import os
import time
import shutil
import inspect
from osgeo import gdal
from spatialist import bbox, intersect
from spatialist.ancillary import finder
from pyroSAR import identify, identify_many, Archive
from s1ard import etad, ard
from s1ard.config import get_config, gdal_conf
from s1ard.ancillary import set_logging
from s1ard import search
from s1ard import ocn
from cesard import dem
import cesard.tile_extraction as tile_ex
from cesard.search import scene_select
from cesard.ancillary import (buffer_time, check_scene_consistency,
check_spacing, get_max_ext, group_by_attr)
from s1ard.processors.registry import load_processor
gdal.UseExceptions()
[docs]
def main(config_file=None, debug=False, **kwargs):
"""
Main function that initiates and controls the processing workflow.
Parameters
----------
config_file: str or None
Full path to a `config.ini` file or `None` to use the package's default file.
debug: bool
Set logging level to DEBUG? Default is False.
**kwargs
extra arguments to override parameters in the config file. E.g. `acq_mode`.
"""
update = False # update existing products? Internal development flag.
config = get_config(config_file=config_file, **kwargs)
log = set_logging(config=config, debug=debug)
config_proc = config['processing']
processor_name = config_proc['processor']
processor = load_processor(processor_name)
config_sar = config[processor_name]
gdal_prms = gdal_conf(config=config)
spacings = {am: config_proc[f'spacing_{am.lower()}']
for am in ['IW', 'SM', 'EW']}
config_sar['spacing'] = spacings[config_proc['acq_mode']]
check_spacing(config_sar['spacing'])
sar_flag = 'sar' in config_proc['mode']
nrb_flag = 'nrb' in config_proc['mode']
orb_flag = 'orb' in config_proc['mode']
# DEM download authentication
username, password = dem.authenticate(dem_type=config_proc['dem_type'],
username=None, password=None)
####################################################################################################################
# scene selection
log.info('collecting scenes')
db_file_set = config_proc['db_file'] is not None
scene_dir_set = config_proc['scene_dir'] is not None
stac_catalog_set = config_proc['stac_catalog'] is not None
stac_collections_set = config_proc['stac_collections'] is not None
parquet_set = config_proc['parquet'] is not None
if db_file_set:
archive = Archive(dbfile=config_proc['db_file'])
if scene_dir_set:
scenes = finder(target=config_proc['scene_dir'],
matchlist=[r'^S1[ABCD].*(SAFE|zip)$'],
regex=True, recursive=True, foldermode=1)
archive.insert(scenes)
elif stac_catalog_set and stac_collections_set:
archive = search.STACArchive(url=config_proc['stac_catalog'],
collections=config_proc['stac_collections'])
elif parquet_set:
archive = search.STACParquetArchive(files=config_proc['parquet'])
else:
raise RuntimeError('could not select a search option. Please check your configuration.')
if config_proc['scene'] is None:
attr_search = ['sensor', 'product', 'mindate', 'maxdate',
'aoi_tiles', 'aoi_geometry', 'date_strict']
dict_search = {k: config_proc[k] for k in attr_search}
dict_search['acquisition_mode'] = config_proc['acq_mode']
if config_proc['datatake'] is not None:
frame_number = [int(x, 16) for x in config_proc['datatake']]
else:
frame_number = None
dict_search['frameNumber'] = frame_number
selection, aoi_tiles = scene_select(archive=archive,
**dict_search)
if len(selection) == 0:
log.error('could not find any scenes')
archive.close()
return
log.info(f'found {len(selection)} scene(s)')
scenes = identify_many(selection, sortkey='start')
else:
if config_proc['mode'] != ['sar']:
raise RuntimeError("if argument 'scene' is set, the processing mode must be 'sar'")
scenes = [identify(config_proc['scene'])]
config_proc['acq_mode'] = scenes[0].acquisition_mode
config_proc['product'] = scenes[0].product
aoi_tiles = []
search.check_acquisition_completeness(scenes=scenes, archive=archive)
# group scenes by datatake
scenes_grouped = group_by_attr(scenes, lambda x: x.meta['frameNumber'])
for scenes in scenes_grouped:
# check that the scenes can really be grouped together
check_scene_consistency(scenes=scenes)
