Newer
Older
"metadata": {},
"outputs": [],
"source": [
"import xarray as xr\n",
"from dask.distributed import Client\n",
"import numpy as np\n",
"from typing import List, Tuple, Union\n",
"import warnings\n",
"import gc\n",
"from parameters.params_samir_class import samir_parameters\n",
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
"metadata": {},
"outputs": [],
"source": [
"\n",
"\n",
"def rasterize_samir_parameters(csv_param_file: str, empty_dataset: xr.Dataset, land_cover_raster: str, chunk_size: dict) -> Tuple[xr.Dataset, dict]:\n",
" \"\"\"\n",
" Creates a raster `xarray` dataset from the csv parameter file, the land cover raster and an empty dataset\n",
" that contains the right structure (emptied ndvi dataset for example). For each parameter, the function loops\n",
" on land cover classes to fill the raster.\n",
"\n",
" ## Arguments\n",
" 1. csv_param_file: `str`\n",
" path to csv paramter file\n",
" 2. empty_dataset: `xr.Dataset`\n",
" empty dataset that contains the right structure (emptied ndvi dataset for example).\n",
" 3. land_cover_raster: `str`\n",
" path to land cover netcdf raster\n",
" 4. chunk_size: `dict`\n",
" chunk_size for dask computation\n",
"\n",
" ## Returns\n",
" 1. parameter_dataset: `xr.Dataset`\n",
" the dataset containing all the rasterized Parameters\n",
" 2. scale_factor: `dict`\n",
" dictionnary containing the scale factors for each parameter\n",
" \"\"\"\n",
" \n",
" # Load samir params into an object\n",
" table_param = samir_parameters(csv_param_file)\n",
" \n",
" # Set general variables\n",
" class_count = table_param.table.shape[1] - 2 # remove dtype and default columns\n",
" \n",
" # Open land cover raster\n",
" land_cover = xr.open_dataarray(land_cover_raster, chunks = chunk_size)\n",
" \n",
" # Create dataset\n",
" parameter_dataset = empty_dataset.copy(deep = True)\n",
" \n",
" # Create dictionnary containing the scale factors\n",
" scale_factor = {}\n",
" \n",
" # Loop on samir parameters and create \n",
" for parameter in table_param.table.index[1:]:\n",
" \n",
" # Create new variable and set attributes\n",
" parameter_dataset[parameter] = land_cover.copy(deep = True).astype('f4')\n",
" parameter_dataset[parameter].attrs['name'] = parameter\n",
" parameter_dataset[parameter].attrs['description'] = 'cf SAMIR Doc for detail'\n",
" parameter_dataset[parameter].attrs['scale factor'] = str(table_param.table.loc[table_param.table.index == parameter]['scale_factor'].values[0])\n",
" \n",
" # Assigne value in dictionnary\n",
" scale_factor[parameter] = 1/int(table_param.table.loc[table_param.table.index == parameter]['scale_factor'].values[0])\n",
" \n",
" # Loop on classes to set parameter values for each class\n",
" for class_val, class_name in zip(range(1, class_count + 1), table_param.table.columns[2:]):\n",
" \n",
" # Parameter values are multiplied by the scale factor in order to store all values as int16 types\n",
" # These values are then rounded to make sure there isn't any decimal point issues when casting the values to int16\n",
" parameter_dataset[parameter].values = np.where(parameter_dataset[parameter].values == class_val, round(table_param.table.loc[table_param.table.index == parameter][class_name].values[0]*table_param.table.loc[table_param.table.index == parameter]['scale_factor'].values[0]), parameter_dataset[parameter].values).astype('f4')\n",
" \n",
" # Return dataset converted to 'int16' data type to reduce memory usage\n",
" # and scale_factor dictionnary for later conversion\n",
" return parameter_dataset, scale_factor\n",
"\n",
"\n",
"def setup_time_loop(calculation_variables_t1: List[str], calculation_variables_t2: List[str], empty_dataset: xr.Dataset) -> Tuple[xr.Dataset, xr.Dataset]:\n",
" \"\"\"\n",
" Creates two temporary `xarray Datasets` that will be used in the SAMIR time loop.\n",
" `variables_t1` corresponds to the variables for the previous day and `variables_t2`\n",
