def test_mst_matrix_as_array(self):
# Verifies MST matrix func returns array with dict/trees in each cell
for i in self.ifgs[3:]:
i.phase_data[0, 1] = 0 # partial stack of NODATA to one cell
for i in self.ifgs:
i.convert_to_nans() # zeros to NaN/NODATA
epochs = algorithm.get_epochs(self.ifgs)[0]
res = mst._mst_matrix_as_array(self.ifgs)
ys, xs = res.shape
for y, x in product(range(ys), range(xs)):
r = res[y, x]
num_nodes = len(r)
self.assertTrue(num_nodes < len(epochs.dates))
stack = array([i.phase_data[y, x]
for i in self.ifgs]) # 17 ifg stack
self.assertTrue(
0 == nsum(stack == 0)) # all 0s should be converted
nc = nsum(isnan(stack))
exp_count = len(epochs.dates) - 1
if nc == 0:
self.assertEqual(num_nodes, exp_count)
elif nc > 5:
# rough test: too many nans must reduce the total tree size
self.assertTrue(num_nodes <= (17 - nc))
def write_timeseries_geotiff(ifgs, params, tsincr, pr_type):
# setup metadata for writing into result files
gt, md, wkt = get_geotiff_header_info(ifgs[0].data_path)
epochlist = algorithm.get_epochs(ifgs)[0]
for i in range(tsincr.shape[2]):
md[ifc.EPOCH_DATE] = epochlist.dates[i + 1]
md['SEQUENCE_POSITION'] = i + 1 # sequence position
data = tsincr[:, :, i]
dest = join(params[cf.OUT_DIR],
pr_type + "_" + str(epochlist.dates[i + 1]) + ".tif")
md[ifc.DATA_TYPE] = pr_type
write_output_geotiff(md, gt, wkt, data, dest, np.nan)
def _create_ifg_dict(dest_tifs, params, tiles):
"""
1. Convert ifg phase data into numpy binary files.
2. Save the preread_ifgs dict with information about the ifgs that are
later used for fast loading of Ifg files in IfgPart class
:param list dest_tifs: List of destination tifs
:param dict params: Config dictionary
:param list tiles: List of all Tile instances
:return: preread_ifgs: Dictionary containing information regarding
interferograms that are used later in workflow
:rtype: dict
"""
ifgs_dict = {}
process_tifs = mpiops.array_split(dest_tifs)
shared.save_numpy_phase(dest_tifs, tiles, params)
for d in process_tifs:
ifg = shared._prep_ifg(d, params)
ifgs_dict[d] = PrereadIfg(path=d,
nan_fraction=ifg.nan_fraction,
master=ifg.master,
slave=ifg.slave,
time_span=ifg.time_span,
nrows=ifg.nrows,
ncols=ifg.ncols,
metadata=ifg.meta_data)
ifg.close()
ifgs_dict = _join_dicts(mpiops.comm.allgather(ifgs_dict))
preread_ifgs_file = join(params[cf.TMPDIR], 'preread_ifgs.pk')
if mpiops.rank == MASTER_PROCESS:
# add some extra information that's also useful later
gt, md, wkt = shared.get_geotiff_header_info(process_tifs[0])
ifgs_dict['epochlist'] = algorithm.get_epochs(ifgs_dict)[0]
ifgs_dict['gt'] = gt
ifgs_dict['md'] = md
ifgs_dict['wkt'] = wkt
# dump ifgs_dict file for later use
cp.dump(ifgs_dict, open(preread_ifgs_file, 'wb'))
mpiops.comm.barrier()
preread_ifgs = OrderedDict(sorted(cp.load(open(preread_ifgs_file,
'rb')).items()))
log.info('Finished converting phase_data to numpy '
'in process {}'.format(mpiops.rank))
return preread_ifgs
def write_linrate_tifs(ifgs, params, res):
# log.info('Writing linrate results')
rate, error, samples = res
gt, md, wkt = get_geotiff_header_info(ifgs[0].data_path)
epochlist = algorithm.get_epochs(ifgs)[0]
dest = join(params[cf.OUT_DIR], "linrate.tif")
md[ifc.EPOCH_DATE] = epochlist.dates
md[ifc.DATA_TYPE] = ifc.LINRATE
write_output_geotiff(md, gt, wkt, rate, dest, np.nan)
dest = join(params[cf.OUT_DIR], "linerror.tif")
md[ifc.DATA_TYPE] = ifc.LINERROR
write_output_geotiff(md, gt, wkt, error, dest, np.nan)
