def test_collocate_collapse_expand(self):
"""Test whether collocating, collapsing and expanding work"""
collocator = Collocator()
test = xr.Dataset({
"time": ("time", np.arange("2000", "2010", dtype="M8[Y]")),
"lat": ("time", np.arange(10)),
"lon": ("time", np.arange(10)),
})
collocations = collocator.collocate(test,
test,
max_interval="30 days",
max_distance="150 miles")
collapsed = collapse(collocations)
expanded = expand(collocations)
def test_flat_to_main_coord(self):
"""Tests Collocator._flat_to_main_coord
This method is crucial since it stacks the whole input datasets for the
collocating routine and makes them collocateable.
"""
collocator = Collocator()
test = xr.Dataset({
"time": ("time", np.arange(10)),
"lat": ("time", np.arange(10)),
"lon": ("time", np.arange(10)),
})
check = xr.Dataset({
"time": ("collocation", np.arange(10)),
"lat": ("collocation", np.arange(10)),
"lon": ("collocation", np.arange(10)),
})
results = collocator._flat_to_main_coord(test)
assert check.equals(results)
test = xr.Dataset({
"time": ("main", np.arange(10)),
"lat": ("main", np.arange(10)),
"lon": ("main", np.arange(10)),
})
check = xr.Dataset({
"time": ("collocation", np.arange(10)),
"lat": ("collocation", np.arange(10)),
"lon": ("collocation", np.arange(10)),
})
results = collocator._flat_to_main_coord(test)
assert check.equals(results)
test = xr.Dataset({
"time": ("scnline", np.arange(5)),
"lat": (("scnline", "scnpos"), np.arange(10).reshape(5, 2)),
"lon": (("scnline", "scnpos"), np.arange(10).reshape(5, 2)),
})
check = test.stack(collocation=("scnline", "scnpos"))
results = collocator._flat_to_main_coord(test)
assert check.equals(results)
def __init__(self,
file=None,
collocator=None,
processes=10,
verbose=0,
sea_mask_file=None,
elevation_file=None):
"""Initialize a SPAREICE object
Args:
file: A JSON file with the coefficients of SPAREICE. If not given,
the standard configuration will be loaded.
collocator: SPARE-ICE requires a collocator when it should be
generated from filesets. You can pass your own
:class:`Collocator` object here if you want.
processes: Number of processes to parallelize the training or
collocation search. 10 is the default. Best value depends on
your machine.
verbose (int): Control ``GridSearchCV`` verbosity. The higher the
value, the more debug messages are printed.
"""
self.verbose = verbose
self.processes = processes
self.name = "SPARE-ICE"
if sea_mask_file is None:
self.sea_mask = None
else:
self.sea_mask = np.flip(
np.array(imageio.imread(sea_mask_file) == 255), axis=0)
if elevation_file is None:
self.elevation_grid = None
else:
ds = xr.open_dataset(elevation_file, decode_times=False)
self.elevation_grid = ds.data.squeeze().values
if collocator is None:
self.collocator = Collocator()
else:
self.collocator = collocator
# SPARE-ICE consists of two retrievals: one neural network for the IWP
# and one decision tree classifier for the ice cloud flag
self._iwp = None
self._ice_cloud = None
# The users can load SPARE-ICE from their own training or the standard
# parameters:
if file is None:
try:
self.load(STANDARD_FILE)
except Exception as e:
warnings.warn(
"Could not load the standard parameters of SPARE-ICE!\n"
"You need to train SPARE-ICE by yourself.")
warnings.warn(str(e))
self._iwp = RetrievalProduct()
self._ice_cloud = RetrievalProduct()
else:
self.load(file)