-
Notifications
You must be signed in to change notification settings - Fork 6
/
ds_gap_filling_algorithms.py
512 lines (420 loc) · 22.7 KB
/
ds_gap_filling_algorithms.py
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
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
import os
import numpy as np
import matplotlib.pyplot as plt
import h5py
from netCDF4 import Dataset
import calendar
import glob
import itertools
import rasterio
import gdal
import osr
import pandas as pd
import datetime
from scipy.optimize import curve_fit
# Ignore runtime warning
import warnings
warnings.filterwarnings("ignore", category=UserWarning)
warnings.filterwarnings("ignore", category=RuntimeWarning)
##############################################################################################################
# (Function 1) Generate lat/lon tables of Geographic projection in world/CONUS
def geo_coord_gen(lat_geo_extent_max, lat_geo_extent_min, lon_geo_extent_max, lon_geo_extent_min, cellsize):
lat_geo_output = np.linspace(lat_geo_extent_max - cellsize / 2, lat_geo_extent_min + cellsize / 2,
num=int((lat_geo_extent_max - lat_geo_extent_min) / cellsize))
lat_geo_output = np.round(lat_geo_output, decimals=3)
lon_geo_output = np.linspace(lon_geo_extent_min + cellsize / 2, lon_geo_extent_max - cellsize / 2,
num=int((lon_geo_extent_max - lon_geo_extent_min) / cellsize))
lon_geo_output = np.round(lon_geo_output, decimals=3)
return lat_geo_output, lon_geo_output
########################################################################################################################
# (Function 2) Define a function to output coefficient and intercept of linear regression fit
def multi_lrm(var_x, var_y):
nan_ind = np.where((np.isnan(var_x)) | (np.isnan(var_y)))
var_x[nan_ind[0], nan_ind[1]] = np.nan
var_y[nan_ind[0], nan_ind[1]] = np.nan
slope = np.nansum((var_x - np.nanmean(var_x, axis=0)) * (var_y - np.nanmean(var_y, axis=0)), axis=0) / \
np.nansum((var_x - np.nanmean(var_x, axis=0)) ** 2, axis=0)
intercept = np.nanmean(var_y, axis=0) - np.nanmean(var_x, axis=0) * slope
var_y_pred = slope * var_x + intercept
r2 = 1 - np.nansum((var_y - var_y_pred) ** 2, axis=0) / np.nansum((var_y - np.nanmean(var_y, axis=0)) ** 2, axis=0)
rmse = np.sqrt(np.nanmean((var_y - var_y_pred) ** 2, axis=0))
return slope, intercept, r2, rmse
########################################################################################################################
# (Function 3) Subset the coordinates table of desired area
def coordtable_subset(lat_input, lon_input, lat_extent_max, lat_extent_min, lon_extent_max, lon_extent_min):
lat_output = lat_input[np.where((lat_input <= lat_extent_max) & (lat_input >= lat_extent_min))]
row_output_ind = np.squeeze(np.array(np.where((lat_input <= lat_extent_max) & (lat_input >= lat_extent_min))))
lon_output = lon_input[np.where((lon_input <= lon_extent_max) & (lon_input >= lon_extent_min))]
col_output_ind = np.squeeze(np.array(np.where((lon_input <= lon_extent_max) & (lon_input >= lon_extent_min))))
return lat_output, row_output_ind, lon_output, col_output_ind
