def test_different_dims(self):
# Source data
source_nx = 100
source_ny = 100
source_x = np.linspace(-180.0, 180.0,
source_nx).astype(np.float64)
source_y = np.linspace(-90, 90.0,
source_ny).astype(np.float64)
source_x, source_y = np.meshgrid(source_x, source_y)
data = np.arange(source_nx * source_ny,
dtype=np.int32).reshape(source_ny, source_nx)
source_cs = ccrs.Geodetic()
# Target grids (different shapes)
target_x_shape = (23, 45)
target_y_shape = (23, 44)
target_x = np.arange(reduce(operator.mul, target_x_shape),
dtype=np.float64).reshape(target_x_shape)
target_y = np.arange(reduce(operator.mul, target_y_shape),
dtype=np.float64).reshape(target_y_shape)
target_proj = ccrs.PlateCarree()
# Attempt regrid
with pytest.raises(ValueError):
im_trans.regrid(data, source_x, source_y, source_cs,
target_proj, target_x, target_y)
def test_different_dims(self):
# Source data
source_nx = 100
source_ny = 100
source_x = np.linspace(-180.0, 180.0,
source_nx).astype(np.float64)
source_y = np.linspace(-90, 90.0,
source_ny).astype(np.float64)
source_x, source_y = np.meshgrid(source_x, source_y)
data = np.arange(source_nx * source_ny,
dtype=np.int32).reshape(source_ny, source_nx)
source_cs = ccrs.Geodetic()
# Target grids (different shapes)
target_x_shape = (23, 45)
target_y_shape = (23, 44)
target_x = np.arange(reduce(operator.mul, target_x_shape),
dtype=np.float64).reshape(target_x_shape)
target_y = np.arange(reduce(operator.mul, target_y_shape),
dtype=np.float64).reshape(target_y_shape)
target_proj = ccrs.PlateCarree()
# Attempt regrid
with self.assertRaises(ValueError):
im_trans.regrid(data, source_x, source_y, source_cs,
target_proj, target_x, target_y)
def test_array_dims(self):
# Source data
source_nx = 100
source_ny = 100
source_x = np.linspace(-180.0, 180.0, source_nx).astype(np.float64)
source_y = np.linspace(-90, 90.0, source_ny).astype(np.float64)
source_x, source_y = np.meshgrid(source_x, source_y)
data = np.arange(source_nx * source_ny,
dtype=np.int32).reshape(source_ny, source_nx)
source_cs = ccrs.Geodetic()
# Target grid
target_nx = 23
target_ny = 45
target_proj = ccrs.PlateCarree()
target_x, target_y, extent = im_trans.mesh_projection(
target_proj, target_nx, target_ny)
# Perform regrid
new_array = im_trans.regrid(data, source_x, source_y, source_cs,
target_proj, target_x, target_y)
# Check dimensions of return array
self.assertEqual(new_array.shape, target_x.shape)
self.assertEqual(new_array.shape, target_y.shape)
self.assertEqual(new_array.shape, (target_ny, target_nx))
def test_gridding_data_outside_projection():
# Data which exists outside the standard projection e.g. [0, 360] rather
# than [-180, 180].
target_prj = ccrs.PlateCarree()
# create 3 data points
lats = np.array([65, 10, -45])
lons = np.array([120, 180, 240])
data = np.array([1, 2, 3])
data_trans = ccrs.Geodetic()
target_x, target_y, extent = img_trans.mesh_projection(target_prj, 8, 4)
image = img_trans.regrid(data,
lons,
lats,
data_trans,
target_prj,
target_x,
target_y,
mask_extrapolated=True)
# The expected image. n.b. on a map the data is reversed in the y axis.
expected = np.array([[3, 3, 3, 3, 3, 2, 2, 2], [3, 3, 3, 3, 1, 1, 2, 2],
[3, 3, 3, 3, 1, 1, 1, 2], [3, 3, 3, 1, 1, 1, 1, 1]],
dtype=np.float64)
expected_mask = np.array(
[[True, True, True, True, True, True, True, True],
[False, False, True, True, True, True, False, False],
[False, False, True, True, True, True, False, False],
[True, True, True, True, True, True, True, True]])
assert_array_equal([-180, 180, -90, 90], extent)
assert_array_equal(expected, image)
assert_array_equal(expected_mask, image.mask)
def test_griding_data_outside_projection():