# Remove scenes with an invalid (0) slice number if others have a valid one (>0).
# This ensures that scenes with a valid slice number are preferred.
slice_numbers = [x.meta['sliceNumber'] for x in scenes]
if None not in slice_numbers:
if min(slice_numbers) == 0 and max(slice_numbers) > 0:
for i in reversed(range(len(scenes))):
if slice_numbers[i] == 0:
del scenes[i]
del slice_numbers[i]
for i in range(1, len(scenes)):
if slice_numbers[i] != slice_numbers[i - 1] + 1:
raise RuntimeError(f"nonconsecutive scene group, "
f"slice numbers: {slice_numbers}")
####################################################################################################################
# Get neighboring GRD scenes to add a buffer to the geocoded scenes.
# Otherwise, there will be a gap between final geocoded images.
# Buffering is only possible if the product composition is 'Sliced'
# (not 'Assembled' or 'Individual') and thus has a sliceNumber attribute.
if config_proc['product'] == 'GRD' and sar_flag:
log.info('collecting GRD neighbors')
neighbors = []
for scenes in scenes_grouped:
if scenes[0].meta['sliceNumber'] is not None:
neighbors_group = []
for scene in scenes:
neighbors_scene = search.collect_neighbors(archive=archive,
scene=scene)
neighbors_group.append(neighbors_scene)
else:
neighbors_group = [None] * len(scenes)
neighbors.append(neighbors_group)
else:
neighbors = [[None] * len(x) for x in scenes_grouped]
####################################################################################################################
# OCN scene selection
if 'wm' in config_proc['annotation']:
log.info('collecting OCN products')
scenes_ocn = []
for scenes in scenes_grouped:
scenes_ocn_group = []
for scene in scenes:
start, stop = buffer_time(scene.start, scene.stop, seconds=2)
result = archive.select(product='OCN', mindate=start,
maxdate=stop, date_strict=True)
if len(result) == 1:
scenes_ocn_group.append(result[0])
else:
if len(result) == 0:
log.error(f'could not find an OCN product for scene {scene.scene}')
else:
msg = f'found multiple OCN products for scene {scene.scene}:\n'
log.error(msg + '\n'.join(result))
archive.close()
return
scenes_ocn_group = identify_many(scenes_ocn_group)
scenes_ocn.append(scenes_ocn_group)
else:
scenes_ocn = [[] for scenes in scenes_grouped]
archive.close()
####################################################################################################################
# annotation layer selection
annotation = config_proc['annotation']
measurement = config_proc['measurement']
export_extra = processor.translate_annotation(annotation=annotation,
measurement=measurement)
####################################################################################################################
# main SAR processing
if sar_flag:
for h, scenes in enumerate(scenes_grouped):
log.info(f'SAR processing of group {h + 1}/{len(scenes_grouped)}')
for i, scene in enumerate(scenes):
scene_base = os.path.splitext(os.path.basename(scene.scene))[0]
out_dir_scene = os.path.join(config_proc['sar_dir'], scene_base)
tmp_dir_scene = os.path.join(config_proc['tmp_dir'], scene_base)
log.info(f'processing scene {i + 1}/{len(scenes)}: {scene.scene}')
if os.path.isdir(out_dir_scene) and not update:
log.info('Already processed - Skip!')
continue
else:
os.makedirs(out_dir_scene, exist_ok=True)
os.makedirs(tmp_dir_scene, exist_ok=True)
########################################################################################################
# Preparation of DEM for SAR processing
dem_prepare_mode = config_sar['dem_prepare_mode']
if dem_prepare_mode is not None:
fname_dem = dem.prepare(scene=scene, dem_type=config_proc['dem_type'],
dir_out=tmp_dir_scene, username=username,
password=password, mode=dem_prepare_mode,
tr=(config_sar['spacing'], config_sar['spacing']))
else:
fname_dem = None
########################################################################################################
# ETAD correction
if config_proc['etad']:
log.info('ETAD correction')
scene = etad.process(scene=scene, etad_dir=config_proc['etad_dir'],
out_dir=tmp_dir_scene)
########################################################################################################
# determination of look factors
if scene.product == 'SLC':
rlks = {'IW': 5,
