def precompute(self, mask=None, radius_of_influence=50000, epsilon=0,
reduce_data=True, nprocs=1,
cache_dir=False, **kwargs):
"""Create bilinear coefficients and store them for later use.
Note: The `mask` keyword should be provided if geolocation may be valid
where data points are invalid. This defaults to the `mask` attribute of
the `data` numpy masked array passed to the `resample` method.
"""
del kwargs
if self.resampler is None:
kwargs = dict(source_geo_def=self.source_geo_def,
target_geo_def=self.target_geo_def,
radius_of_influence=radius_of_influence,
neighbours=32,
epsilon=epsilon,
reduce_data=reduce_data)
self.resampler = XArrayResamplerBilinear(**kwargs)
try:
self.load_bil_info(cache_dir, **kwargs)
LOG.debug("Loaded bilinear parameters")
except IOError:
LOG.debug("Computing bilinear parameters")
self.resampler.get_bil_info()
self.save_bil_info(cache_dir, **kwargs)
def test_query_resample_kdtree(self, qnd):
"""Test that query_no_distance is called in _query_resample_kdtree()."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
res, none = resampler._query_resample_kdtree(1, 2, 3, 4, reduce_data=5)
qnd.assert_called_with(2, 3, 4, 1, resampler.neighbours,
resampler.epsilon,
resampler.radius_of_influence)
def test_create_resample_kdtree(self, KDTree):
"""Test that KDTree creation is called."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
vii, kdtree = resampler._create_resample_kdtree()
self.assertEqual(np.sum(vii), 2700)
self.assertEqual(vii.size, self.source_def.size)
KDTree.assert_called_once()
def test_add_missing_coordinates(self):
"""Test coordinate updating."""
import dask.array as da
from xarray import DataArray
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
bands = ['R', 'G', 'B']
data = DataArray(da.ones((3, 10, 10)),
dims=('bands', 'y', 'x'),
coords={
'bands': bands,
'y': np.arange(10),
'x': np.arange(10)
})
resampler._add_missing_coordinates(data)
# X and Y coordinates should not change
self.assertIsNone(resampler.out_coords_x)
self.assertIsNone(resampler.out_coords_y)
self.assertIsNone(resampler._out_coords['x'])
self.assertIsNone(resampler._out_coords['y'])
self.assertTrue('bands' in resampler._out_coords)
self.assertTrue(np.all(resampler._out_coords['bands'] == bands))
# Available coordinates from self.out_coords_x and self.out_coords_y
# should be set to self._out_coords
resampler.out_coords_x = [1]
resampler.out_coords_y = [2]
resampler._add_missing_coordinates(data)
self.assertEqual(resampler._out_coords['x'], resampler.out_coords_x)
self.assertEqual(resampler._out_coords['y'], resampler.out_coords_y)
def test_create_empty_bil_info(self):
"""Test creation of empty bilinear info."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
resampler._create_empty_bil_info()
self.assertEqual(resampler.bilinear_t.shape, (self.target_def.size, ))
self.assertEqual(resampler.bilinear_s.shape, (self.target_def.size, ))
self.assertEqual(resampler._index_array.shape,
(self.target_def.size, 4))
self.assertTrue(resampler._index_array.dtype == np.int32)
self.assertEqual(resampler._valid_input_index.shape,
(self.source_def.size, ))
self.assertTrue(resampler._valid_input_index.dtype == np.bool)
def resample_dataset(data,
target_grid,
radius_of_influence=100e3,
resample_cache=None,
return_neighbor_info=False,
neighbours=1,
epsilon=0,
interp='nearest'):
# first get the source grid definition
try:
if 'area' in data.attrs:
source_grid = data.attrs['area']
else:
raise RuntimeError
except RuntimeError:
print('Must include pyresample.gemoetry in the data.attrs area_def or '
'area')
return
# check for SwathDefinition or AreaDefinition
# if swath ensure it is xarray.DataArray and not numpy for chunking
source_grid = _check_swath_or_area(source_grid)
# set kwargs for XArrayResamplerNN
kwargs = dict(source_geo_def=source_grid,
target_geo_def=target_grid,
radius_of_influence=radius_of_influence,
neighbours=neighbours,
epsilon=epsilon)
if interp is 'nearest':
resampler = XArrayResamplerNN(**kwargs)
else:
resampler = XArrayResamplerBilinear(**kwargs)
# check if resample cash is none else assume it is a dict with keys
#[valid_input_index, valid_output_index, index_array, distance_array]
