def run(self):
self.parse_arguments()
config = Config(os.path.expanduser("~/.datacube/config"))
_log.debug(config.to_str())
# Clear stack files
# TODO - filename consistency and safety and so on
if self.stack_vrt:
for satellite, dataset_type in itertools.product(self.satellites, self.dataset_types):
path = os.path.join(self.output_directory, get_filename_file_list(satellite, dataset_type, self.x, self.y))
check_overwrite_remove_or_fail(path, self.overwrite)
# TODO once WOFS is in the cube
for tile in list_tiles(x=[self.x], y=[self.y], acq_min=self.acq_min, acq_max=self.acq_max,
satellites=[satellite for satellite in self.satellites],
dataset_types=intersection(self.dataset_types, dataset_type_database),
database=config.get_db_database(),
user=config.get_db_username(),
password=config.get_db_password(),
host=config.get_db_host(), port=config.get_db_port()):
if self.list_only:
_log.info("Would retrieve datasets [%s]", [tile.datasets[t].path for t in intersection(self.dataset_types, dataset_type_database)])
continue
pqa = None
# Apply PQA if specified
if self.apply_pqa_filter:
pqa = tile.datasets[DatasetType.PQ25]
for dataset_type in intersection(self.dataset_types, dataset_type_database):
retrieve_data(tile.datasets[dataset_type], pqa, self.pqa_mask, self.get_output_filename(tile.datasets[dataset_type]), tile.x, tile.y, self.overwrite, self.stack_vrt)
nbar = tile.datasets[DatasetType.ARG25]
self.generate_derived_nbar(intersection(self.dataset_types, dataset_type_derived_nbar), nbar, pqa, self.pqa_mask, self.overwrite)
# Generate VRT stack
if self.stack_vrt:
for satellite, dataset_type in itertools.product(self.satellites, self.dataset_types):
path = os.path.join(self.output_directory, get_filename_file_list(satellite, dataset_type, self.x, self.y))
if os.path.exists(path):
for band in BANDS[dataset_type, satellite]:
path_vrt = os.path.join(self.output_directory, get_filename_stack_vrt(satellite, dataset_type, self.x, self.y, band))
_log.info("Generating VRT file [%s] for band [%s]", path_vrt, band)
# gdalbuildrt -separate -b <band> -input_file_list <input file> <vrt file>
subprocess.call(["gdalbuildvrt", "-separate", "-b", str(band.value), "-input_file_list", path, path_vrt])
def go(self):
# If we are applying a vector mask then calculate it (once as it is the same for all tiles)
mask = None
if self.mask_vector_apply:
mask = get_mask_vector_for_cell(self.x, self.y,
self.mask_vector_file, self.mask_vector_layer, self.mask_vector_feature)
for tile in self.get_tiles():
if self.list_only:
_log.info("Would retrieve datasets [%s]", "\n".join([tile.datasets[t].path for t in
intersection(self.dataset_types,
[d for d in tile.datasets])]))
continue
pqa = (self.mask_pqa_apply and DatasetType.PQ25 in tile.datasets) and tile.datasets[DatasetType.PQ25] or None
wofs = (self.mask_wofs_apply and DatasetType.WATER in tile.datasets) and tile.datasets[DatasetType.WATER] or None
for dataset_type in self.dataset_types:
if dataset_type not in tile.datasets:
_log.debug("No [%s] dataset present for [%s] - skipping", dataset_type.name, tile.end_datetime)
continue
dataset = tile.datasets[dataset_type]
filename = os.path.join(self.output_directory,
get_dataset_filename(dataset,
output_format=self.output_format,
mask_pqa_apply=self.mask_pqa_apply,
mask_wofs_apply=self.mask_wofs_apply,
mask_vector_apply=self.mask_vector_apply))
retrieve_data(tile.x, tile.y, tile.end_datetime, dataset, pqa, self.mask_pqa_mask,
wofs, self.mask_wofs_mask, filename, self.output_format, self.overwrite, mask=mask)
def go(self):
import numpy
from datacube.api.query import list_cells_as_list, list_tiles_as_list
from datacube.config import Config
x_min, x_max, y_max, y_min = self.extract_bounds_from_vector()
_log.debug("The bounds are [%s]", (x_min, x_max, y_min, y_max))
cells_vector = self.extract_cells_from_vector()
_log.debug("Intersecting cells_vector are [%d] [%s]", len(cells_vector), cells_vector)
config = Config()
_log.debug(config.to_str())
x_list = range(x_min, x_max + 1)
y_list = range(y_min, y_max + 1)
_log.debug("x = [%s] y=[%s]", x_list, y_list)
cells_db = list()
for cell in list_cells_as_list(x=x_list, y=y_list, acq_min=self.acq_min, acq_max=self.acq_max,
satellites=[satellite for satellite in self.satellites],
