# (2.5 or 3.5 generally is sufficient)
dilatePixels = 2.5
# Choose the resampling method: 'near', 'bilinear', or 'bicubic'
# Defaults to 'near'
# If method other than 'near' is chosen, any map drawn on the fly that is not
# reprojected, will appear blurred
# Use .reproject to view the actual resulting image (this will slow it down)
resampleMethod = 'near'
# If available, bring in preComputed cloudScore offsets and TDOM stats
# Set to null if computing on-the-fly is wanted
# These have been pre-computed for all CONUS for Landsat and Setinel 2 (separately)
# and are appropriate to use for any time period within the growing season
# The cloudScore offset is generally some lower percentile of cloudScores on a pixel-wise basis
preComputedCloudScoreOffset = getImagesLib.getPrecomputedCloudScoreOffsets(
cloudScorePctl)['landsat']
# The TDOM stats are the mean and standard deviations of the two IR bands used in TDOM
# By default, TDOM uses the nir and swir1 bands
preComputedTDOMStats = getImagesLib.getPrecomputedTDOMStats()
preComputedTDOMIRMean = preComputedTDOMStats['landsat']['mean']
preComputedTDOMIRStdDev = preComputedTDOMStats['landsat']['stdDev']
# correctIllumination: Choose if you want to correct the illumination using
# Sun-Canopy-Sensor+C correction. Additionally, choose the scale at which the
# correction is calculated in meters.
correctIllumination = False
correctScale = 250 #Choose a scale to reduce on- 250 generally works well
# Export params
# Whether to export composites
def getTimeSeriesSample(startYear,
endYear,
startJulian,
endJulian,
compositePeriod,
exportBands,
studyArea,
nSamples,
output_table_name,
showGEEViz,
maskSnow=False,
programs=['Landsat', 'Sentinel2']):
check_dir(os.path.dirname(output_table_name))
#If available, bring in preComputed cloudScore offsets and TDOM stats
#Set to null if computing on-the-fly is wanted
#These have been pre-computed for all CONUS for Landsat and Setinel 2 (separately)
#and are appropriate to use for any time period within the growing season
#The cloudScore offset is generally some lower percentile of cloudScores on a pixel-wise basis
preComputedCloudScoreOffset = getImagesLib.getPrecomputedCloudScoreOffsets(
10)
preComputedLandsatCloudScoreOffset = preComputedCloudScoreOffset['landsat']
preComputedSentinel2CloudScoreOffset = preComputedCloudScoreOffset[
'sentinel2']
#The TDOM stats are the mean and standard deviations of the two bands used in TDOM
#By default, TDOM uses the nir and swir1 bands
preComputedTDOMStats = getImagesLib.getPrecomputedTDOMStats()
preComputedLandsatTDOMIRMean = preComputedTDOMStats['landsat']['mean']
preComputedLandsatTDOMIRStdDev = preComputedTDOMStats['landsat']['stdDev']
preComputedSentinel2TDOMIRMean = preComputedTDOMStats['sentinel2']['mean']
preComputedSentinel2TDOMIRStdDev = preComputedTDOMStats['sentinel2'][
'stdDev']
#####################################################################################
#Function Calls
#Get all images
try:
saBounds = studyArea.geometry().bounds()
except:
saBounds = studyArea.bounds()
#Sample the study area
randomSample = ee.FeatureCollection.randomPoints(studyArea, nSamples, 0,
50)
Map.addLayer(randomSample, {"layerType": "geeVector"}, 'Samples', True)
dummyImage = None
for yr in range(startYear, endYear + 1):
output_table_nameT = '{}_{}_{}_{}-{}_{}_{}{}'.format(
os.path.splitext(output_table_name)[0], '-'.join(programs), yr,
startJulian, endJulian, compositePeriod, nSamples,
os.path.splitext(output_table_name)[1])
if not os.path.exists(output_table_nameT):
if 'Landsat' in programs and 'Sentinel2' in programs:
if dummyImage == None:
dummyImage = ee.Image(getImagesLib.getProcessedLandsatAndSentinel2Scenes(saBounds,\
2019,\
2020,\
1,\
365).first())
images = getImagesLib.getProcessedLandsatAndSentinel2Scenes(saBounds,\
yr,\
yr,\
startJulian,\
endJulian,\
toaOrSR = 'TOA',
includeSLCOffL7 = True,
)
elif 'Sentinel2' in programs:
if dummyImage == None:
dummyImage = ee.Image(getImagesLib.getProcessedSentinel2Scenes(saBounds,\
2019,\
2020,\
1,\
365).first())
images = getImagesLib.etProcessedSentinel2Scenes(saBounds,\
yr,\
yr,\
startJulian,\
endJulian)
elif 'Landsat' in programs:
if dummyImage == None:
dummyImage = ee.Image(getImagesLib.getProcessedLandsatScenes(saBounds,\
2019,\
2020,\
1,\
365).first())
images = getImagesLib.getProcessedLandsatScenes(
saBounds,\
yr,\
yr,\
startJulian,\
endJulian,\
toaOrSR = 'TOA',
includeSLCOffL7 = True)
images = getImagesLib.fillEmptyCollections(images, dummyImage)
#Vizualize the median of the images
# Map.addLayer(images.median(),vizParamsFalse,'Median Composite '+str(yr),False)
#Mask snow
if maskSnow:
print('Masking snow')
images = images.map(getImagesLib.sentinel2SnowMask)
#Add greenness ratio/indices
images = images.map(getImagesLib.HoCalcAlgorithm2)
# Map.addLayer(images.select(exportBands),{},'Raw Time Series ' + str(yr),True)
#Convert to n day composites
composites = getImagesLib.nDayComposites(images, yr, yr, 1, 365,
compositePeriod)
# Map.addLayer(composites.select(exportBands),{},str(compositePeriod) +' day composites '+str(yr))
#Convert to a stack
stack = composites.select(exportBands).toBands()
#Fix band names to be yyyy_mm_dd
bns = stack.bandNames()
bns = bns.map(
lambda bn: ee.String(bn).split('_').slice(1, None).join('_'))
stack = stack.rename(bns)
#Start export table thread
tt = threading.Thread(target=getTableWrapper,
args=(stack, randomSample,
output_table_nameT))
tt.start()
time.sleep(0.1)
threadLimit = 1
if showGEEViz:
#Vizualize outputs
Map.addLayer(studyArea, {'strokeColor': '00F'}, 'Study Area')
Map.centerObject(studyArea)
Map.view()
threadLimit = 2
limitThreads(threadLimit)
# Defaults to 'aggregate'
# Aggregate is generally useful for aggregating pixels when reprojecting instead of resampling
# A good example would be reprojecting S2 data to 30 m
# If method other than 'near' is chosen, any map drawn on the fly that is not
# reprojected, will appear blurred or not really represented properly
# Use .reproject to view the actual resulting image (this will slow it down)
resampleMethod = 'aggregate'
# If available, bring in preComputed cloudScore offsets and TDOM stats
# Set to None if computing on-the-fly is wanted
# These have been pre-computed for all CONUS for Landsat and Setinel 2 (separately)
# and are appropriate to use for any time period within the growing season
# The cloudScore offset is generally some lower percentile of cloudScores on a pixel-wise basis
preComputedCloudScoreOffset = getImagesLib.getPrecomputedCloudScoreOffsets(
cloudScorePctl)['sentinel2']
# The TDOM stats are the mean and standard deviations of the two IR bands used in TDOM
# By default, TDOM uses the nir and swir1 bands
preComputedTDOMStats = getImagesLib.getPrecomputedTDOMStats()
preComputedTDOMIRMean = preComputedTDOMStats['sentinel2']['mean']
preComputedTDOMIRStdDev = preComputedTDOMStats['sentinel2']['stdDev']
# Export params
# Whether to export composites
exportComposites = False
# Set up Names for the export
outputName = 'Sentinel2'
#final deliverables for local viewing #Intended to work within the geeViz package # python -m pip install geeViz #Also requires gdal #################################################################################################### import geeViz.getImagesLib as getImagesLib import geeViz.taskManagerLib as tml ee = getImagesLib.ee Map = getImagesLib.Map import make_esri_viewer as mew import json, os, threading, time, glob from osgeo import gdal ################################################## #Bring in some existing assets to use for cloud masking preComputedCloudScoreOffset = getImagesLib.getPrecomputedCloudScoreOffsets(10) preComputedLandsatCloudScoreOffset = preComputedCloudScoreOffset['landsat'] preComputedSentinel2CloudScoreOffset = preComputedCloudScoreOffset['sentinel2'] #The TDOM stats are the mean and standard deviations of the two bands used in TDOM #By default, TDOM uses the nir and swir1 bands preComputedTDOMStats = getImagesLib.getPrecomputedTDOMStats() preComputedLandsatTDOMIRMean = preComputedTDOMStats['landsat']['mean'] preComputedLandsatTDOMIRStdDev = preComputedTDOMStats['landsat']['stdDev'] preComputedSentinel2TDOMIRMean = preComputedTDOMStats['sentinel2']['mean'] preComputedSentinel2TDOMIRStdDev = preComputedTDOMStats['sentinel2']['stdDev'] ############################################################################ #On-the-fly basic water masking method