# else generate the data
if resample_cache is None:
valid_input_index, valid_output_index, index_array, distance_array = resampler.get_neighbour_info(
)
else:
resampler.valid_input_index = resample_cache['valid_input_index']
resampler.valid_output_index = resample_cache['valid_output_index']
resampler.index_array = resample_cache['index_array']
resampler.distance_array = resample_cache['distance_array']
# now store the resampled data temporarily in temp
temp = resampler.get_sample_from_neighbour_info(data)
# reformat data from temp
out = _reformat_resampled_data(data, temp, target_grid)
if return_neighbor_info:
resample_cache = dict(valid_input_index=valid_input_index,
valid_output_index=valid_output_index,
index_array=index_array,
distance_array=distance_array)
return out, resample_cache
else:
return out
def test_init(self):
"""Test that the resampler has been initialized correctly."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
# With defaults
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
self.assertTrue(resampler.source_geo_def == self.source_def)
self.assertTrue(resampler.target_geo_def == self.target_def)
self.assertEqual(resampler.radius_of_influence, self.radius)
self.assertEqual(resampler.neighbours, 32)
self.assertEqual(resampler.epsilon, 0)
self.assertTrue(resampler.reduce_data)
# These should be None
self.assertIsNone(resampler.valid_input_index)
self.assertIsNone(resampler.valid_output_index)
self.assertIsNone(resampler.index_array)
self.assertIsNone(resampler.distance_array)
self.assertIsNone(resampler.bilinear_t)
self.assertIsNone(resampler.bilinear_s)
self.assertIsNone(resampler.slices_x)
self.assertIsNone(resampler.slices_y)
self.assertIsNone(resampler.mask_slices)
self.assertIsNone(resampler.out_coords_x)
self.assertIsNone(resampler.out_coords_y)
# self.slices_{x,y} are used in self.slices dict
self.assertTrue(resampler.slices['x'] is resampler.slices_x)
self.assertTrue(resampler.slices['y'] is resampler.slices_y)
# self.out_coords_{x,y} are used in self.out_coords dict
self.assertTrue(resampler.out_coords['x'] is resampler.out_coords_x)
self.assertTrue(resampler.out_coords['y'] is resampler.out_coords_y)
# Override defaults
resampler = XArrayResamplerBilinear(self.source_def,
self.target_def,
self.radius,
neighbours=16,
epsilon=0.1,
reduce_data=False)
self.assertEqual(resampler.neighbours, 16)
self.assertEqual(resampler.epsilon, 0.1)
self.assertFalse(resampler.reduce_data)
def test_get_sample_from_bil_info(self):
"""Test bilinear interpolation as a whole."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
_ = resampler.get_bil_info()
# Sample from data1
res = resampler.get_sample_from_bil_info(self.data1)
res = res.compute()
# Check couple of values
self.assertEqual(res.values[1, 1], 1.)
self.assertTrue(np.isnan(res.values[0, 3]))
# Check that the values haven't gone down or up a lot
self.assertAlmostEqual(np.nanmin(res.values), 1.)
self.assertAlmostEqual(np.nanmax(res.values), 1.)
# Check that dimensions are the same
self.assertEqual(res.dims, self.data1.dims)
# Sample from data1, custom fill value
res = resampler.get_sample_from_bil_info(self.data1, fill_value=-1.0)
res = res.compute()
self.assertEqual(np.nanmin(res.values), -1.)
# Sample from integer data
res = resampler.get_sample_from_bil_info(self.data1.astype(np.uint8),
fill_value=None)
res = res.compute()
# Five values should be filled with zeros, which is the
# default fill_value for integer data
self.assertEqual(np.sum(res == 0), 6)
def test_get_sample_from_bil_info(self):
"""Test bilinear interpolation as a whole."""
import dask.array as da
from xarray import DataArray
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
resampler.get_bil_info()
# Sample from data1
res = resampler.get_sample_from_bil_info(self.data1)
res = res.compute()
# Check couple of values
self.assertEqual(res.values[1, 1], 1.)
self.assertTrue(np.isnan(res.values[0, 3]))
# Check that the values haven't gone down or up a lot
self.assertAlmostEqual(np.nanmin(res.values), 1.)
self.assertAlmostEqual(np.nanmax(res.values), 1.)