dest = join(params[cf.OUT_DIR], "linsamples.tif")
md[ifc.DATA_TYPE] = ifc.LINSAMP
write_output_geotiff(md, gt, wkt, samples, dest, np.nan)
write_linrate_numpy_files(error, rate, samples, params)
def _time_series_setup(ifgs, mst, params):
"""
Convenience function for setting up time series computation parameters
"""
if len(ifgs) < 1:
msg = 'Time series requires 2+ interferograms'
raise TimeSeriesError(msg)
# if mst is not a single tree then do interpolation
interp = 0 if mst_module.mst_from_ifgs(ifgs)[1] else 1
# Parallel Processing parameters
parallel = params[cf.PARALLEL]
# Time Series parameters
tsmethod = params[cf.TIME_SERIES_METHOD]
pthresh, smfactor, smorder = _validate_params(params, tsmethod)
epochlist = get_epochs(ifgs)[0]
nrows = ifgs[0].nrows
ncols = ifgs[0].ncols
nifgs = len(ifgs)
span = diff(epochlist.spans)
nepoch = len(epochlist.dates) # epoch number
nvelpar = nepoch - 1 # velocity parameters number
# nlap = nvelpar - smorder # Laplacian observations number
mast_slave_ids = master_slave_ids(epochlist.dates)
imaster = [mast_slave_ids[ifg.master] for ifg in ifgs]
islave = [mast_slave_ids[ifg.slave] for ifg in ifgs]
imaster = min(imaster, islave)
islave = max(imaster, islave)
b0_mat = zeros((nifgs, nvelpar))
for i in range(nifgs):
b0_mat[i, imaster[i]:islave[i]] = span[imaster[i]:islave[i]]
# change the sign if slave is earlier than master
isign = where(imaster > islave)
b0_mat[isign[0], :] = -b0_mat[isign[0], :]
tsvel_matrix = np.empty(shape=(nrows, ncols, nvelpar),
dtype=float32)
ifg_data = np.zeros((nifgs, nrows, ncols), dtype=float32)
for ifg_num in range(nifgs):
ifg_data[ifg_num] = ifgs[ifg_num].phase_data
if mst is None:
mst = ~isnan(ifg_data)
return b0_mat, interp, pthresh, smfactor, smorder, tsmethod, ifg_data, \
mst, ncols, nrows, nvelpar, parallel, span, tsvel_matrix
def _ts_to_ifgs(tsincr, preread_ifgs):
"""
Function that converts an incremental displacement time series into
interferometric phase observations. Used to re-construct an interferogram
network from a time series.
:param ndarray tsincr: incremental time series array of size
(ifg.shape, nepochs-1)
:param dict preread_ifgs: Dictionary of shared.PrereadIfg class instances
:return: None, interferograms are saved to disk
"""
log.info('Converting time series to ifgs')
ifgs = list(OrderedDict(sorted(preread_ifgs.items())).values())
_, n = get_epochs(ifgs)
index_master, index_slave = n[:len(ifgs)], n[len(ifgs):]
for i, ifg in enumerate(ifgs):
phase = np.sum(tsincr[:, :, index_master[i]:index_slave[i]], axis=2)
_save_aps_corrected_phase(ifg.path, phase)
def get_nml(ifg_list_instance, nodata_value, nan_conversion=False):
"""
A reproduction of getnml.m, the function in Matlab Pirate.
Note: the Matlab version tested does not have nan's.
replaces the ifg_list_instance in place.
:param ifg_list_instance: xxxx
:param nodata_value: xxxx
:param nan_conversion: xxxx
:return xxxxx
"""
_epoch_list, n = get_epochs(ifg_list_instance.ifgs)
ifg_list_instance.reshape_n(n)
if nan_conversion:
ifg_list_instance.update_nan_frac(nodata_value)
# turn on for nan conversion
ifg_list_instance.convert_nans(nan_conversion=nan_conversion)
ifg_list_instance.make_data_stack()
return ifg_list_instance, _epoch_list
def get_nml(ifg_list_instance, nodata_value, nan_conversion=False):
"""
A translation of the Matlab Pirate 'getnml.m' function.
Note: the Matlab version tested does not have nan's.