#########################################################################################################################
# # (Function 4) Define a function for temporal Fourier analysis
omega = 2*np.pi/365
def tclm_model(input_x, input_y):
ind_nonnan = np.where(~np.isnan(input_x) & ~np.isnan(input_y))[0]
input_x_valid = input_x[ind_nonnan]
input_y_valid = input_y[ind_nonnan]
y_mean = np.nanmean(input_y_valid)
def func(input_x_valid, a1, a2, a3, phi1, phi2, phi3):
return y_mean + a1 * np.cos(np.radians(omega * input_x_valid - phi1)) + \
a2 * np.cos(np.radians(2*omega * input_x_valid - phi2)) + \
a3 * np.cos(np.radians(3*omega * input_x_valid - phi3))
param, covariance = curve_fit(func, input_x_valid, input_y_valid, maxfev=10**9, ftol=1e-4, xtol=1e-4)
y_pred = y_mean + param[0] * np.cos(np.radians(omega * input_x - param[3])) + \
param[1] * np.cos(np.radians(2*omega * input_x - param[4])) + \
param[2] * np.cos(np.radians(3*omega * input_x - param[5]))
residuals = input_y - y_pred
ss_res = np.nansum(residuals**2)
ss_tot = np.nansum((input_y - np.nanmean(input_y))**2)
rmse = np.sqrt(ss_tot/365)
r_squared = 1 - (ss_res / ss_tot)
return y_pred, r_squared, rmse
########################################################################################################################
# 0. Input variables
# Specify file paths
# Path of current workspace
path_workspace = '/Users/binfang/Documents/SMAP_Project/smap_codes'
# Path of source GLDAS data
path_gldas = '/Volumes/MyPassport/SMAP_Project/Datasets/GLDAS'
# Path of source LTDR NDVI data
path_ltdr = '/Volumes/MyPassport/SMAP_Project/Datasets/LTDR/Ver5'
# Path of processed data
path_procdata = '/Volumes/MyPassport/SMAP_Project/Datasets/processed_data'
# Path of model data
path_model = '/Volumes/MyPassport/SMAP_Project/Datasets/model_data'
# Path of Land mask
path_lmask = '/Volumes/MyPassport/SMAP_Project/Datasets/Lmask'
# Path of AMSR2 data
path_amsr2 = '/Volumes/MyPassport/SMAP_Project/Datasets/AMSR2'
# Path of 1 km MODIS LST
path_modis_1km = '/Volumes/My Book/MODIS/Model_Input/MYD11A1/'
path_modis_input = '/Volumes/MyPassport/SMAP_Project/Datasets/MODIS/HDF_Data'
path_modis_input_hdf = '/Users/binfang/Downloads/Processing/SMAP_Downscale/HDF_Data'
path_modis_input_hdf_proc = '/Users/binfang/Downloads/Processing/SMAP_Downscale/HDF_Data_processed'
path_modis_input_hdf_tfa = '/Users/binfang/Downloads/Processing/SMAP_Downscale/HDF_Data_TFA'
path_modis_input_tif = '/Volumes/MyPassport/SMAP_Project/Datasets/MODIS/Model_Input/MYD11A1'
path_modis_lrm_output = '/Volumes/MyPassport/SMAP_Project/Datasets/model_data/LRM_output'
path_modis_lrm_output_proc = '/Volumes/MyPassport/SMAP_Project/Datasets/model_data/LRM_output'
path_modis_prediction = '/Users/binfang/Downloads/Processing/SMAP_Downscale/MODIS_LST_pred'
path_modis_lst = '/Volumes/WD My Book /MODIS/Model_Input/MYD11A1'
path_modis_lst_rev = '/Volumes/MyPassport/SMAP_Project/Datasets/MODIS/Model_Input/MYD11A1_ver2'
path_amsr2_lst_rev = '/Volumes/MyPassport/SMAP_Project/Datasets/MODIS/Model_Input/MYD11A1_AMSR2'
# Path of SMAP SM
path_smap = '/Volumes/MyPassport/SMAP_Project/Datasets/SMAP'
path_smap_ip = '/Users/binfang/Downloads/Processing/MODIS/Model_Output'