# Check that dimensions are the same
self.assertEqual(res.dims, self.data1.dims)
# Sample from data1, custom fill value
res = resampler.get_sample_from_bil_info(self.data1, fill_value=-1.0)
res = res.compute()
self.assertEqual(np.nanmin(res.values), -1.)
# Sample from integer data
res = resampler.get_sample_from_bil_info(self.data1.astype(np.uint8),
fill_value=None)
res = res.compute()
# Five values should be filled with zeros, which is the
# default fill_value for integer data
self.assertEqual(np.sum(res == 0), 6)
# Output coordinates should have been set
self.assertTrue(isinstance(resampler._out_coords, dict))
self.assertTrue(
np.all(resampler._out_coords['x'] == resampler.out_coords_x))
self.assertTrue(
np.all(resampler._out_coords['y'] == resampler.out_coords_y))
# 3D data
data = da.moveaxis(da.dstack((self.data1, self.data1)), -1, 0)
data = DataArray(data, dims=('bands', 'y', 'x'))
res = resampler.get_sample_from_bil_info(data)
assert res.shape == (2, ) + self.target_def.shape
assert res.dims == data.dims
def test_get_index_array(self, qnd, ria):
"""Test that query_no_distance is called in __get_index_array()."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
qnd.return_value = 'foo'
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
resampler._target_lons = 1
resampler._target_lats = 2
resampler._resample_kdtree = 3
resampler._get_index_array()
qnd.assert_called_with(1, 2, True, 3, resampler._neighbours,
resampler._epsilon,
resampler._radius_of_influence)
ria.assert_called_with(qnd.return_value)
def test_compute_indices(self, mock_setattr):
"""Test running .compute() for indices."""
from pyresample.bilinear.xarr import (XArrayResamplerBilinear,
CACHE_INDICES)
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
# Set indices to Numpy arrays
for idx in CACHE_INDICES:
setattr(resampler, idx, np.array([]))
resampler._compute_indices()
# None of the indices shouldn't have been reassigned
mock_setattr.assert_not_called()
# Set indices to a Mock object
arr = mock.MagicMock()
for idx in CACHE_INDICES:
setattr(resampler, idx, arr)
resampler._compute_indices()
# All the indices should have been reassigned
self.assertEqual(mock_setattr.call_count, len(CACHE_INDICES))
# The compute should have been called the same amount of times
self.assertEqual(arr.compute.call_count, len(CACHE_INDICES))
def test_get_bil_info(self):
"""Test calculation of bilinear info."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
def _check_ts(t__, s__, nans):
for i, _ in enumerate(t__):
# Just check the exact value for one pixel
if i == 5:
self.assertAlmostEqual(t__[i], 0.730659147133, 5)
self.assertAlmostEqual(s__[i], 0.310314173004, 5)
# These pixels are outside the area
elif i in nans:
self.assertTrue(np.isnan(t__[i]))
self.assertTrue(np.isnan(s__[i]))
# All the others should have values between 0.0 and 1.0
else:
self.assertTrue(t__[i] >= 0.0)
self.assertTrue(s__[i] >= 0.0)
self.assertTrue(t__[i] <= 1.0)
self.assertTrue(s__[i] <= 1.0)
# Data reduction enabled (default)
resampler = XArrayResamplerBilinear(self.source_def,
self.target_def,
self.radius,
reduce_data=True)
(t__, s__, slices, mask_slices, out_coords) = resampler.get_bil_info()
_check_ts(t__.compute(), s__.compute(), [3, 10, 12, 13, 14, 15])
# Nothing should be masked based on coordinates
self.assertTrue(np.all(~mask_slices))
# Four values per output location
self.assertEqual(mask_slices.shape, (self.target_def.size, 4))
# self.slices_{x,y} are used in self.slices dict so they
# should be the same (object)
self.assertTrue(isinstance(slices, dict))
self.assertTrue(resampler.slices['x'] is resampler.slices_x)
self.assertTrue(np.all(resampler.slices['x'] == slices['x']))
self.assertTrue(resampler.slices['y'] is resampler.slices_y)