" corresponds to the variables for the current day. After each loop, `variables_t1`\n",
" takes the value of `variables_t2` for the corresponding variables.\n",
"\n",
" ## Arguments\n",
" 1. calculation_variables_t1: `List[str]`\n",
" list of strings containing the variable names\n",
" for the previous day dataset\n",
" 2. calculation_variables_t2: `List[str]`\n",
" list of strings containing the variable names\n",
" for the current day dataset\n",
" 3. empty_dataset: `xr.Dataset`\n",
" empty dataset that contains the right structure\n",
"\n",
" ## Returns\n",
" 1. variables_t1: `xr.Dataset`\n",
" output dataset for previous day\n",
" 2. variables_t2: `xr.Dataset`\n",
" output dataset for current day\n",
" \"\"\"\n",
" \n",
" # Create new dataset\n",
" variables_t1 = empty_dataset.copy(deep = True)\n",
" \n",
" # Create empty DataArray for each variable\n",
" for variable in calculation_variables_t1:\n",
" \n",
" # Assign new empty DataArray\n",
" variables_t1[variable] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'float32'))\n",
" variables_t1[variable].attrs['name'] = variable # set name in attributes\n",
" \n",
" # Create new dataset\n",
" variables_t2 = empty_dataset.copy(deep = True)\n",
" \n",
" # Create empty DataArray for each variable\n",
" for variable in calculation_variables_t2:\n",
" \n",
" # Assign new empty DataArray\n",
" variables_t2[variable] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'float32'))\n",
" variables_t2[variable].attrs['name'] = variable # set name in attributes\n",
" \n",
" return variables_t1, variables_t2\n",
"\n",
"\n",
"def prepare_outputs(empty_dataset: xr.Dataset, additional_outputs: List[str] = None) -> xr.Dataset:\n",
" \"\"\"\n",
" Creates the `xarray Dataset` containing the outputs of the SAMIR model that will be saved.\n",
" Additional variables can be saved by adding their names to the `additional_outputs` list.\n",
"\n",
" ## Arguments\n",
" 1. empty_dataset: `xr.Dataset`\n",
" empty dataset that contains the right structure\n",
" 2. additional_outputs: `List[str]`\n",
" list of additional variable names to be saved\n",
"\n",
" ## Returns\n",
" 1. model_outputs: `xr.Dataset`\n",
" model outputs to be saved\n",
" \"\"\"\n",
" \n",
" # Evaporation and Transpiraion\n",
" model_outputs = empty_dataset.copy(deep = True)\n",
" \n",
" model_outputs['E'] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" model_outputs['E'].attrs['units'] = 'mm'\n",
" model_outputs['E'].attrs['standard_name'] = 'Evaporation'\n",
" model_outputs['E'].attrs['description'] = 'Accumulated daily evaporation in milimeters'\n",
" model_outputs['E'].attrs['scale factor'] = '1'\n",
" \n",
" model_outputs['Tr'] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" model_outputs['Tr'].attrs['units'] = 'mm'\n",
" model_outputs['Tr'].attrs['standard_name'] = 'Transpiration'\n",
" model_outputs['Tr'].attrs['description'] = 'Accumulated daily plant transpiration in milimeters'\n",
" model_outputs['Tr'].attrs['scale factor'] = '1'\n",
" \n",
" # Soil Water Content\n",
" model_outputs['SWCe'] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" model_outputs['SWCe'].attrs['units'] = 'mm'\n",
" model_outputs['SWCe'].attrs['standard_name'] = 'Soil Water Content of the evaporative zone'\n",
" model_outputs['SWCe'].attrs['description'] = 'Soil water content of the evaporative zone in milimeters'\n",
" model_outputs['SWCe'].attrs['scale factor'] = '1'\n",
" \n",
" model_outputs['SWCr'] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" model_outputs['SWCr'].attrs['units'] = 'mm'\n",
" model_outputs['SWCr'].attrs['standard_name'] = 'Soil Water Content of the root zone'\n",
" model_outputs['SWCr'].attrs['description'] = 'Soil water content of the root zone in milimeters'\n",
" model_outputs['SWCr'].attrs['scale factor'] = '1'\n",
" \n",
" # Irrigation\n",
" model_outputs['Irr'] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" model_outputs['Irr'].attrs['units'] = 'mm'\n",
