1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
|
# cython: language_level=3, boundscheck=False, cdivision=True, wraparound=False, initializedcheck=False, nonecheck=False
cimport cython
from ._modis_utils cimport floating
from ._modis_utils cimport lonlat2xyz, xyz2lonlat
from ._modis_utils import rows_per_scan_for_resolution
cimport numpy as np
import numpy as np
from scipy.ndimage import map_coordinates
np.import_array()
def interpolate_geolocation_cartesian(
np.ndarray[floating, ndim=2] lon_array,
np.ndarray[floating, ndim=2] lat_array,
unsigned int coarse_resolution,
unsigned int fine_resolution):
lon_array = np.ascontiguousarray(lon_array)
lat_array = np.ascontiguousarray(lat_array)
cdef unsigned int rows_per_scan = rows_per_scan_for_resolution(coarse_resolution)
cdef unsigned int res_factor = coarse_resolution // fine_resolution
cdef Py_ssize_t num_rows = lon_array.shape[0]
cdef Py_ssize_t num_cols = lon_array.shape[1]
cdef unsigned int num_scans = num_rows // rows_per_scan
# SciPy's map_coordinates requires the x/y dimension to be first
cdef np.ndarray[floating, ndim=3] coordinates = np.empty(
(2, res_factor * rows_per_scan, res_factor * num_cols), dtype=lon_array.dtype)
cdef floating[:, :, ::1] coordinates_view = coordinates
_compute_yx_coordinate_arrays(res_factor, coordinates_view)
cdef np.ndarray[floating, ndim=3] xyz_result = np.empty(
(res_factor * rows_per_scan, num_cols * res_factor, 3), dtype=lon_array.dtype)
cdef floating[:, :, ::1] xyz_result_view = xyz_result
cdef np.ndarray[floating, ndim=3] xyz_in = np.empty(
(rows_per_scan, num_cols, 3), dtype=lon_array.dtype)
cdef floating[:, :, ::1] xyz_in_view = xyz_in
cdef floating[:, ::1] lon_in_view = lon_array
cdef floating[:, ::1] lat_in_view = lat_array
cdef np.ndarray[floating, ndim=2] new_lons = np.empty((res_factor * num_rows, res_factor * num_cols),
dtype=lon_array.dtype)
cdef np.ndarray[floating, ndim=2] new_lats = np.empty((res_factor * num_rows, res_factor * num_cols),
dtype=lon_array.dtype)
cdef floating[:, ::1] new_lons_view = new_lons
cdef floating[:, ::1] new_lats_view = new_lats
# Interpolate each scan, one at a time, otherwise the math doesn't work well
cdef Py_ssize_t scan_idx, j0, j1, k0, k1, comp_index
with nogil:
for scan_idx in range(num_scans):
# Calculate indexes
j0 = rows_per_scan * scan_idx
j1 = j0 + rows_per_scan
k0 = rows_per_scan * res_factor * scan_idx
k1 = k0 + rows_per_scan * res_factor
lonlat2xyz(lon_in_view[j0:j1, :], lat_in_view[j0:j1, :], xyz_in_view)
_compute_interpolated_xyz_scan(
res_factor, coordinates_view, xyz_in_view,
xyz_result_view)
xyz2lonlat(xyz_result_view, new_lons_view[k0:k1], new_lats_view[k0:k1], low_lat_z=True)
return new_lons, new_lats
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef void _compute_yx_coordinate_arrays(
unsigned int res_factor,
floating[:, :, ::1] coordinates,
) noexcept nogil:
cdef Py_ssize_t i, j
for j in range(coordinates.shape[1]):
for i in range(coordinates.shape[2]):
# y coordinate - 0.375 for 250m, 0.25 for 500m
coordinates[0, j, i] = j * (1.0 / res_factor) - (res_factor * (1.0 / 16) + (1.0 / 8))
# x coordinate
coordinates[1, j, i] = i * (1.0 / res_factor)
@cython.boundscheck(False)
cdef void _compute_interpolated_xyz_scan(
unsigned int res_factor,
floating[:, :, ::1] coordinates_view,
floating[:, :, ::1] xyz_input_view,
floating[:, :, ::1] xyz_result_view,
) noexcept nogil:
cdef Py_ssize_t comp_index
cdef floating[:, :] input_view, result_view
with gil:
for comp_index in range(3):
input_view = xyz_input_view[:, :, comp_index]
result_view = xyz_result_view[:, :, comp_index]
_call_map_coordinates(
input_view,
coordinates_view,
result_view,
)
if res_factor == 4:
for comp_index in range(3):
result_view = xyz_result_view[:, :, comp_index]
_extrapolate_xyz_rightmost_columns(result_view, 3)
_interpolate_xyz_250(
result_view,
coordinates_view,
)
else:
for comp_index in range(3):
result_view = xyz_result_view[:, :, comp_index]
_extrapolate_xyz_rightmost_columns(result_view, 1)
_interpolate_xyz_500(
result_view,
coordinates_view,
)
cdef void _call_map_coordinates(
floating[:, :] nav_array_view,