'SM': 6,
'EW': 3}[config_proc['acq_mode']]
rlks *= int(config_sar['spacing'] / 10)
azlks = {'IW': 1,
'SM': 6,
'EW': 1}[config_proc['acq_mode']]
azlks *= int(config_sar['spacing'] / 10)
else:
rlks = azlks = None
########################################################################################################
# main processing routine
start_time = time.time()
try:
log.info('starting SAR processing')
proc_args = {'scene': scene.scene,
'outdir': config_proc['sar_dir'],
'measurement': measurement,
'tmpdir': config_proc['tmp_dir'],
'dem': fname_dem,
'neighbors': neighbors[h][i],
'export_extra': export_extra,
'rlks': rlks, 'azlks': azlks}
proc_args.update(config_sar)
sig = inspect.signature(processor.process)
accepted_params = set(sig.parameters.keys())
proc_args = {k: v for k, v in proc_args.items() if k in accepted_params}
processor.process(**proc_args)
t = round((time.time() - start_time), 2)
log.info(f'SAR processing finished in {t} seconds')
except Exception as e:
log.error(msg=e)
raise
####################################################################################################################
# OCN preparation
if sum(len(x) for x in scenes_ocn) > 0:
log.info('extracting OCN products')
for scenes in scenes_ocn:
for scene in scenes:
if scene.compression is not None:
scene.unpack(directory=config_proc['tmp_dir'], exist_ok=True)
basename = os.path.basename(scene.scene).replace('.SAFE', '')
outdir = os.path.join(config_proc['sar_dir'], basename)
os.makedirs(outdir, exist_ok=True)
for v in ['owiNrcsCmod', 'owiEcmwfWindSpeed', 'owiEcmwfWindDirection']:
out = os.path.join(outdir, f'{v}.tif')
if not os.path.isfile(out):
ocn.extract(src=scene.scene, dst=out, variable=v)
####################################################################################################################
# ARD - final product generation
if nrb_flag or orb_flag:
product_type = 'NRB' if nrb_flag else 'ORB'
log.info(f'starting {product_type} production')
for s, scenes in enumerate(scenes_grouped):
log.info(f'ARD processing of group {s + 1}/{len(scenes_grouped)}')
log.info('preparing WBM tiles')
vec = [x.geometry() for x in scenes]
extent = get_max_ext(geometries=vec)
with bbox(coordinates=extent, crs=4326) as box:
dem.retile(vector=box, threads=gdal_prms['threads'],
dem_dir=None, wbm_dir=config_proc['wbm_dir'],
dem_type=config_proc['dem_type'],
tilenames=aoi_tiles, username=username, password=password,
dem_strict=True)
# get the geometries of all tiles that overlap with the current scene group
tiles = tile_ex.tile_from_aoi(vector=vec,
return_geometries=True,
tilenames=aoi_tiles)
del vec
t_total = len(tiles)
for t, tile in enumerate(tiles):
# select all scenes from the group whose footprint overlaps with the current tile
scenes_sub = [x for x in scenes if intersect(tile, x.geometry())]
scenes_sub_fnames = [x.scene for x in scenes_sub]
fname_wbm = os.path.join(config_proc['wbm_dir'],
config_proc['dem_type'],
'{}_WBM.tif'.format(tile.mgrs))
if not os.path.isfile(fname_wbm):
fname_wbm = None
add_dem = True # add the DEM as output layer?
dem_type = config_proc['dem_type'] if add_dem else None
extent = tile.extent
epsg = tile.getProjection('epsg')
log.info(f'creating product {t + 1}/{t_total}')
log.info(f'selected scene(s): {scenes_sub_fnames}')
prod_meta = ard.product_info(
product_type=product_type, src_ids=scenes_sub,
tile_id=tile.mgrs, extent=extent, epsg=epsg,
dir_ard=config_proc['ard_dir'], update=update
)
if prod_meta is None:
continue
log.info(f"product name: {prod_meta['dir_ard_product']}")
try:
src_ids, sar_assets = ard.get_datasets(
scenes=scenes_sub_fnames,
sar_dir=config_proc['sar_dir'],
extent=extent, epsg=epsg,
processor_name=processor_name
)
if len(src_ids) > 0:
ard_assets = ard.format(
config=config, prod_meta=prod_meta,
src_ids=src_ids, sar_assets=sar_assets,
tile=tile.mgrs, extent=extent, epsg=epsg,
wbm=fname_wbm, dem_type=dem_type,
compress='LERC_ZSTD',
multithread=gdal_prms['multithread'],
annotation=annotation
)
ard.append_metadata(
config=config, prod_meta=prod_meta,
src_ids=src_ids, assets=ard_assets,
compression='LERC_ZSTD'
)
else:
shutil.rmtree(prod_meta['dir_ard_product'])
except Exception as e:
log.error(msg=e)
raise
del tiles
gdal.SetConfigOption('GDAL_NUM_THREADS', gdal_prms['threads_before'])