" model_outputs['Irr'].attrs['standard_name'] = 'Irrigation'\n",
" model_outputs['Irr'].attrs['description'] = 'Simulated daily irrigation in milimeters'\n",
" model_outputs['Irr'].attrs['scale factor'] = '1'\n",
" \n",
" # Deep Percolation\n",
" model_outputs['DP'] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" model_outputs['DP'].attrs['units'] = 'mm'\n",
" model_outputs['DP'].attrs['standard_name'] = 'Deep Percolation'\n",
" model_outputs['DP'].attrs['description'] = 'Simulated daily Deep Percolation in milimeters'\n",
" model_outputs['DP'].attrs['scale factor'] = '1'\n",
" \n",
" if additional_outputs:\n",
" for var in additional_outputs:\n",
" model_outputs[var] = (empty_dataset.dims, np.zeros(tuple(empty_dataset.dims[d] for d in list(empty_dataset.dims)), dtype = 'int16'))\n",
" \n",
" return model_outputs\n",
"\n",
"\n",
"def xr_maximum(ds: xr.DataArray, value: Union[xr.DataArray, float, int]) -> xr.DataArray:\n",
" \"\"\"\n",
" Equivalent of `numpy.maximum(ds, value)` for xarray DataArrays\n",
"\n",
" ## Arguments\n",
" 1. ds: `xr.DataArray`\n",
" datarray to compare\n",
" 2. value: `Union[xr.DataArray, float, int]`\n",
" value (scalar or dataarray) to compare\n",
"\n",
" ## Returns\n",
" 1. output: `xr.DataArray`\n",
" resulting dataarray with maximum value element-wise\n",
" \"\"\"\n",
" return xr.where(ds <= value, value, ds, keep_attrs = True)\n",
"\n",
"\n",
"def xr_minimum(ds: xr.DataArray, value: Union[xr.DataArray, float, int]) -> xr.DataArray:\n",
" \"\"\"\n",
" Equivalent of `numpy.minimum(ds, value)` for xarray DataArrays\n",
"\n",
" ## Arguments\n",
" 1. ds: `xr.DataArray`\n",
" datarray to compare\n",
" 2. value: `Union[xr.DataArray, float, int]`\n",
" value (scalar or dataarray) to compare\n",
"\n",
" ## Returns\n",
" 1. output: `xr.DataArray`\n",
" resulting dataarray with minimum value element-wise\n",
" \"\"\"\n",
" return xr.where(ds >= value, value, ds, keep_attrs = True)\n",
"\n",
"\n",
"def calculate_diff_re(TAW: xr.DataArray, Dr: xr.DataArray, Zr: xr.DataArray, RUE: xr.DataArray, De: xr.DataArray, FCov: xr.DataArray, Ze_: xr.DataArray, DiffE_: xr.DataArray, scale_dict: dict) -> xr.DataArray:\n",
" \"\"\"\n",
" Calculates the diffusion between the top soil layer and the root layer.\n",
"\n",
" ## Arguments\n",
" 1. TAW: `xr.DataArray`\n",
" water capacity of root layer\n",
" 2. Dr: `xr.DataArray`\n",
" depletion of root layer\n",
" 3. Zr: `xr.DataArray`\n",
" height of root layer\n",
" 4. RUE: `xr.DataArray`\n",
" total available surface water\n",
" 5. De: `xr.DataArray`\n",
" depletion of the evaporative layer\n",
" 6. FCov: `xr.DataArray`\n",
" fraction cover of plants\n",
" 7. Ze_: `xr.DataArray`\n",
" height of evaporative layer (paramter)\n",
" 8. DiffE_: `xr.DataArray`\n",
" diffusion coefficient between evaporative\n",
" and root layers (unitless, parameter)\n",
" 9. scale_dict: `dict`\n",
" dictionnary containing the scale factors for\n",
" the rasterized parameters\n",
"\n",
" ## Returns\n",
" 1. diff_re: `xr.Dataset`\n",
" the diffusion between the top soil layer and\n",
" the root layer\n",
" \"\"\"\n",
" \n",
" # Temporary variables to make calculation easier to read\n",
" tmp1 = (((TAW - Dr) / Zr - (RUE - De) / (scale_dict['Ze'] * Ze_)) / FCov) * (scale_dict['DiffE'] * DiffE_)\n",
" tmp2 = ((TAW * scale_dict['Ze'] * Ze_) - (RUE - De - Dr) * Zr) / (Zr + scale_dict['Ze'] * Ze_) - Dr\n",
" \n",
" # Calculate diffusion according to SAMIR equation\n",
" diff_re = xr.where(tmp1 < 0, xr_maximum(tmp1, tmp2), xr_minimum(tmp1, tmp2))\n",
"\n",
" # Return zero values where the 'DiffE' parameter is equal to 0\n",
" return xr.where(DiffE_ == 0, 0, diff_re)\n",
"\n",
"\n",
"def calculate_diff_dr(TAW: xr.DataArray, TDW: xr.DataArray, Dr: xr.DataArray, Zr: xr.DataArray, Dd: xr.DataArray, FCov: xr.DataArray, Zsoil_: xr.DataArray, DiffR_: xr.DataArray, scale_dict: dict) -> xr.DataArray:\n",
" \"\"\"\n",