# Path of source output MODIS data
path_smap_op = '/Volumes/Elements/MODIS/Model_Output'
# Load in variables
os.chdir(path_workspace)
f = h5py.File("ds_parameters.hdf5", "r")
varname_list = ['lat_world_ease_1km', 'lon_world_ease_1km', 'lat_conus_ease_1km', 'lat_conus_ease_1km']
for x in range(len(varname_list)):
var_obj = f[varname_list[x]][()]
exec(varname_list[x] + '= var_obj')
f.close()
del (var_obj, f, varname_list)
# World extent corner coordinates
lat_world_max = 90
lat_world_min = -90
lon_world_max = 180
lon_world_min = -180
# CONUS extent corner coordinates
lat_conus_max = 53
lat_conus_min = 25
lon_conus_max = -67
lon_conus_min = -125
cellsize_10km = 0.1
cellsize_1km = 0.01
size_world_ease_1km = np.array([14616, 34704])
# Generate 1 km lat/lon tables and corresponding row/col indices in the world lat/lon table
[lat_world_geo_10km, lon_world_geo_10km] = geo_coord_gen \
(lat_world_max, lat_world_min, lon_world_max, lon_world_min, cellsize_10km)
# Generate a sequence of string between start and end dates (Year + DOY)
start_date = '2015-01-01'
end_date = '2020-12-31'
year = 2019 - 2015 + 1
start_date = datetime.datetime.strptime(start_date, '%Y-%m-%d').date()
end_date = datetime.datetime.strptime(end_date, '%Y-%m-%d').date()
delta_date = end_date - start_date
date_seq = []
date_seq_doy = []
for i in range(delta_date.days + 1):
date_str = start_date + datetime.timedelta(days=i)
date_seq.append(date_str.strftime('%Y%m%d'))
date_seq_doy.append(str(date_str.timetuple().tm_year) + str(date_str.timetuple().tm_yday).zfill(3))
# Count how many days for a specific year
yearname = np.linspace(2015, 2020, 6, dtype='int')
monthnum = np.linspace(1, 12, 12, dtype='int')
monthname = np.arange(1, 13)
monthname = [str(i).zfill(2) for i in monthname]
daysofyear = []
for idt in range(len(yearname)):
f_date = datetime.date(yearname[idt], monthnum[0], 1)
l_date = datetime.date(yearname[idt], monthnum[-1], 31)
delta_1y = l_date - f_date
daysofyear.append(delta_1y.days + 1)
# print(delta_1y.days + 1)
daysofyear = np.asarray(daysofyear)
# Find the indices of each month in the list of days
nlpyear = 1999 # non-leap year
lpyear = 2000 # leap year
daysofmonth_nlp = np.array([calendar.monthrange(nlpyear, x)[1] for x in range(1, len(monthnum) + 1)])
ind_nlp = [np.arange(daysofmonth_nlp[0:x].sum(), daysofmonth_nlp[0:x + 1].sum()) for x in range(0, len(monthnum))]
daysofmonth_lp = np.array([calendar.monthrange(lpyear, x)[1] for x in range(1, len(monthnum) + 1)])
ind_lp = [np.arange(daysofmonth_lp[0:x].sum(), daysofmonth_lp[0:x + 1].sum()) for x in range(0, len(monthnum))]
ind_iflpr = np.array([int(calendar.isleap(yearname[x])) for x in range(len(yearname))]) # Find out leap years
# Generate a sequence of the days of months for all years
daysofmonth_seq = np.array([np.tile(daysofmonth_nlp[x], len(yearname)) for x in range(0, len(monthnum))])
daysofmonth_seq[1, :] = daysofmonth_seq[1, :] + ind_iflpr # Add leap days to February
# print(datetime.datetime.now().strftime("%H:%M:%S"))
########################################################################################################################