dataset_types=[self.dataset_type]):
cells_db.append((cell.x, cell.y))
_log.debug("Cells from DB are [%d] [%s]", len(cells_db), cells_db)
cells = intersection(cells_vector, cells_db)
_log.debug("Combined cells are [%d] [%s]", len(cells), cells)
for (x, y) in cells:
_log.info("Processing cell [%3d/%4d]", x, y)
tiles = list_tiles_as_list(x=x_list, y=y_list, acq_min=self.acq_min, acq_max=self.acq_max,
satellites=[satellite for satellite in self.satellites],
dataset_types=[self.dataset_type])
_log.info("There are [%d] tiles", len(tiles))
if self.list_only:
for tile in tiles:
_log.info("Would process [%s]", tile.datasets[self.dataset_type].path)
continue
# Calculate the mask for the cell
mask_aoi = self.get_mask_aoi_cell(x, y)
pixel_count = 4000 * 4000
pixel_count_aoi = (mask_aoi == False).sum()
_log.debug("mask_aoi is [%s]\n[%s]", numpy.shape(mask_aoi), mask_aoi)
metadata = None
with self.get_output_file() as csv_file:
csv_writer = csv.writer(csv_file)
import operator
header = reduce(operator.add, [["DATE", "INSTRUMENT", "# PIXELS", "# PIXELS IN AOI"]] + [
["%s - # DATA PIXELS" % band_name,
"%s - # DATA PIXELS AFTER PQA" % band_name,
"%s - # DATA PIXELS AFTER PQA WOFS" % band_name,
"%s - # DATA PIXELS AFTER PQA WOFS AOI" % band_name,
"%s - MIN" % band_name, "%s - MAX" % band_name, "%s - MEAN" % band_name] for band_name in self.bands])
csv_writer.writerow(header)
for tile in tiles:
_log.info("Processing tile [%s]", tile.datasets[self.dataset_type].path)
if self.list_only:
continue
if not metadata:
metadata = get_dataset_metadata(tile.datasets[self.dataset_type])
# Apply PQA if specified
pqa = None
mask_pqa = None
if self.mask_pqa_apply and DatasetType.PQ25 in tile.datasets:
pqa = tile.datasets[DatasetType.PQ25]
mask_pqa = get_mask_pqa(pqa, self.mask_pqa_mask)
_log.debug("mask_pqa is [%s]\n[%s]", numpy.shape(mask_pqa), mask_pqa)
# Apply WOFS if specified
wofs = None
mask_wofs = None
if self.mask_wofs_apply and DatasetType.WATER in tile.datasets:
wofs = tile.datasets[DatasetType.WATER]
mask_wofs = get_mask_wofs(wofs, self.mask_wofs_mask)
_log.debug("mask_wofs is [%s]\n[%s]", numpy.shape(mask_wofs), mask_wofs)
dataset = tile.datasets[self.dataset_type]
bands = []
dataset_band_names = [b.name for b in dataset.bands]
for b in self.bands:
if b in dataset_band_names:
bands.append(dataset.bands[b])
data = get_dataset_data(tile.datasets[self.dataset_type], bands=bands)
_log.debug("data is [%s]\n[%s]", numpy.shape(data), data)
pixel_count_data = dict()
pixel_count_data_pqa = dict()
pixel_count_data_pqa_wofs = dict()
pixel_count_data_pqa_wofs_aoi = dict()
mmin = dict()
mmax = dict()
mmean = dict()
for band_name in self.bands:
# Add "zeroed" entries for non-present bands - should only be if outputs for those bands have been explicitly requested
if band_name not in dataset_band_names:
pixel_count_data[band_name] = 0
pixel_count_data_pqa[band_name] = 0
pixel_count_data_pqa_wofs[band_name] = 0
pixel_count_data_pqa_wofs_aoi[band_name] = 0
mmin[band_name] = numpy.ma.masked
mmax[band_name] = numpy.ma.masked
mmean[band_name] = numpy.ma.masked
continue
band = dataset.bands[band_name]
data[band] = numpy.ma.masked_equal(data[band], NDV)
_log.debug("masked data is [%s] [%d]\n[%s]", numpy.shape(data), numpy.ma.count(data), data)
pixel_count_data[band_name] = numpy.ma.count(data[band])
if pqa:
data[band].mask = numpy.ma.mask_or(data[band].mask, mask_pqa)
_log.debug("PQA masked data is [%s] [%d]\n[%s]", numpy.shape(data[band]), numpy.ma.count(data[band]), data[band])
pixel_count_data_pqa[band_name] = numpy.ma.count(data[band])
if wofs:
data[band].mask = numpy.ma.mask_or(data[band].mask, mask_wofs)
_log.debug("WOFS masked data is [%s] [%d]\n[%s]", numpy.shape(data[band]), numpy.ma.count(data[band]), data[band])
pixel_count_data_pqa_wofs[band_name] = numpy.ma.count(data[band])
data[band].mask = numpy.ma.mask_or(data[band].mask, mask_aoi)
_log.debug("AOI masked data is [%s] [%d]\n[%s]", numpy.shape(data[band]), numpy.ma.count(data[band]), data[band])
pixel_count_data_pqa_wofs_aoi[band_name] = numpy.ma.count(data[band])
mmin[band_name] = numpy.ma.min(data[band])
mmax[band_name] = numpy.ma.max(data[band])
mmean[band_name] = numpy.ma.mean(data[band])