:param xxx(eg str, tuple, int, float...) ifg_list_instance: xxxx
:param float nodata_value: No data value in image
:param bool nan_conversion: Convert NaNs
:return: ifg_list_instance: replaces in place
:rtype: list
:return: _epoch_list: list of epochs
:rtype: list
"""
_epoch_list, n = algorithm.get_epochs(ifg_list_instance.ifgs)
ifg_list_instance.reshape_n(n)
if nan_conversion:
ifg_list_instance.update_nan_frac(nodata_value)
# turn on for nan conversion
ifg_list_instance.convert_nans(nan_conversion=nan_conversion)
ifg_list_instance.make_data_stack()
return ifg_list_instance, _epoch_list
def test_get_epochs(self):
def str2date(s):
segs = s[:4], s[4:6], s[6:] # year, month, day
return date(*[int(sg) for sg in segs])
raw_date = ['20060619', '20060828', '20061002', '20061106', '20061211',
'20070115', '20070219', '20070326', '20070430', '20070604',
'20070709', '20070813', '20070917']
exp_dates = [str2date(d) for d in raw_date]
exp_repeat = [1, 1, 3, 3, 4, 3, 3, 3, 3, 3, 3, 2, 2]
exp_spans = [0, 0.1916, 0.2875, 0.3833, 0.4791, 0.5749, 0.6708, 0.7666,
0.8624, 0.9582, 1.0541, 1.1499, 1.2457]
ifms = join(SML_TEST_TIF, "ifms_17")
ifgs = [Ifg(join(SML_TEST_TIF, p)) for p in parse_namelist(ifms)]
for i in ifgs:
i.open()
epochs = get_epochs(ifgs)[0]
self.assertTrue((exp_dates == epochs.dates).all())
self.assertTrue((exp_repeat == epochs.repeat).all())
assert_array_almost_equal(exp_spans, epochs.spans, decimal=4)
def test_timeseries_linrate_mpi(mpisync, tempdir, modify_config,
ref_est_method, row_splits, col_splits,
get_crop, orbfit_lks, orbfit_method,
orbfit_degrees):
params = modify_config
outdir = mpiops.run_once(tempdir)
params[cf.OUT_DIR] = outdir
params[cf.TMPDIR] = os.path.join(params[cf.OUT_DIR], cf.TMPDIR)
params[cf.DEM_HEADER_FILE] = SML_TEST_DEM_HDR_GAMMA
params[cf.REF_EST_METHOD] = ref_est_method
params[cf.IFG_CROP_OPT] = get_crop
params[cf.ORBITAL_FIT_LOOKS_Y] = orbfit_lks
params[cf.ORBITAL_FIT_LOOKS_X] = orbfit_lks
params[cf.ORBITAL_FIT_METHOD] = orbfit_method
params[cf.ORBITAL_FIT_DEGREE] = orbfit_degrees
xlks, ylks, crop = cf.transform_params(params)
if xlks * col_splits > 45 or ylks * row_splits > 70:
print('skipping test because lks and col_splits are not compatible')
return
# skip some tests in travis to run CI faster
if TRAVIS and (xlks % 2 or row_splits % 2 or col_splits % 2
or orbfit_lks % 2):
print('Skipping in travis env for faster CI run')
return
print("xlks={}, ref_est_method={}, row_splits={}, col_splits={}, "
"get_crop={}, orbfit_lks={}, orbfit_method={}, "
"rank={}".format(xlks, ref_est_method, row_splits, col_splits,
get_crop, orbfit_lks, orbfit_method, orbfit_degrees,
mpiops.rank))
base_unw_paths = cf.original_ifg_paths(params[cf.IFG_FILE_LIST])
# dest_paths are tifs that have been geotif converted and multilooked
dest_paths = cf.get_dest_paths(base_unw_paths, crop, params, xlks)
# run prepifg, create the dest_paths files
if mpiops.rank == 0:
run_prepifg.gamma_prepifg(base_unw_paths, params)
mpiops.comm.barrier()
(refpx,
refpy), maxvar, vcmt = run_pyrate.process_ifgs(ifg_paths=dest_paths,
params=params,
rows=row_splits,
cols=col_splits)
tiles = mpiops.run_once(run_pyrate.get_tiles,
dest_paths[0],
rows=row_splits,