self.assertTrue(np.all(resampler.slices['y'] == slices['y']))
# self.slices_{x,y} are used in self.slices dict so they
# should be the same (object)
self.assertTrue(isinstance(out_coords, dict))
self.assertTrue(resampler.out_coords['x'] is resampler.out_coords_x)
self.assertTrue(np.all(resampler.out_coords['x'] == out_coords['x']))
self.assertTrue(resampler.out_coords['y'] is resampler.out_coords_y)
self.assertTrue(np.all(resampler.out_coords['y'] == out_coords['y']))
# Also some other attributes should have been set
self.assertTrue(t__ is resampler.bilinear_t)
self.assertTrue(s__ is resampler.bilinear_s)
self.assertIsNotNone(resampler.valid_output_index)
self.assertIsNotNone(resampler.index_array)
self.assertIsNotNone(resampler.valid_input_index)
# Data reduction disabled
resampler = XArrayResamplerBilinear(self.source_def,
self.target_def,
self.radius,
reduce_data=False)
(t__, s__, slices, mask_slices, out_coords) = resampler.get_bil_info()
_check_ts(t__.compute(), s__.compute(), [10, 12, 13, 14, 15])
class BilinearResampler(BaseResampler):
"""Resample using bilinear."""
def __init__(self, source_geo_def, target_geo_def):
super(BilinearResampler, self).__init__(source_geo_def, target_geo_def)
self.resampler = None
def precompute(self, mask=None, radius_of_influence=50000, epsilon=0,
reduce_data=True, nprocs=1,
cache_dir=False, **kwargs):
"""Create bilinear coefficients and store them for later use.
Note: The `mask` keyword should be provided if geolocation may be valid
where data points are invalid. This defaults to the `mask` attribute of
the `data` numpy masked array passed to the `resample` method.
"""
del kwargs
if self.resampler is None:
kwargs = dict(source_geo_def=self.source_geo_def,
target_geo_def=self.target_geo_def,
radius_of_influence=radius_of_influence,
neighbours=32,
epsilon=epsilon,
reduce_data=reduce_data)
self.resampler = XArrayResamplerBilinear(**kwargs)
try:
self.load_bil_info(cache_dir, **kwargs)
LOG.debug("Loaded bilinear parameters")
except IOError:
LOG.debug("Computing bilinear parameters")
self.resampler.get_bil_info()
self.save_bil_info(cache_dir, **kwargs)
def load_bil_info(self, cache_dir, **kwargs):
if cache_dir:
filename = self._create_cache_filename(cache_dir,
prefix='resample_lut_bil_',
**kwargs)
cache = np.load(filename)
for elt in ['bilinear_s', 'bilinear_t', 'valid_input_index',
'index_array']:
if isinstance(cache[elt], tuple):
setattr(self.resampler, elt, cache[elt][0])
else:
setattr(self.resampler, elt, cache[elt])
cache.close()
else:
raise IOError
def save_bil_info(self, cache_dir, **kwargs):
if cache_dir:
filename = self._create_cache_filename(cache_dir,
prefix='resample_lut_bil_',
**kwargs)
LOG.info('Saving kd_tree neighbour info to %s', filename)
cache = {'bilinear_s': self.resampler.bilinear_s,
'bilinear_t': self.resampler.bilinear_t,
'valid_input_index': self.resampler.valid_input_index,
'index_array': self.resampler.index_array}
np.savez(filename, **cache)
def compute(self, data, fill_value=None, **kwargs):
"""Resample the given data using bilinear interpolation"""
del kwargs
if fill_value is None:
fill_value = data.attrs.get('_FillValue')
target_shape = self.target_geo_def.shape
res = self.resampler.get_sample_from_bil_info(data,
fill_value=fill_value,
output_shape=target_shape)
return res
class BilinearResampler(BaseResampler):
"""Resample using bilinear."""
def __init__(self, source_geo_def, target_geo_def):
"""Init BilinearResampler."""
super(BilinearResampler, self).__init__(source_geo_def, target_geo_def)
self.resampler = None
def precompute(self, mask=None, radius_of_influence=50000, epsilon=0,
reduce_data=True, cache_dir=False, **kwargs):
"""Create bilinear coefficients and store them for later use.