# Data which exists outside the standard projection e.g. [0, 360] rather
# than [-180, 180].
target_prj = ccrs.PlateCarree()
# create 3 data points
lats = np.array([65, 10, -45])
lons = np.array([120, 180, 240])
data = np.array([1, 2, 3])
data_trans = ccrs.Geodetic()
target_x, target_y, extent = img_trans.mesh_projection(target_prj, 8, 4)
image = img_trans.regrid(data, lons, lats, data_trans, target_prj,
target_x, target_y,
mask_extrapolated=True)
# The expected image. n.b. on a map the data is reversed in the y axis.
expected = np.array(
[[3, 3, 3, 3, 3, 3, 3, 3],
[3, 3, 3, 3, 3, 1, 2, 2],
[2, 2, 3, 1, 1, 1, 1, 2],
[1, 1, 1, 1, 1, 1, 1, 1]], dtype=np.float64)
expected_mask = np.array(
[[True, True, True, True, True, True, True, True],
[False, False, True, True, True, True, False, False],
[False, False, True, True, True, True, False, False],
[True, True, True, True, True, True, True, True]])
assert_array_equal([-180, 180, -90, 90], extent)
assert_array_equal(expected, image)
assert_array_equal(expected_mask, image.mask)
def test_array_dims(self):
# Source data
source_nx = 100
source_ny = 100
source_x = np.linspace(-180.0,
180.0,
source_nx).astype(np.float64)
source_y = np.linspace(-90, 90.0, source_ny).astype(np.float64)
source_x, source_y = np.meshgrid(source_x, source_y)
data = np.arange(source_nx * source_ny,
dtype=np.int32).reshape(source_ny, source_nx)
source_cs = ccrs.Geodetic()
# Target grid
target_nx = 23
target_ny = 45
target_proj = ccrs.PlateCarree()
target_x, target_y, extent = im_trans.mesh_projection(target_proj,
target_nx,
target_ny)
# Perform regrid
new_array = im_trans.regrid(data, source_x, source_y, source_cs,
target_proj, target_x, target_y)
# Check dimensions of return array
self.assertEqual(new_array.shape, target_x.shape)
self.assertEqual(new_array.shape, target_y.shape)
self.assertEqual(new_array.shape, (target_ny, target_nx))
def data(self, da, projection):
""""""
# Reproject the data
if projection and (da[self.xname].ndim == 2):
projection = getattr(cartopy.crs, projection)()
# Set the central longitude (if applicable)
if self.c_lon:
try:
projection = projection.__class__(central_longitude=self.c_lon)
except:
pass
# TODO: Need to detect and add cyclic point insertion
x, y = img_transform.mesh_projection(projection, nx=self.nx, ny=self.ny)[:2]
field = img_transform.regrid(da.to_masked_array(), da[self.xname].values, da[self.yname].values,
self.ref_crs, projection, x, y)
plt.gcf().add_subplot(self.subplot, projection=projection)
# Subsample the data and plot on grid coords
else:
# NOTE: Assume here that dims are (y, x); I don't think auto-transpose happens in mpl so should be
# obvious from figure if this assumption is wrong
da = da[slice(None, None, max(1, da.shape[0] / self.ny)),
slice(None, None, max(1, da.shape[1] / self.nx))]
x = da[self.xname]
y = da[self.yname]
field = da.to_masked_array()
plt.gcf().add_subplot(self.subplot)
return x, y, field
def test_regrid_image():
# Source data
fname = os.path.join(config["repo_data_dir"], 'raster', 'natural_earth',
'50-natural-earth-1-downsampled.png')
nx = 720
ny = 360
source_proj = ccrs.PlateCarree()
source_x, source_y, _ = im_trans.mesh_projection(source_proj, nx, ny)
data = plt.imread(fname)
# Flip vertically to match source_x/source_y orientation
data = data[::-1]
# Target grid
target_nx = 300
target_ny = 300
target_proj = ccrs.InterruptedGoodeHomolosine(emphasis='land')
target_x, target_y, target_extent = im_trans.mesh_projection(
target_proj, target_nx, target_ny)