cols=col_splits)
postprocessing.postprocess_linrate(row_splits, col_splits, params)
postprocessing.postprocess_timeseries(row_splits, col_splits, params)
ifgs_mpi_out_dir = params[cf.OUT_DIR]
ifgs_mpi = small_data_setup(datafiles=dest_paths)
# single process timeseries/linrate calculation
if mpiops.rank == 0:
params_old = modify_config
params_old[cf.OUT_DIR] = tempdir()
params_old[cf.REF_EST_METHOD] = ref_est_method
params_old[cf.IFG_CROP_OPT] = get_crop
params_old[cf.ORBITAL_FIT_LOOKS_Y] = orbfit_lks
params_old[cf.ORBITAL_FIT_LOOKS_X] = orbfit_lks
params_old[cf.ORBITAL_FIT_METHOD] = orbfit_method
params_old[cf.ORBITAL_FIT_DEGREE] = orbfit_degrees
xlks, ylks, crop = cf.transform_params(params_old)
base_unw_paths = cf.original_ifg_paths(params_old[cf.IFG_FILE_LIST])
dest_paths = cf.get_dest_paths(base_unw_paths, crop, params_old, xlks)
run_prepifg.gamma_prepifg(base_unw_paths, params_old)
ifgs = shared.pre_prepare_ifgs(dest_paths, params_old)
mst_grid = tests.common.mst_calculation(dest_paths, params_old)
refy, refx = refpixel.ref_pixel(ifgs, params_old)
assert (refx == refpx) and (refy == refpy) # both must match
pyrate.orbital.remove_orbital_error(ifgs, params_old)
ifgs = common.prepare_ifgs_without_phase(dest_paths, params_old)
rpe.estimate_ref_phase(ifgs, params_old, refx, refy)
ifgs = shared.pre_prepare_ifgs(dest_paths, params_old)
maxvar_s = [vcm.cvd(i, params_old)[0] for i in ifgs]
vcmt_s = vcm.get_vcmt(ifgs, maxvar)
tsincr, tscum, _ = tests.common.compute_time_series(
ifgs, mst_grid, params, vcmt)
rate, error, samples = tests.common.calculate_linear_rate(
ifgs, params_old, vcmt, mst_grid)
mst_mpi = reconstruct_mst(ifgs[0].shape, tiles, params[cf.TMPDIR])
np.testing.assert_array_almost_equal(mst_grid, mst_mpi)
tsincr_mpi, tscum_mpi = reconstruct_times_series(
ifgs[0].shape, tiles, params[cf.TMPDIR])
rate_mpi, error_mpi, samples_mpi = \
[reconstruct_linrate(ifgs[0].shape, tiles, params[cf.TMPDIR], t)
for t in ['linrate', 'linerror', 'linsamples']]
np.testing.assert_array_almost_equal(maxvar, maxvar_s)
np.testing.assert_array_almost_equal(vcmt, vcmt_s)
for i, j in zip(ifgs, ifgs_mpi):
np.testing.assert_array_almost_equal(i.phase_data, j.phase_data)
np.testing.assert_array_almost_equal(tsincr, tsincr_mpi, decimal=4)
np.testing.assert_array_almost_equal(tscum, tscum_mpi, decimal=4)
np.testing.assert_array_almost_equal(rate, rate_mpi, decimal=4)
np.testing.assert_array_almost_equal(error, error_mpi, decimal=4)
np.testing.assert_array_almost_equal(samples, samples_mpi, decimal=4)
# assert linear rate output tifs are same
_tifs_same(ifgs_mpi_out_dir, params_old[cf.OUT_DIR], 'linrate.tif')
_tifs_same(ifgs_mpi_out_dir, params_old[cf.OUT_DIR], 'linerror.tif')
_tifs_same(ifgs_mpi_out_dir, params_old[cf.OUT_DIR], 'linsamples.tif')
# assert time series output tifs are same
epochlist = algorithm.get_epochs(ifgs)[0]
for i in range(tsincr.shape[2]):
_tifs_same(ifgs_mpi_out_dir, params_old[cf.OUT_DIR],
'tsincr' + '_' + str(epochlist.dates[i + 1]) + ".tif")
# 12 timeseries outputs
assert i + 1 == tsincr.shape[2]
shutil.rmtree(ifgs_mpi_out_dir) # remove mpi out dir
shutil.rmtree(params_old[cf.OUT_DIR]) # remove serial out dir