Note: The `mask` keyword should be provided if geolocation may be valid
where data points are invalid. This defaults to the `mask` attribute of
the `data` numpy masked array passed to the `resample` method.
"""
del kwargs
if self.resampler is None:
kwargs = dict(source_geo_def=self.source_geo_def,
target_geo_def=self.target_geo_def,
radius_of_influence=radius_of_influence,
neighbours=32,
epsilon=epsilon,
reduce_data=reduce_data)
self.resampler = XArrayResamplerBilinear(**kwargs)
try:
self.load_bil_info(cache_dir, **kwargs)
LOG.debug("Loaded bilinear parameters")
except IOError:
LOG.debug("Computing bilinear parameters")
self.resampler.get_bil_info()
self.save_bil_info(cache_dir, **kwargs)
def load_bil_info(self, cache_dir, **kwargs):
"""Load bilinear resampling info from cache directory."""
if cache_dir:
filename = self._create_cache_filename(cache_dir,
prefix='bil_lut-',
**kwargs)
for val in BIL_COORDINATES.keys():
try:
cache = da.from_zarr(filename, val)
if val == 'valid_input_index':
# valid input index array needs to be boolean
cache = cache.astype(np.bool)
# Compute the cache arrays
cache = cache.compute()
except ValueError:
raise IOError
setattr(self.resampler, val, cache)
else:
raise IOError
def save_bil_info(self, cache_dir, **kwargs):
"""Save bilinear resampling info to cache directory."""
if cache_dir:
filename = self._create_cache_filename(cache_dir,
prefix='bil_lut-',
**kwargs)
LOG.info('Saving BIL neighbour info to %s', filename)
zarr_out = xr.Dataset()
for idx_name, coord in BIL_COORDINATES.items():
var = getattr(self.resampler, idx_name)
if isinstance(var, np.ndarray):
var = da.from_array(var, chunks=CHUNK_SIZE)
var = var.rechunk(CHUNK_SIZE)
zarr_out[idx_name] = (coord, var)
zarr_out.to_zarr(filename)
def compute(self, data, fill_value=None, **kwargs):
"""Resample the given data using bilinear interpolation."""
del kwargs
if fill_value is None:
fill_value = data.attrs.get('_FillValue')
target_shape = self.target_geo_def.shape
res = self.resampler.get_sample_from_bil_info(data,
fill_value=fill_value,
output_shape=target_shape)
return update_resampled_coords(data, res, self.target_geo_def)
def test_get_bil_info(self):
"""Test calculation of bilinear info."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
def _check_ts(t__, s__, nans):
for i, _ in enumerate(t__):
# Just check the exact value for one pixel
if i == 5:
self.assertAlmostEqual(t__[i], 0.730659147133, 5)
self.assertAlmostEqual(s__[i], 0.310314173004, 5)
# These pixels are outside the area
elif i in nans:
self.assertTrue(np.isnan(t__[i]))
self.assertTrue(np.isnan(s__[i]))
# All the others should have values between 0.0 and 1.0
else:
self.assertTrue(t__[i] >= 0.0)
self.assertTrue(s__[i] >= 0.0)
self.assertTrue(t__[i] <= 1.0)
self.assertTrue(s__[i] <= 1.0)
# Data reduction enabled (default)
resampler = XArrayResamplerBilinear(self.source_def,
self.target_def,
self.radius,
reduce_data=True)
resampler.get_bil_info()
t__ = resampler.bilinear_t
s__ = resampler.bilinear_s
mask_slices = resampler.mask_slices
_check_ts(t__, s__, [3, 10, 12, 13, 14, 15])
# Nothing should be masked based on coordinates
self.assertTrue(np.all(~mask_slices))
# Four values per output location
self.assertEqual(mask_slices.shape, (self.target_def.size, 4))
