load_data.py

[Kurt Feigl]

I was having issues reading the HDF5 file named "inputs/geometryGeo.h5" in Matlab. The compression method "lzf" is not available:

     H5Z_pipeline    required filter 'lzf' is not registered

Apparently, in load_data.py the compression filter was hard-wired to "lzf". To work around this issue, I changed two lines in load_data.py below and then set 

      mintpy.load.compression     = gzip

#################################################################
def load_data(inps):
    """load data into HDF5 files."""

    ## 0. read input
    start_time = time.time()
    iDict = read_inps2dict(inps)

    ## 1. prepare metadata
    prepare_metadata(iDict)
    extraDict = get_extra_metadata(iDict)

    # skip data writing as it is included in prep_aria/nisar
    if iDict['processor'] in ['aria', 'nisar']:
        return

    ## 2. search & write data files
    print('-'*50)
    print('updateMode : {}'.format(iDict['updateMode']))
    print('compression: {}'.format(iDict['compression']))
    print('multilook x/ystep: {}/{}'.format(iDict['xstep'], iDict['ystep']))
    print('multilook method : {}'.format(iDict['method']))
    kwargs = dict(updateMode=iDict['updateMode'], xstep=iDict['xstep'], ystep=iDict['ystep'])

    # read subset info [need the metadata from above]
    iDict = read_subset_box(iDict)

    # geometry in geo / radar coordinates
    geom_dset_name2template_key = {
        **GEOM_DSET_NAME2TEMPLATE_KEY,
        **IFG_DSET_NAME2TEMPLATE_KEY,
        **OFF_DSET_NAME2TEMPLATE_KEY,
    }
    geom_geo_obj, geom_radar_obj = read_inps_dict2geometry_dict_object(iDict, geom_dset_name2template_key)
    geom_geo_file = os.path.abspath('./inputs/geometryGeo.h5')
    geom_radar_file = os.path.abspath('./inputs/geometryRadar.h5')

    if run_or_skip(geom_geo_file, geom_geo_obj, iDict['box4geo'], **kwargs) == 'run':
        geom_geo_obj.write2hdf5(
            outputFile=geom_geo_file,
            access_mode='w',
            box=iDict['box4geo'],
            xstep=iDict['xstep'],
            ystep=iDict['ystep'],
            compression=iDict['compression'])  # KF use specified compression='lzf')

    if run_or_skip(geom_radar_file, geom_radar_obj, iDict['box'], **kwargs) == 'run':
        geom_radar_obj.write2hdf5(
            outputFile=geom_radar_file,
            access_mode='w',
            box=iDict['box'],
            xstep=iDict['xstep'],
            ystep=iDict['ystep'],
            extra_metadata=extraDict,
            compression=iDict['compression'])  # KF use specified compression='lzf')


    # observations: ifgram, ion or offset
    # loop over obs stacks
    stack_ds_name2tmpl_key_list = [
        IFG_DSET_NAME2TEMPLATE_KEY,
        ION_DSET_NAME2TEMPLATE_KEY,
        OFF_DSET_NAME2TEMPLATE_KEY,
    ]
    stack_files = ['ifgramStack.h5', 'ionStack.h5', 'offsetStack.h5']
    stack_files = [os.path.abspath(os.path.join('./inputs', x)) for x in stack_files]
    for ds_name2tmpl_opt, stack_file in zip(stack_ds_name2tmpl_key_list, stack_files):

        # initiate dict objects
        stack_obj = read_inps_dict2ifgram_stack_dict_object(iDict, ds_name2tmpl_opt)

        # use geom_obj as size reference while loading ionosphere
        geom_obj = None
        if os.path.basename(stack_file).startswith('ion'):
            geom_obj = geom_geo_obj if iDict['geocoded'] else geom_radar_obj

        # write dict objects to HDF5 files
        if run_or_skip(stack_file, stack_obj, iDict['box'], geom_obj=geom_obj, **kwargs) == 'run':
            stack_obj.write2hdf5(
                outputFile=stack_file,
                access_mode='w',
                box=iDict['box'],
                xstep=iDict['xstep'],
                ystep=iDict['ystep'],
                mli_method=iDict['method'],
                compression=iDict['compression'],
                extra_metadata=extraDict,
                geom_obj=geom_obj)

    # used time
    m, s = divmod(time.time()-start_time, 60)
    print(f'time used: {m:02.0f} mins {s:02.1f} secs.\n')

    return

[Zhang Yunjun]

Thank you for the suggestion. This has been implemented in https://github.com/insarlab/MintPy/pull/1338.

Cheers,

Yunjun