# 0. Subset CONUS region from the world
# Load in the coord table variables
f = h5py.File(path_model + '/gap_filling/coord_world_1km_ind.hdf5', "r")
varname_list = list(f.keys())
varname_list = [varname_list[0], varname_list[3]]
for x in range(len(varname_list)):
var_obj = f[varname_list[x]][()]
exec(varname_list[x] + '= var_obj')
del (var_obj)
f.close()
del (f, varname_list)
[lat_conus_ease_1km, row_conus_1km_ind, lon_conus_ease_1km, col_conus_1km_ind] = \
coordtable_subset(lat_world_ease_1km, lon_world_ease_1km,
lat_conus_max, lat_conus_min, lon_conus_max, lon_conus_min)
col_conus_1km_ind, row_conus_1km_ind = np.meshgrid(col_conus_1km_ind, row_conus_1km_ind)
col_conus_1km_ind = col_conus_1km_ind.reshape(1, -1)
row_conus_1km_ind = row_conus_1km_ind.reshape(1, -1)
coord_conus_ease_1km_mesh_ind = \
np.ravel_multi_index(np.array([row_conus_1km_ind[0], col_conus_1km_ind[0]]),
(len(lat_world_ease_1km), len(lon_world_ease_1km)))
# Find the matched 1km land pixels in world coord table
coord_conus_1km = np.where(np.in1d(coord_world_1km_ind, coord_conus_ease_1km_mesh_ind))[0] # Subset
coord_conus_1km_ind = coord_world_1km_ind[coord_conus_1km] # Original
# Find the matched land pixels in subset CONUS coord table (indexed by world coord table indices)
coord_conus_1km_match = np.where(np.in1d(coord_conus_ease_1km_mesh_ind, coord_conus_1km_ind))[0]
# Find the matched land pixels in 10 km AMSR2 coord table
coord_conus_amsr2_ind = coord_world_1km_land_ind[coord_conus_1km] # AMSR2 coord table (original)
coord_conus_amsr2_ind_unique = np.unique(coord_conus_amsr2_ind)
coord_conus_amsr2_ind_unique_2d = \
np.unravel_index(coord_conus_amsr2_ind_unique, (len(lat_world_geo_10km), len(lon_world_geo_10km)))
coord_conus_amsr2_2d_size = \
np.array([len(np.unique(coord_conus_amsr2_ind_unique_2d[0])), len(np.unique(coord_conus_amsr2_ind_unique_2d[1]))])
coord_conus_amsr2_ind_match = coord_world_1km_land_ind_match[coord_conus_1km] # AMSR2 coord table (subset)
coord_conus_amsr2_ind_unique_match = np.unique(coord_conus_amsr2_ind_match)
########################################################################################################################
# 1. Read AMSR2 Tb and MODIS LST data
modis_files_all = sorted(glob.glob(path_modis_input + '/*hdf5*'))
modis_files_month = np.array(
[int(os.path.basename(modis_files_all[x]).split('_')[2][4:6]) for x in range(len(modis_files_all))])
modis_files_month_ind = [np.where(modis_files_month == x)[0].tolist() for x in range(1, 13)]
modis_files_year = np.array(
[int(os.path.basename(modis_files_all[x]).split('_')[2][0:4]) for x in range(len(modis_files_all))])
modis_files_year_ind = [np.where(modis_files_year == x)[0].tolist() for x in np.unique(modis_files_year)]
# len_modis = len(coord_world_1km_group_divide[0]) * 99 + len(coord_world_1km_group_divide[-1])
# coord_world_1km_group_divide_ind_rev = \
# np.concatenate(([0], coord_world_1km_group_divide_ind))
# coord_world_1km_group_divide_ind_rev[-1] = len_modis
# # Divide MODIS LST data into 10 blocks
modis_lst_divide_ind = \
np.arange(0, len(coord_conus_1km), len(coord_conus_1km) // 10)[1:]
modis_lst_group_divide = np.split(np.arange(len(coord_conus_1km)), modis_lst_divide_ind)