# Also some other attributes should have been set
self.assertTrue(t__ is resampler.bilinear_t)
self.assertTrue(s__ is resampler.bilinear_s)
self.assertIsNotNone(resampler._index_array)
self.assertIsNotNone(resampler._valid_input_index)
self.assertIsNotNone(resampler.out_coords_x)
self.assertIsNotNone(resampler.out_coords_y)
self.assertTrue(
np.allclose(resampler.out_coords_x,
[-1070912.72, -470912.72, 129087.28, 729087.28]))
self.assertTrue(
np.allclose(resampler.out_coords_y,
[1190031.36, 590031.36, -9968.64, -609968.64]))
# Data reduction disabled
resampler = XArrayResamplerBilinear(self.source_def,
self.target_def,
self.radius,
reduce_data=False)
resampler.get_bil_info()
t__ = resampler.bilinear_t
s__ = resampler.bilinear_s
mask_slices = resampler.mask_slices
_check_ts(t__, s__, [10, 12, 13, 14, 15])
# Target area and source data do not overlap
resampler = XArrayResamplerBilinear(self.source_def,
self.target_def_outside,
self.radius,
reduce_data=False)
resampler.get_bil_info()
self.assertEqual(resampler.bilinear_t.shape,
(self.target_def_outside.size, ))
self.assertEqual(resampler.bilinear_s.shape,
(self.target_def_outside.size, ))
self.assertEqual(resampler.slices_x.shape,
(self.target_def_outside.size, 4))
self.assertEqual(resampler.slices_y.shape,
(self.target_def_outside.size, 4))
self.assertEqual(resampler.out_coords_x.shape,
(self.target_def_outside.shape[1], ))
self.assertEqual(resampler.out_coords_y.shape,
(self.target_def_outside.shape[0], ))
self.assertEqual(resampler.mask_slices.shape,
(self.target_def_outside.size, 4))
def test_get_slices(self, meshgrid):
"""Test slice array creation."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
meshgrid.return_value = (self.cols, self.lines)
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
resampler.valid_input_index = self.valid_input_index
resampler.index_array = self.index_array
resampler._get_slices()
self.assertIsNotNone(resampler.out_coords_x)
self.assertIsNotNone(resampler.out_coords_y)
self.assertTrue(resampler.out_coords_x is resampler.out_coords['x'])
self.assertTrue(resampler.out_coords_y is resampler.out_coords['y'])
self.assertTrue(
np.allclose(resampler.out_coords_x,
[-1070912.72, -470912.72, 129087.28, 729087.28]))
self.assertTrue(
np.allclose(resampler.out_coords_y,
[1190031.36, 590031.36, -9968.64, -609968.64]))
self.assertIsNotNone(resampler.slices_x)
self.assertIsNotNone(resampler.slices_y)
self.assertTrue(resampler.slices_x is resampler.slices['x'])
self.assertTrue(resampler.slices_y is resampler.slices['y'])
self.assertTrue(resampler.slices_x.shape == (self.target_def.size, 32))
self.assertTrue(resampler.slices_y.shape == (self.target_def.size, 32))
self.assertEqual(np.sum(resampler.slices_x), 12471)
self.assertEqual(np.sum(resampler.slices_y), 2223)
self.assertFalse(np.any(resampler.mask_slices))
# Ensure that source geo def is used in masking
# Setting target_geo_def to 0-size shouldn't cause any masked values
resampler.target_geo_def = np.array([])
resampler._get_slices()
self.assertFalse(np.any(resampler.mask_slices))
# Setting source area def to 0-size should mask all values
resampler.source_geo_def = np.array([[]])
resampler._get_slices()
self.assertTrue(np.all(resampler.mask_slices))
def test_slice_data(self):
"""Test slicing the data."""
