def create_tasks(self):
x_list = range(self.x_min, self.x_max + 1)
y_list = range(self.y_min, self.y_max + 1)
dataset_types = [self.dataset_type]
if self.mask_pqa_apply:
dataset_types.append(DatasetType.PQ25)
from itertools import product
for (acq_min, acq_max), season in product(self.get_epochs(), self.get_seasons()):
_log.info("%s %s %s", acq_min, acq_max, season)
acq_min_extended, acq_max_extended, criteria = build_season_date_criteria(acq_min, acq_max, season,
seasons=SEASONS,
extend=True)
_log.info("\tcriteria is %s", criteria)
for cell in list_cells_as_generator(x=x_list, y=y_list, satellites=self.satellites,
acq_min=acq_min_extended, acq_max=acq_max_extended,
dataset_types=dataset_types, include=criteria):
_log.info("\t%3d %4d", cell.x, cell.y)
yield self.create_task(x=cell.x, y=cell.y, acq_min=acq_min, acq_max=acq_max, season=season)
def create_tasks(self):
x_list = range(self.x_min, self.x_max + 1)
y_list = range(self.y_min, self.y_max + 1)
dataset_types = [self.dataset_type]
if self.mask_pqa_apply:
dataset_types.append(DatasetType.PQ25)
from itertools import product
for (acq_min, acq_max), season in product(self.get_epochs(),
self.get_seasons()):
_log.info("%s %s %s", acq_min, acq_max, season)
acq_min_extended, acq_max_extended, criteria = build_season_date_criteria(
acq_min, acq_max, season, seasons=SEASONS, extend=True)
_log.info("\tcriteria is %s", criteria)
for cell in list_cells_as_generator(x=x_list,
y=y_list,
satellites=self.satellites,
acq_min=acq_min_extended,
acq_max=acq_max_extended,
dataset_types=dataset_types,
include=criteria):
_log.info("\t%3d %4d", cell.x, cell.y)
yield self.create_task(x=cell.x,
y=cell.y,
acq_min=acq_min,
acq_max=acq_max,
season=season)
def run(self):
_log.info("Creating stack for band [%s]", self.band.name)
data_type = get_dataset_type_datatype(self.dataset_type)
ndv = get_dataset_type_ndv(self.dataset_type)
metadata = None
driver = None
raster = None
acq_min, acq_max, criteria = build_season_date_criteria(self.acq_min, self.acq_max, self.season,
seasons=SEASONS, extend=True)
_log.info("\tacq %s to %s criteria is %s", acq_min, acq_max, criteria)
dataset_types = [self.dataset_type]
if self.mask_pqa_apply:
dataset_types.append(DatasetType.PQ25)
tiles = list_tiles_as_list(x=[self.x], y=[self.y], satellites=self.satellites,
acq_min=acq_min, acq_max=acq_max,
dataset_types=dataset_types, include=criteria)
for index, tile in enumerate(tiles, start=1):
dataset = tile.datasets[self.dataset_type]
assert dataset
# band = dataset.bands[self.band]
# assert band
band = self.band
pqa = (self.mask_pqa_apply and DatasetType.PQ25 in tile.datasets) and tile.datasets[DatasetType.PQ25] or None
if self.dataset_type not in tile.datasets:
_log.debug("No [%s] dataset present for [%s] - skipping", self.dataset_type.name, tile.end_datetime)
continue
filename = self.output().path
if not metadata:
metadata = get_dataset_metadata(dataset)
assert metadata
if not driver:
if self.output_format == OutputFormat.GEOTIFF:
driver = gdal.GetDriverByName("GTiff")
elif self.output_format == OutputFormat.ENVI:
driver = gdal.GetDriverByName("ENVI")
assert driver
if not raster:
if self.output_format == OutputFormat.GEOTIFF:
raster = driver.Create(filename, metadata.shape[0], metadata.shape[1], len(tiles), data_type, options=["BIGTIFF=YES", "INTERLEAVE=BAND"])
elif self.output_format == OutputFormat.ENVI:
raster = driver.Create(filename, metadata.shape[0], metadata.shape[1], len(tiles), data_type, options=["INTERLEAVE=BSQ"])
