/call a function in landsatFactTools_GDAL called extractProductForCompare()
#loop through quadTiffList2Process and get unique list of data to download
for quad_pair in quadTiffList2Process:
base_quad = quad_pair[1]
diff_quad = quad_pair[0]
inNewSceneTar = os.path.join(tarStorage, tar) # check to see if the tar file has been extracted, if not extract the files extractedPath = landsatFactTools_GDAL.checkExisting(inNewSceneTar, tiffsStorage) # run Fmask # jdm 4/22/15: after spending a couple of days trying to get FMASK installed on cloud4 # I have not been able to get it to work. Therefore, for now I am commenting this out # rasterAnalysis_GDAL.runFmask(extractedPath,Fmaskexe) #BM's original print extractedPath rasterAnalysis_GDAL.runFmask(extractedPath,fmaskShellCall) # get DN min number from each band in the scene and write to database dnminExists = landsatFactTools_GDAL.checkForDNminExist(extractedPath) # May not be needed in final design, used during testing if dnminExists == False: dnMinDict = rasterAnalysis_GDAL.getDNmin(extractedPath) landsatFactTools_GDAL.writeDNminToDB(dnMinDict,extractedPath) # create quads from the input scene quadPaths = rasterAnalysis_GDAL.cropToQuad(extractedPath,quadsFolder) landsatFactTools_GDAL.writeQuadToDB(quadPaths) # get cloud cover percentage for each quad quadCCDict = landsatFactTools_GDAL.getQuadCCpercent(quadPaths) # ========================================================================= # write input scene quads cloud cover percentage to the landsat_metadata table in the database landsatFactTools_GDAL.writeQuadsCCtoDB(quadCCDict,extractedPath) # for each quad this finds the closest scene that passes the cloud cover threshold for processing quadTiffList2Process = landsatFactTools_GDAL.getNextBestQuad(quadCCDict,cloudCoverThreshold) # ========================================================================= #print "quadTiffList2Process: ", quadTiffList2Process # checks the list of quads, if 0 then there were no quads under the cloud cover threshold # so there is nothing to process for that scene if len(quadTiffList2Process) > 0: #print "quadTiffList2Process: ", quadTiffList2Process # for each quad pair perform the change analysis