import dask.array as da
from xarray import DataArray
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
# Too many dimensions
data = DataArray(da.ones((1, 3, 10, 10)))
with self.assertRaises(ValueError):
_ = resampler._slice_data(data, np.nan)
# 2D data
data = DataArray(da.ones((10, 10)))
resampler.slices_x = np.random.randint(0, 10, (100, 4))
resampler.slices_y = np.random.randint(0, 10, (100, 4))
resampler.mask_slices = np.zeros((100, 4), dtype=np.bool)
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertEqual(p_1.shape, (100, ))
self.assertTrue(p_1.shape == p_2.shape == p_3.shape == p_4.shape)
self.assertTrue(
np.all(p_1 == 1.0) and np.all(p_2 == 1.0) and np.all(p_3 == 1.0)
and np.all(p_4 == 1.0))
# 2D data with masking
resampler.mask_slices = np.ones((100, 4), dtype=np.bool)
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertTrue(
np.all(np.isnan(p_1)) and np.all(np.isnan(p_2))
and np.all(np.isnan(p_3)) and np.all(np.isnan(p_4)))
# 3D data
data = DataArray(da.ones((3, 10, 10)))
resampler.slices_x = np.random.randint(0, 10, (100, 4))
resampler.slices_y = np.random.randint(0, 10, (100, 4))
resampler.mask_slices = np.zeros((100, 4), dtype=np.bool)
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertEqual(p_1.shape, (3, 100))
self.assertTrue(p_1.shape == p_2.shape == p_3.shape == p_4.shape)
# 3D data with masking
resampler.mask_slices = np.ones((100, 4), dtype=np.bool)
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertTrue(
np.all(np.isnan(p_1)) and np.all(np.isnan(p_2))
and np.all(np.isnan(p_3)) and np.all(np.isnan(p_4)))
def test_slice_data(self):
"""Test slicing the data."""
import dask.array as da
from xarray import DataArray
from pyresample.bilinear.xarr import XArrayResamplerBilinear
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
resampler.get_bil_info()
# Too many dimensions
data = DataArray(da.ones((1, 3) + self.source_def.shape))
with self.assertRaises(ValueError):
_ = resampler._slice_data(data, np.nan)
# 2D data
data = DataArray(da.ones(self.source_def.shape))
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertEqual(p_1.shape, resampler.bilinear_s.shape)
self.assertTrue(p_1.shape == p_2.shape == p_3.shape == p_4.shape)
self.assertTrue(
np.all(p_1 == 1.0) and np.all(p_2 == 1.0) and np.all(p_3 == 1.0)
and np.all(p_4 == 1.0))
# 2D data with masking
resampler.mask_slices[:, :] = True
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertTrue(
np.all(np.isnan(p_1)) and np.all(np.isnan(p_2))
and np.all(np.isnan(p_3)) and np.all(np.isnan(p_4)))
# 3D data
data = DataArray(da.ones((3, ) + self.source_def.shape))
resampler.mask_slices[:, :] = False
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertEqual(p_1.shape, (3, ) + resampler.bilinear_s.shape)
self.assertTrue(p_1.shape == p_2.shape == p_3.shape == p_4.shape)
# 3D data with masking
resampler.mask_slices[:, :] = True
p_1, p_2, p_3, p_4 = resampler._slice_data(data, np.nan)
self.assertTrue(
np.all(np.isnan(p_1)) and np.all(np.isnan(p_2))
and np.all(np.isnan(p_3)) and np.all(np.isnan(p_4)))
def test_get_slices(self, meshgrid):
"""Test slice array creation."""
from pyresample.bilinear.xarr import XArrayResamplerBilinear
meshgrid.return_value = (self.cols, self.lines)
resampler = XArrayResamplerBilinear(self.source_def, self.target_def,
self.radius)
resampler._valid_input_index = self._valid_input_index
resampler._index_array = self._index_array
resampler._get_slices()
self.assertIsNotNone(resampler.slices_x)
self.assertIsNotNone(resampler.slices_y)
self.assertTrue(resampler.slices_x.shape == (self.target_def.size, 32))
self.assertTrue(resampler.slices_y.shape == (self.target_def.size, 32))
self.assertEqual(np.sum(resampler.slices_x), 12471)
self.assertEqual(np.sum(resampler.slices_y), 2223)
self.assertFalse(np.any(resampler.mask_slices))
# Ensure that source geo def is used in masking
# Setting target_geo_def to 0-size shouldn't cause any masked values
resampler._target_geo_def = np.array([])
resampler._get_slices()
self.assertFalse(np.any(resampler.mask_slices))
# Setting source area def to 0-size should mask all values
resampler._source_geo_def = np.array([[]])
resampler._get_slices()
self.assertTrue(np.all(resampler.mask_slices))