assert raster
# NOTE: could do this without the metadata!!
raster.SetGeoTransform(metadata.transform)
raster.SetProjection(metadata.projection)
raster.SetMetadata(self.generate_raster_metadata())
mask = None
if pqa:
mask = get_mask_pqa(pqa, self.mask_pqa_mask, mask=mask)
_log.info("Stacking [%s] band data from [%s] with PQA [%s] and PQA mask [%s] to [%s]",
band.name, dataset.path,
pqa and pqa.path or "", pqa and self.mask_pqa_mask or "",
filename)
data = get_dataset_data_masked(dataset, mask=mask, ndv=ndv)
_log.debug("data is [%s]", data)
stack_band = raster.GetRasterBand(index)
stack_band.SetDescription(os.path.basename(dataset.path))
stack_band.SetNoDataValue(ndv)
stack_band.WriteArray(data[band])
stack_band.ComputeStatistics(True)
stack_band.SetMetadata({"ACQ_DATE": format_date(tile.end_datetime), "SATELLITE": dataset.satellite.name})
stack_band.FlushCache()
del stack_band
if raster:
raster.FlushCache()
del raster
raster = None
def run(self):
_log.info("Calculating statistics for chunk")
ndv = get_dataset_type_ndv(self.dataset_type)
acq_min, acq_max, criteria = build_season_date_criteria(self.acq_min, self.acq_max, self.season,
seasons=SEASONS, extend=True)
_log.info("\tcriteria is %s", criteria)
dataset_types = [self.dataset_type]
if self.mask_pqa_apply:
dataset_types.append(DatasetType.PQ25)
tiles = list_tiles_as_generator(x=[self.x], y=[self.y], satellites=self.satellites,
acq_min=acq_min, acq_max=acq_max,
dataset_types=dataset_types, include=criteria)
stack = get_dataset_data_stack(tiles, self.dataset_type, self.band.name, ndv=ndv,
x=self.x_offset, y=self.y_offset,
x_size=self.x_chunk_size, y_size=self.y_chunk_size,
mask_pqa_apply=self.mask_pqa_apply, mask_pqa_mask=self.mask_pqa_mask)
if len(stack) == 0:
return
# TODO get statistics to be generated from command line argument
if Statistic.COUNT in self.statistics:
log_mem("Before COUNT")
# COUNT
print "COUNT"
stack_stat = calculate_stack_statistic_count(stack=stack, ndv=ndv)
numpy.save(self.get_statistic_filename(Statistic.COUNT), stack_stat)
del stack_stat
if Statistic.MIN in self.statistics:
log_mem("Before MIN")
# MIN
print "MIN"
stack_stat = calculate_stack_statistic_min(stack=stack, ndv=ndv)
numpy.save(self.get_statistic_filename(Statistic.MIN), stack_stat)
del stack_stat
if Statistic.MAX in self.statistics:
log_mem("Before MAX")
# MAX
print "MAX"
stack_stat = calculate_stack_statistic_max(stack=stack, ndv=ndv)
numpy.save(self.get_statistic_filename(Statistic.MAX), stack_stat)
del stack_stat
if Statistic.MEAN in self.statistics:
log_mem("Before MEAN")
# MEAN
print "MEAN"
stack_stat = calculate_stack_statistic_mean(stack=stack, ndv=ndv)
numpy.save(self.get_statistic_filename(Statistic.MEAN), stack_stat)
del stack_stat
for percentile in [Statistic.PERCENTILE_25, Statistic.PERCENTILE_50, Statistic.PERCENTILE_75, Statistic.PERCENTILE_90, Statistic.PERCENTILE_95]:
if percentile in self.statistics:
log_mem("Before {p}".format(p=percentile.name))
print "Before {p}".format(p=percentile.name)
stack_stat = calculate_stack_statistic_percentile(stack=stack, percentile=PERCENTILE[percentile], ndv=ndv)
numpy.save(self.get_statistic_filename(percentile), stack_stat)