floating[:, :, ::1] coordinates_view,
floating[:, :] result_view,
):
cdef np.ndarray[floating, ndim=2] nav_array = np.asarray(nav_array_view)
cdef np.ndarray[floating, ndim=3] coordinates_array = np.asarray(coordinates_view)
cdef np.ndarray[floating, ndim=2] result_array = np.asarray(result_view)
# Use bilinear interpolation for all 250 meter pixels
map_coordinates(nav_array, coordinates_array,
output=result_array,
order=1, mode='nearest')
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef void _extrapolate_xyz_rightmost_columns(
floating[:, :] result_view,
int num_columns,
) noexcept nogil:
cdef Py_ssize_t row_idx, col_offset
cdef floating last_interp_col_diff
for row_idx in range(result_view.shape[0]):
last_interp_col_diff = result_view[row_idx, result_view.shape[1] - num_columns - 1] - \
result_view[row_idx, result_view.shape[1] - num_columns - 2]
for col_offset in range(num_columns):
# map_coordinates repeated the last columns value, we now add more to it as an "extrapolation"
result_view[row_idx, result_view.shape[1] - num_columns + col_offset] += last_interp_col_diff * (col_offset + 1)
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef void _interpolate_xyz_250(
floating[:, :] result_view,
floating[:, :, ::1] coordinates_view,
) noexcept nogil:
cdef Py_ssize_t col_idx
cdef floating m, b
cdef floating[:] result_col_view
cdef floating[:, ::1] y_coordinates = coordinates_view[0]
for col_idx in range(result_view.shape[1]):
result_col_view = result_view[:, col_idx]
# Use linear extrapolation for the first two 250 meter pixels along track
m = _calc_slope_250(result_col_view,
y_coordinates,
2)
b = _calc_offset_250(result_col_view,
y_coordinates,
m,
2)
result_view[0, col_idx] = m * y_coordinates[0, 0] + b
result_view[1, col_idx] = m * y_coordinates[1, 0] + b
# Use linear extrapolation for the last two 250 meter pixels along track
# m = (result_array[k0 + 37, :] - result_array[k0 + 34, :]) / (y[37, 0] - y[34, 0])
# b = result_array[k0 + 37, :] - m * y[37, 0]
m = _calc_slope_250(result_col_view,
y_coordinates,
34)
b = _calc_offset_250(result_col_view,
y_coordinates,
m,
34)
result_view[38, col_idx] = m * y_coordinates[38, 0] + b
result_view[39, col_idx] = m * y_coordinates[39, 0] + b
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef void _interpolate_xyz_500(
floating[:, :] result_view,
floating[:, :, ::1] coordinates_view,
) noexcept nogil:
cdef Py_ssize_t col_idx
cdef floating m, b
for col_idx in range(result_view.shape[1]):
# Use linear extrapolation for the first two 250 meter pixels along track
m = _calc_slope_500(
result_view[:, col_idx],
coordinates_view[0],
1)
b = _calc_offset_500(
result_view[:, col_idx],
coordinates_view[0],
m,
1)
result_view[0, col_idx] = m * coordinates_view[0, 0, 0] + b
# Use linear extrapolation for the last two 250 meter pixels along track
m = _calc_slope_500(
result_view[:, col_idx],
coordinates_view[0],
17)
b = _calc_offset_500(
result_view[:, col_idx],
coordinates_view[0],
m,
17)
result_view[19, col_idx] = m * coordinates_view[0, 19, 0] + b
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef inline floating _calc_slope_250(
floating[:] result_view,
floating[:, ::1] y,
Py_ssize_t offset,
) noexcept nogil:
return (result_view[offset + 3] - result_view[offset]) / \
(y[offset + 3, 0] - y[offset, 0])
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef inline floating _calc_offset_250(
floating[:] result_view,
floating[:, ::1] y,
floating m,
Py_ssize_t offset,
) noexcept nogil:
return result_view[offset + 3] - m * y[offset + 3, 0]
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef inline floating _calc_slope_500(
floating[:] result_view,
floating[:, ::1] y,
Py_ssize_t offset,
) noexcept nogil:
return (result_view[offset + 1] - result_view[offset]) / \
(y[offset + 1, 0] - y[offset, 0])
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
@cython.initializedcheck(False)
cdef inline floating _calc_offset_500(
floating[:] result_view,
floating[:, ::1] y,
floating m,
Py_ssize_t offset,
) noexcept nogil:
return result_view[offset + 1] - m * y[offset + 1, 0]
|