modis_lst_group_divide[-2] = np.concatenate((modis_lst_group_divide[-2], modis_lst_group_divide[-1]))
del (modis_lst_group_divide[-1])
var_name = ['modis_lstd_all_div', 'modis_lstn_all_div']
for iyr in range(2, len(yearname) - 1):
modis_lstd_all = []
modis_lstn_all = []
for imo in range(len(monthname)):
modis_hdf_file = h5py.File(modis_files_all[modis_files_year_ind[iyr][imo]], "r")
modis_varname_list = list(modis_hdf_file.keys())
modis_lstd = modis_hdf_file[modis_varname_list[0]][()]
modis_lstd = modis_lstd[:, coord_conus_1km]
modis_lstn = modis_hdf_file[modis_varname_list[1]][()]
modis_lstn = modis_lstn[:, coord_conus_1km]
modis_lstd_all.append(modis_lstd)
modis_lstn_all.append(modis_lstn)
print(modis_hdf_file)
del (modis_hdf_file, modis_varname_list, modis_lstd, modis_lstn)
modis_lstd_all = np.concatenate(modis_lstd_all, axis=0)
modis_lstn_all = np.concatenate(modis_lstn_all, axis=0)
for ife in range(len(modis_lst_group_divide)):
modis_lstd_all_div = modis_lstd_all[:, modis_lst_group_divide[ife]]
modis_lstn_all_div = modis_lstn_all[:, modis_lst_group_divide[ife]]
with h5py.File(path_modis_input_hdf + '/modis_lst_conus_' + str(ife) + '_' + str(yearname[iyr]) + '.hdf5',
'w') as f:
for idv in range(len(var_name)):
f.create_dataset(var_name[idv], data=eval(var_name[idv]), compression='lzf')
f.close()
print(ife)
del (modis_lstd_all_div, modis_lstn_all_div)
del (modis_lstd_all, modis_lstn_all)
########################################################################################################################
# 2. Read the annual MODIS data and subset CONUS region
modis_hdf_files_all = sorted(glob.glob(path_modis_input_hdf + '/*hdf5*'))
for ife in range(10):
modis_lstd_month = []
modis_lstn_month = []
for iyr in range(len(yearname) - 1):
modis_hdf_file = h5py.File(modis_hdf_files_all[iyr + ife * 5], "r")
modis_hdf_varname_list = list(modis_hdf_file.keys())
# 2 layers (day/night)
modis_lstd = modis_hdf_file[modis_hdf_varname_list[0]][()]
modis_lstn = modis_hdf_file[modis_hdf_varname_list[1]][()]
if iyr == 0 or iyr == 4:
modis_lstd = np.delete(modis_lstd, 59, 0)
modis_lstn = np.delete(modis_lstn, 59, 0)
modis_lstd_month.append(modis_lstd)
modis_lstn_month.append(modis_lstn)
print(modis_hdf_files_all[iyr + ife * 5])
modis_lstd_month = np.nanmean(np.stack(modis_lstd_month, axis=0), axis=0)
modis_lstn_month = np.nanmean(np.stack(modis_lstn_month, axis=0), axis=0)
# Save and load the data
with h5py.File(path_modis_input_hdf_proc + '/modis_lst_conus_' + str(ife) + '.hdf5', 'w') as f:
f.create_dataset('modis_lstd_month', data=modis_lstd_month)
f.create_dataset('modis_lstn_month', data=modis_lstn_month)
f.close()
# modis_lstd_2d = np.copy(lmask_init)
# modis_lstd_2d[0, coord_conus_1km_match] = modis_lstd
# modis_lstd_2d = modis_lstd_2d.reshape((len(lat_conus_ease_1km), len(lon_conus_ease_1km)))
# modis_lstd_2d[modis_lstd_2d == 0] = np.nan
#
# amsr2_tbd_2d = np.copy(lmask_init_10km)
# amsr2_tbd_2d[0, coord_conus_amsr2_ind_unique] = amsr2_tbd_1
# amsr2_tbd_2d = amsr2_tbd_2d.reshape((len(lat_world_geo_10km), len(lon_world_geo_10km)))
# amsr2_tbd_2d[amsr2_tbd_2d == 0] = np.nan
########################################################################################################################