del stack_stat
if Statistic.COUNT_OBSERVED in self.statistics:
log_mem("Before OBSERVED COUNT")
# COUNT OBSERVED - note the copy=False is modifying the array so this is done last
print "COUNT OBSERVED"
stack_stat = calculate_stack_statistic_count_observed(stack=stack, ndv=ndv)
numpy.save(self.get_statistic_filename(Statistic.COUNT_OBSERVED), stack_stat)
del stack_stat
log_mem("DONE")
def run(self):
_log.info("Creating stack for band [%s]", self.band.name)
data_type = get_dataset_type_datatype(self.dataset_type)
ndv = get_dataset_type_ndv(self.dataset_type)
metadata = None
driver = None
raster = None
acq_min, acq_max, criteria = build_season_date_criteria(
self.acq_min,
self.acq_max,
self.season,
seasons=SEASONS,
extend=True)
_log.info("\tacq %s to %s criteria is %s", acq_min, acq_max, criteria)
dataset_types = [self.dataset_type]
if self.mask_pqa_apply:
dataset_types.append(DatasetType.PQ25)
tiles = list_tiles_as_list(x=[self.x],
y=[self.y],
satellites=self.satellites,
acq_min=acq_min,
acq_max=acq_max,
dataset_types=dataset_types,
include=criteria)
for index, tile in enumerate(tiles, start=1):
dataset = tile.datasets[self.dataset_type]
assert dataset
# band = dataset.bands[self.band]
# assert band
band = self.band
pqa = (self.mask_pqa_apply and DatasetType.PQ25 in tile.datasets
) and tile.datasets[DatasetType.PQ25] or None
if self.dataset_type not in tile.datasets:
_log.debug("No [%s] dataset present for [%s] - skipping",
self.dataset_type.name, tile.end_datetime)
continue
filename = self.output().path
if not metadata:
metadata = get_dataset_metadata(dataset)
assert metadata
if not driver:
if self.output_format == OutputFormat.GEOTIFF:
driver = gdal.GetDriverByName("GTiff")
elif self.output_format == OutputFormat.ENVI:
driver = gdal.GetDriverByName("ENVI")
assert driver
if not raster:
if self.output_format == OutputFormat.GEOTIFF:
raster = driver.Create(
filename,
metadata.shape[0],
metadata.shape[1],
len(tiles),
data_type,
options=["BIGTIFF=YES", "INTERLEAVE=BAND"])
elif self.output_format == OutputFormat.ENVI:
raster = driver.Create(filename,
metadata.shape[0],
metadata.shape[1],
len(tiles),
data_type,
options=["INTERLEAVE=BSQ"])
assert raster
# NOTE: could do this without the metadata!!
raster.SetGeoTransform(metadata.transform)
raster.SetProjection(metadata.projection)
raster.SetMetadata(self.generate_raster_metadata())
mask = None
if pqa:
mask = get_mask_pqa(pqa, self.mask_pqa_mask, mask=mask)
_log.info(
"Stacking [%s] band data from [%s] with PQA [%s] and PQA mask [%s] to [%s]",
band.name, dataset.path, pqa and pqa.path or "",
pqa and self.mask_pqa_mask or "", filename)
data = get_dataset_data_masked(dataset, mask=mask, ndv=ndv)
_log.debug("data is [%s]", data)
stack_band = raster.GetRasterBand(index)
stack_band.SetDescription(os.path.basename(dataset.path))
stack_band.SetNoDataValue(ndv)
stack_band.WriteArray(data[band])
stack_band.ComputeStatistics(True)
stack_band.SetMetadata({
"ACQ_DATE": format_date(tile.end_datetime),
"SATELLITE": dataset.satellite.name
})
stack_band.FlushCache()
del stack_band
if raster:
raster.FlushCache()
del raster
raster = None