# Perform regrid
new_array = im_trans.regrid(data, source_x, source_y, source_proj,
target_proj, target_x, target_y)
# Plot
fig = plt.figure(figsize=(10, 10))
gs = mpl.gridspec.GridSpec(nrows=4, ncols=1, hspace=1.5, wspace=0.5)
# Set up axes and title
ax = fig.add_subplot(gs[0], projection=target_proj)
ax.imshow(new_array, origin='lower', extent=target_extent)
ax.coastlines()
# Plot each color slice (tests masking)
cmaps = {'red': 'Reds', 'green': 'Greens', 'blue': 'Blues'}
for i, color in enumerate(['red', 'green', 'blue']):
ax = fig.add_subplot(gs[i + 1], projection=target_proj)
ax.imshow(new_array[:, :, i],
extent=target_extent,
origin='lower',
cmap=cmaps[color])
ax.coastlines()
# Tighten up layout
gs.tight_layout(fig)
return fig
def test_regrid_image():
# Source data
fname = os.path.join(config["repo_data_dir"], 'raster', 'natural_earth',
'50-natural-earth-1-downsampled.png')
nx = 720
ny = 360
source_proj = ccrs.PlateCarree()
source_x, source_y, _ = im_trans.mesh_projection(source_proj, nx, ny)
data = plt.imread(fname)
# Flip vertically to match source_x/source_y orientation
data = data[::-1]
# Target grid
target_nx = 300
target_ny = 300
target_proj = ccrs.InterruptedGoodeHomolosine()
target_x, target_y, target_extent = im_trans.mesh_projection(target_proj,
target_nx,
target_ny)
# Perform regrid
new_array = im_trans.regrid(data, source_x, source_y, source_proj,
target_proj, target_x, target_y)
# Plot
fig = plt.figure(figsize=(10, 10))
gs = matplotlib.gridspec.GridSpec(nrows=4, ncols=1,
hspace=1.5, wspace=0.5)
# Set up axes and title
ax = plt.subplot(gs[0], frameon=False, projection=target_proj)
plt.imshow(new_array, origin='lower', extent=target_extent)
ax.coastlines()
# Plot each colour slice (tests masking)
cmaps = {'red': 'Reds', 'green': 'Greens', 'blue': 'Blues'}
for i, colour in enumerate(['red', 'green', 'blue']):
ax = plt.subplot(gs[i + 1], frameon=False, projection=target_proj)
ax.set_title(colour)
plt.imshow(new_array[:, :, i], extent=target_extent, origin='lower',
cmap=cmaps[colour])
ax.coastlines()
# Tighten up layout
gs.tight_layout(plt.gcf())
def test_griding_data():
target_prj = ccrs.PlateCarree()
# create 3 data points
lats = np.array([45, 20, -45])
lons = np.array([-90, 90, 0])
data = np.array([1, 2, 3])
data_trans = ccrs.Geodetic()
target_x, target_y, extent = img_trans.mesh_projection(target_prj, 8, 4)
image = img_trans.regrid(data, lons, lats, data_trans, target_prj,
target_x, target_y)
# The expected image. n.b. on a map the data is reversed in the y axis.
expected = np.array(
[[3., 3., 3., 3., 3., 3., 3., 3.], [1., 1., 3., 3., 3., 2., 2., 2.],
[1., 1., 1., 1., 2., 2., 2., 2.], [1., 1., 1., 1., 1., 2., 2., 1.]],
dtype=np.float64)
assert_array_equal([-180, 180, -90, 90], extent)
assert_array_equal(expected, image)
def test_griding_data():
target_prj = ccrs.PlateCarree()
# create 3 data points
lats = np.array([45, 20, -45])
lons = np.array([-90, 90, 0])
data = np.array([1, 2, 3])
data_trans = ccrs.Geodetic()
target_x, target_y, extent = img_trans.mesh_projection(target_prj, 8, 4)
image = img_trans.regrid(data, lons, lats, data_trans, target_prj,
target_x, target_y)
# The expected image. n.b. on a map the data is reversed in the y axis.
expected = np.array([[3., 3., 3., 3., 3., 3., 3., 3.],
[1., 1., 3., 3., 3., 2., 2., 2.],
[1., 1., 1., 1., 2., 2., 2., 2.],
[1., 1., 1., 1., 1., 2., 2., 1.]], dtype=np.float64)
assert_array_equal([-180, 180, -90, 90], extent)
assert_array_equal(expected, image)