# 3. Read the annual AMSR2 data and subset CONUS region
# MODIS and AMSR2 HDF files have the same file name specification
amsr2_files_all = sorted(glob.glob(path_amsr2 + '/Tb_10km_monthly_land/*hdf5*'))
amsr2_files_all = amsr2_files_all[9:]
# Read AMSR2 Tb data and calculate multiple-year average Tb of one Julian year
amsr2_tbd_all = []
amsr2_tbn_all = []
for imo in range(len(monthname)):
amsr2_tbd_month = []
amsr2_tbn_month = []
daysofmonth = np.arange(daysofmonth_nlp[imo])
for iyr in range(len(yearname) - 1):
amsr2_hdf_file = h5py.File(amsr2_files_all[imo + iyr * 12], "r")
amsr2_varname_list = list(amsr2_hdf_file.keys())
# 2 layers (day/night)
amsr2_tbd = amsr2_hdf_file[amsr2_varname_list[0]][daysofmonth,]
amsr2_tbd = amsr2_tbd[:, coord_conus_amsr2_ind_unique_match]
amsr2_tbn = amsr2_hdf_file[amsr2_varname_list[1]][daysofmonth,]
amsr2_tbn = amsr2_tbn[:, coord_conus_amsr2_ind_unique_match]
amsr2_tbd_month.append(amsr2_tbd)
amsr2_tbn_month.append(amsr2_tbn)
del (amsr2_hdf_file, amsr2_varname_list, amsr2_tbd, amsr2_tbn)
print(os.path.basename(amsr2_files_all[imo + iyr * 12]))
amsr2_tbd_month = np.nanmean(np.stack(amsr2_tbd_month, axis=0), axis=0)
amsr2_tbn_month = np.nanmean(np.stack(amsr2_tbn_month, axis=0), axis=0)
amsr2_tbd_all.append(amsr2_tbd_month)
amsr2_tbn_all.append(amsr2_tbn_month)
del (amsr2_tbd_month, amsr2_tbn_month, daysofmonth)
amsr2_tbd_all = np.concatenate(amsr2_tbd_all, axis=0)
amsr2_tbn_all = np.concatenate(amsr2_tbn_all, axis=0)
# Save and load the data
with h5py.File(path_modis_input_hdf_proc + '/amsr2_tb_avg.hdf5', 'w') as f:
f.create_dataset('amsr2_tbd_all', data=amsr2_tbd_all)
f.create_dataset('amsr2_tbn_all', data=amsr2_tbn_all)
f.close()
########################################################################################################################
# 4. Generate TFA coefficients for AMSR2 and MODIS data
# 4.1 AMSR2 variables
f = h5py.File(path_modis_input_hdf_proc + '/amsr2_tb_avg.hdf5', "r")
varname_list = list(f.keys())
for x in range(len(varname_list)):
var_obj = f[varname_list[x]][()]
exec(varname_list[x] + '= var_obj')
del (var_obj)
f.close()
del(f, varname_list)
x = np.arange(365)
y_d = amsr2_tbd_all
y_n = amsr2_tbn_all
model_output_d_all = []
model_output_n_all = []
for ipt in range(y_d.shape[1]):
model_output_d = tclm_model(x, y_d[:, ipt])
model_output_n = tclm_model(x, y_n[:, ipt])
model_output_d_all.append(model_output_d)
model_output_n_all.append(model_output_n)
del(model_output_d, model_output_n)
print(ipt)
y_pred_d = np.stack([model_output_d_all[n][0] for n in range(y_d.shape[1])], axis=1)
r_squared_d = np.stack([model_output_d_all[n][1] for n in range(y_d.shape[1])])
rmse_d = np.stack([model_output_d_all[n][2] for n in range(y_d.shape[1])])
y_pred_n = np.stack([model_output_n_all[n][0] for n in range(y_n.shape[1])], axis=1)
r_squared_n = np.stack([model_output_n_all[n][1] for n in range(y_n.shape[1])])
rmse_n = np.stack([model_output_n_all[n][2] for n in range(y_n.shape[1])])
# Save and load the data
var_name = ['y_pred_d', 'r_squared_d', 'rmse_d', 'y_pred_n', 'r_squared_n', 'rmse_n']