# tsincr matrix from matlab using svd timeseries method
tsincr_svd = sio.loadmat(
os.path.join(SML_TEST_DIR, 'matlab_aps', 'tsincr_svd.mat'))
# prepread ifgs pickle file
preread_pk = 'preread_ifgs.pk' if PY3 else 'preread_ifgs_py2.pk'
ifgs_pk = os.path.join(SML_TEST_DIR, 'matlab_aps', preread_pk)
ifgs_pk = pickle.load(open(ifgs_pk, 'rb'))
# tlp filter output from matlab
ts_hp = sio.loadmat(os.path.join(SML_TEST_DIR, 'matlab_aps', 'ts_hp.mat'))
ts_hp_m2 = sio.loadmat(os.path.join(SML_TEST_DIR, 'matlab_aps',
'ts_hp_m2.mat'))
ts_hp_m3 = sio.loadmat(os.path.join(SML_TEST_DIR, 'matlab_aps',
'ts_hp_m3.mat'))
epochlist = get_epochs(ifgs_pk)[0]
params = cf.get_config_params(os.path.join(TEST_CONF_GAMMA))
@pytest.fixture(params=[1, 2, 3])
def tlpfilter_method(request):
return request.param
@pytest.fixture(params=[1, 2])
def slpfilter_method(request):
return request.param
def test_tlpfilter_matlab(tlpfilter_method):
params[cf.TLPF_METHOD] = tlpfilter_method
def time_series_setup(ifgs, mst, params):
"""
Convenience function setting up time series computation parameters.
:param ifgs: xxxx
:param mst: xxxx
:param params: xxxx
:return xxxx
"""
if len(ifgs) < 1:
msg = 'Time series requires 2+ interferograms'
raise TimeSeriesError(msg)
# if mst is not a single tree then do interpolation
interp = 0 if mst_module.mst_from_ifgs(ifgs)[1] else 1
# Parallel Processing parameters
parallel = params[cf.PARALLEL]
# Time Series parameters
tsmethod = params[cf.TIME_SERIES_METHOD]
if tsmethod == 1 and params[cf.TIME_SERIES_SM_ORDER] is None:
missing_option_error(cf.TIME_SERIES_SM_ORDER)
else:
smorder = params[cf.TIME_SERIES_SM_ORDER]
if tsmethod == 1 and params[cf.TIME_SERIES_SM_FACTOR] is None:
missing_option_error(cf.TIME_SERIES_SM_FACTOR)
else:
smfactor = np.power(10, params[cf.TIME_SERIES_SM_FACTOR])
if params[cf.TIME_SERIES_PTHRESH] is None:
missing_option_error(cf.TIME_SERIES_PTHRESH)
else:
pthresh = params[cf.TIME_SERIES_PTHRESH]
if pthresh < 0.0 or pthresh > 1000:
raise ValueError(
"minimum number of coherent observations for a pixel"
" TIME_SERIES_PTHRESH setting must be >= 0.0 and <= 1000")
epochlist = get_epochs(ifgs)[0]
nrows = ifgs[0].nrows
ncols = ifgs[0].ncols
nifgs = len(ifgs)
span = diff(epochlist.spans)
nepoch = len(epochlist.dates) # epoch number
nvelpar = nepoch - 1 # velocity parameters number
# nlap = nvelpar - smorder # Laplacian observations number
mast_slave_ids = master_slave_ids(epochlist.dates)
imaster = [mast_slave_ids[ifg.master] for ifg in ifgs]
islave = [mast_slave_ids[ifg.slave] for ifg in ifgs]
imaster = min(imaster, islave)
islave = max(imaster, islave)
b0_mat = zeros((nifgs, nvelpar))
for i in range(nifgs):
b0_mat[i, imaster[i]:islave[i]] = span[imaster[i]:islave[i]]
# change the sign if slave is earlier than master
isign = where(imaster > islave)
b0_mat[isign[0], :] = -b0_mat[isign[0], :]
tsvel_matrix = np.empty(shape=(nrows, ncols, nvelpar), dtype=float32)
ifg_data = np.zeros((nifgs, nrows, ncols), dtype=float32)
for ifg_num in range(nifgs):
ifg_data[ifg_num] = ifgs[ifg_num].phase_data
if mst is None:
mst = ~isnan(ifg_data)
return b0_mat, interp, pthresh, smfactor, smorder, tsmethod, ifg_data, \
mst, ncols, nrows, nvelpar, parallel, span, tsvel_matrix