# Convert the mean to an int...taking into account masking....
if not numpy.ma.is_masked(mmean[band_name]):
mmean[band_name] = mmean[band_name].astype(numpy.int16)
pixel_count_data_pqa_wofs_aoi_all_bands = reduce(operator.add, pixel_count_data_pqa_wofs_aoi.itervalues())
if pixel_count_data_pqa_wofs_aoi_all_bands == 0 and not self.output_no_data:
_log.info("Skipping dataset with no non-masked data values in ANY band")
continue
row = reduce(
operator.add,
[[tile.end_datetime,
self.decode_satellite_as_instrument(tile.datasets[self.dataset_type].satellite),
pixel_count, pixel_count_aoi]] +
[[pixel_count_data[band_name], pixel_count_data_pqa[band_name],
pixel_count_data_pqa_wofs[band_name], pixel_count_data_pqa_wofs_aoi[band_name],
mmin[band_name], mmax[band_name], mmean[band_name]] for band_name in self.bands])
csv_writer.writerow(row)
def go(self):
import numpy
from datacube.api.query import list_cells_as_list, list_tiles_as_list
from datacube.config import Config
# Verify that all the requested satellites have the same band combinations
dataset_bands = get_bands(self.dataset_type, self.satellites[0])
_log.info("dataset bands is [%s]", " ".join([b.name for b in dataset_bands]))
for satellite in self.satellites:
if dataset_bands != get_bands(self.dataset_type, satellite):
_log.error("Satellites [%s] have differing bands", " ".join([satellite.name for satellite in self.satellites]))
raise Exception("Satellites with different band combinations selected")
bands = []
dataset_bands_list = list(dataset_bands)
if not self.bands:
bands = dataset_bands_list
else:
for b in self.bands:
bands.append(dataset_bands_list[b - 1])
_log.info("Using bands [%s]", " ".join(band.name for band in bands))
x_min, x_max, y_max, y_min = self.extract_bounds_from_vector()
_log.debug("The bounds are [%s]", (x_min, x_max, y_min, y_max))
cells_vector = self.extract_cells_from_vector()
_log.debug("Intersecting cells_vector are [%d] [%s]", len(cells_vector), cells_vector)
config = Config(os.path.expanduser("~/.datacube/config"))
_log.debug(config.to_str())
x_list = range(x_min, x_max + 1)
y_list = range(y_min, y_max + 1)
_log.debug("x = [%s] y=[%s]", x_list, y_list)
cells_db = list()
for cell in list_cells_as_list(x=x_list, y=y_list, acq_min=self.acq_min, acq_max=self.acq_max,
satellites=[satellite for satellite in self.satellites],
dataset_types=[self.dataset_type]):
cells_db.append((cell.x, cell.y))
_log.debug("Cells from DB are [%d] [%s]", len(cells_db), cells_db)
cells = intersection(cells_vector, cells_db)
_log.debug("Combined cells are [%d] [%s]", len(cells), cells)
for (x, y) in cells:
_log.info("Processing cell [%3d/%4d]", x, y)
tiles = list_tiles_as_list(x=x_list, y=y_list, acq_min=self.acq_min, acq_max=self.acq_max,
satellites=[satellite for satellite in self.satellites],
dataset_types=[self.dataset_type])
_log.info("There are [%d] tiles", len(tiles))
if self.list_only:
for tile in tiles:
_log.info("Would process [%s]", tile.datasets[self.dataset_type].path)
continue
# Calculate the mask for the cell
mask_aoi = self.get_mask_aoi_cell(x, y)
pixel_count = 4000 * 4000
pixel_count_aoi = (mask_aoi == False).sum()
_log.debug("mask_aoi is [%s]\n[%s]", numpy.shape(mask_aoi), mask_aoi)
metadata = None
with self.get_output_file() as csv_file:
csv_writer = csv.writer(csv_file)
import operator