with h5py.File(path_modis_input_hdf_tfa + '/amsr2_tb_tfa.hdf5', 'w') as f:
for idv in range(len(var_name)):
f.create_dataset(var_name[idv], data=eval(var_name[idv]), compression='lzf')
f.close()
# plt.plot(x, y_n[:, -50])
# plt.plot(x, y_pred_n[:, -50], 'r-')
# 4.2 MODIS variables
os.chdir(path_modis_input_hdf_proc)
modis_hdf_proc_files = sorted(glob.glob('*modis*'))
x = np.arange(365)
nan_array = np.empty(365)
nan_array[:] = np.nan
nan_fill = (nan_array, np.nan, np.nan)
for ife in range(9, len(modis_hdf_proc_files)):
f = h5py.File(modis_hdf_proc_files[ife], "r")
varname_list = list(f.keys())
for idx in range(len(varname_list)):
var_obj = f[varname_list[idx]][()]
exec(varname_list[idx] + '= var_obj')
del (var_obj)
f.close()
del(f, varname_list)
y_d = modis_lstd_month
y_n = modis_lstn_month
del(modis_lstd_month, modis_lstn_month)
pixel_divide_ind = \
np.arange(0, y_d.shape[1], y_d.shape[1] // 10)[1:]
pixel_group_divide = np.split(np.arange(y_d.shape[1]), pixel_divide_ind)
pixel_group_divide[-2] = np.concatenate((pixel_group_divide[-2], pixel_group_divide[-1]))
del(pixel_group_divide[-1])
for igp in range(len(pixel_group_divide)):
model_output_d_all = []
model_output_n_all = []
for ipt in range(len(pixel_group_divide[igp])):
if len(np.where(~np.isnan(y_d[:, pixel_group_divide[igp][ipt]]))[0]) != 0 and \
len(np.where(~np.isnan(y_n[:, pixel_group_divide[igp][ipt]]))[0]) != 0:
model_output_d = tclm_model(x, y_d[:, pixel_group_divide[igp][ipt]])
model_output_n = tclm_model(x, y_n[:, pixel_group_divide[igp][ipt]])
else:
model_output_d = nan_fill
model_output_n = nan_fill
model_output_d_all.append(model_output_d)
model_output_n_all.append(model_output_n)
del(model_output_d, model_output_n)
print(str(ife)+'/'+str(igp)+'/'+str(ipt))
y_pred_d = np.stack([model_output_d_all[n][0] for n in range(len(model_output_d_all))], axis=1)
r_squared_d = np.stack([model_output_d_all[n][1] for n in range(len(model_output_d_all))])
rmse_d = np.stack([model_output_d_all[n][2] for n in range(len(model_output_d_all))])
y_pred_n = np.stack([model_output_n_all[n][0] for n in range(len(model_output_n_all))], axis=1)
r_squared_n = np.stack([model_output_n_all[n][1] for n in range(len(model_output_n_all))])
rmse_n = np.stack([model_output_n_all[n][2] for n in range(len(model_output_n_all))])
# Save and load the data
var_name = ['y_pred_d', 'r_squared_d', 'rmse_d', 'y_pred_n', 'r_squared_n', 'rmse_n']
with h5py.File(path_modis_input_hdf_tfa + '/' + os.path.basename(modis_hdf_proc_files[ife]).split('.')[0] + \
'_' + str(igp) + '_tfa.hdf5', 'w') as f:
for idv in range(len(var_name)):
f.create_dataset(var_name[idv], data=eval(var_name[idv]), compression='lzf')
f.close()
del(model_output_d_all, model_output_n_all, y_pred_d, r_squared_d, rmse_d, y_pred_n, r_squared_n, rmse_n)
del(y_d, y_n)
# # Reshape to 2d matrix
# lmask_init = np.empty((len(lat_conus_ease_1km), len(lon_conus_ease_1km)), dtype='float32')
# lmask_init = lmask_init.reshape(1, -1)
# lmask_init[:] = 0
#
# lmask_init_10km = np.empty((len(lat_world_geo_10km), len(lon_world_geo_10km)), dtype='float32')
# lmask_init_10km = lmask_init_10km.reshape(1, -1)
# lmask_init_10km[:] = 0