header = reduce(operator.add, [["DATE", "INSTRUMENT", "# PIXELS", "# PIXELS IN AOI"]] + [
["%s - # DATA PIXELS" % b.name,
"%s - # DATA PIXELS AFTER PQA" % b.name,
"%s - # DATA PIXELS AFTER PQA WOFS" % b.name,
"%s - # DATA PIXELS AFTER PQA WOFS AOI" % b.name,
"%s - MIN" % b.name, "%s - MAX" % b.name, "%s - MEAN" % b.name] for b in bands])
csv_writer.writerow(header)
for tile in tiles:
_log.info("Processing tile [%s]", tile.datasets[self.dataset_type].path)
if self.list_only:
continue
if not metadata:
metadata = get_dataset_metadata(tile.datasets[self.dataset_type])
# Apply PQA if specified
pqa = None
mask_pqa = None
if self.mask_pqa_apply and DatasetType.PQ25 in tile.datasets:
pqa = tile.datasets[DatasetType.PQ25]
mask_pqa = get_mask_pqa(pqa, self.mask_pqa_mask)
_log.debug("mask_pqa is [%s]\n[%s]", numpy.shape(mask_pqa), mask_pqa)
# Apply WOFS if specified
wofs = None
mask_wofs = None
if self.mask_wofs_apply and DatasetType.WATER in tile.datasets:
wofs = tile.datasets[DatasetType.WATER]
mask_wofs = get_mask_wofs(wofs, self.mask_wofs_mask)
_log.debug("mask_wofs is [%s]\n[%s]", numpy.shape(mask_wofs), mask_wofs)
data = get_dataset_data(tile.datasets[self.dataset_type], bands=bands)
_log.debug("data is [%s]\n[%s]", numpy.shape(data), data)
pixel_count_data = dict()
pixel_count_data_pqa = dict()
pixel_count_data_pqa_wofs = dict()
pixel_count_data_pqa_wofs_aoi = dict()
mmin = dict()
mmax = dict()
mmean = dict()
for band in bands:
data[band] = numpy.ma.masked_equal(data[band], NDV)
_log.debug("masked data is [%s] [%d]\n[%s]", numpy.shape(data), numpy.ma.count(data), data)
pixel_count_data[band] = numpy.ma.count(data[band])
if pqa:
data[band].mask = numpy.ma.mask_or(data[band].mask, mask_pqa)
_log.debug("PQA masked data is [%s] [%d]\n[%s]", numpy.shape(data[band]), numpy.ma.count(data[band]), data[band])
pixel_count_data_pqa[band] = numpy.ma.count(data[band])
if wofs:
data[band].mask = numpy.ma.mask_or(data[band].mask, mask_wofs)
_log.debug("WOFS masked data is [%s] [%d]\n[%s]", numpy.shape(data[band]), numpy.ma.count(data[band]), data[band])
pixel_count_data_pqa_wofs[band] = numpy.ma.count(data[band])
data[band].mask = numpy.ma.mask_or(data[band].mask, mask_aoi)
_log.debug("AOI masked data is [%s] [%d]\n[%s]", numpy.shape(data[band]), numpy.ma.count(data[band]), data[band])
pixel_count_data_pqa_wofs_aoi[band] = numpy.ma.count(data[band])
mmin[band] = numpy.ma.min(data[band])
mmax[band] = numpy.ma.max(data[band])
mmean[band] = numpy.ma.mean(data[band])
# Convert the mean to an int...which is actually trickier than you would expect due to masking....
if numpy.ma.count(mmean[band]) != 0:
mmean[band] = mmean[band].astype(numpy.int16)
# Should we output if no data values found?
pixel_count_data_pqa_wofs_aoi_all_bands = reduce(operator.add, pixel_count_data_pqa_wofs_aoi.itervalues())
if pixel_count_data_pqa_wofs_aoi_all_bands == 0 and not self.output_no_data:
_log.info("Skipping dataset with no non-masked data values in ANY band")
continue
row = reduce(
operator.add,
[[tile.end_datetime,
self.decode_satellite_as_instrument(tile.datasets[self.dataset_type].satellite),
pixel_count, pixel_count_aoi]] +
[[pixel_count_data[band], pixel_count_data_pqa[band],
pixel_count_data_pqa_wofs[band], pixel_count_data_pqa_wofs_aoi[band],
mmin[band], mmax[band], mmean[band]] for band in bands])
csv_writer.writerow(row)