" Calculates the diffusion between the root layer and the deep layer.\n",
"\n",
" ## Arguments\n",
" 1. TAW: `xr.DataArray`\n",
" water capacity of root layer\n",
" 2. TDW: `xr.DataArray`\n",
" water capacity of deep layer\n",
" 3. Dr: `xr.DataArray`\n",
" depletion of root layer\n",
" 4. Zr: `xr.DataArray`\n",
" height of root layer\n",
" 5. Dd: `xr.DataArray`\n",
" depletion of deep layer\n",
" 6. FCov: `xr.DataArray`\n",
" fraction cover of plants\n",
" 7. Zsoil_: `xr.DataArray`\n",
" total height of soil (paramter)\n",
" 8. DiffR_: `xr.DataArray`\n",
" Diffusion coefficient between root\n",
" and deep layers (unitless, parameter)\n",
" 9. scale_dict: `dict`\n",
" dictionnary containing the scale factors for\n",
" the rasterized parameters\n",
"\n",
" ## Returns\n",
" 1. diff_dr: `xr.Dataset`\n",
" the diffusion between the root layer and the\n",
" deep layer\n",
" \"\"\"\n",
" \n",
" # Temporary variables to make calculation easier to read\n",
" tmp1 = (((TDW - Dd) / (scale_dict['Zsoil'] * Zsoil_ - Zr) - (TAW - Dr) / Zr) / FCov) * scale_dict['DiffR'] * DiffR_\n",
" tmp2 = (TDW *Zr - (TAW - Dr - Dd) * (scale_dict['Zsoil'] * Zsoil_ - Zr)) / (scale_dict['Zsoil'] * Zsoil_) - Dd\n",
" \n",
" # Calculate diffusion according to SAMIR equation\n",
" diff_dr = xr.where(tmp1 < 0, xr_maximum(tmp1, tmp2), xr_minimum(tmp1, tmp2))\n",
" \n",
" # Return zero values where the 'DiffR' parameter is equal to 0\n",
" return xr.where(DiffR_ == 0, 0, diff_dr)\n",
"\n",
"\n",
"def calculate_W(TEW: xr.DataArray, Dei: xr.DataArray, Dep: xr.DataArray, fewi: xr.DataArray, fewp: xr.DataArray) -> xr.DataArray:\n",
" \"\"\"\n",
" Calculate W, the weighting factor to split the energy available\n",
" for evaporation depending on the difference in water availability\n",
" in the two evaporation components, ensuring that the larger and\n",
" the wetter, the more the evaporation occurs from that component\n",
"\n",
" ## Arguments\n",
" 1. TEW: `xr.DataArray`\n",
" water capacity of evaporative layer\n",
" 2. Dei: `xr.DataArray`\n",
" depletion of the evaporative layer\n",
" (irrigation part)\n",
" 3. Dep: `xr.DataArray`\n",
" depletion of the evaporative layer\n",
" (precipitation part)\n",
" 4. fewi: `xr.DataArray`\n",
" soil fraction which is wetted by irrigation\n",
" and exposed to evaporation\n",
" 5. fewp: `xr.DataArray`\n",
" soil fraction which is wetted by precipitation\n",
" and exposed to evaporation\n",
"\n",
" ## Returns\n",
" 1. W: `xr.DataArray`\n",
" weighting factor W\n",
" \"\"\"\n",
" \n",
" # Temporary variables to make calculation easier to read\n",
" tmp = fewi * (TEW - Dei)\n",
" \n",
" # Calculate the weighting factor to split the energy available for evaporation\n",
" W = 1 / (1 + (fewp * (TEW - Dep) / tmp ))\n",
"\n",
" # Return W \n",
" return xr.where(tmp > 0, W, 0)\n",
"\n",
"\n",
"def calculate_Kr(TEW: xr.DataArray, De: xr.DataArray, REW_: xr.DataArray, scale_dict: dict) -> xr.DataArray:\n",
" \"\"\"\n",
" calculates of the reduction coefficient for evaporation dependent \n",
" on the amount of water in the soil using the FAO-56 method\n",
"\n",
" ## Arguments\n",
" 1. TEW: `xr.DataArray`\n",
" water capacity of evaporative layer\n",
" 2. De: `xr.DataArray`\n",
" depletion of evaporative layer\n",
" 3. REW_: `xr.DataArray`\n",
" readily evaporable water\n",
" 4. scale_dict: `dict`\n",
" dictionnary containing the scale factors for\n",
" the rasterized parameters\n",
"\n",
" ## Returns\n",
" 1. Kr: `xr.DataArray`\n",
" Kr coefficient\n",
" \"\"\"\n",
" \n",
" # Formula for calculating Kr\n",
" Kr = (TEW - De) / (TEW - scale_dict['REW'] * REW_)\n",
" \n",
" # Return Kr\n",
" return xr_maximum(0, xr_minimum(Kr, 1))\n",
"\n",
"\n",
"def update_Dr(TAW: xr.DataArray, TDW: xr.DataArray, Zr: xr.DataArray, TAW0: xr.DataArray, TDW0: xr.DataArray, Dr0: xr.DataArray, Dd0: xr.DataArray, Zr0: xr.DataArray) -> xr.DataArray:\n",
" \"\"\"\n",
" Return the updated depletion for the root layer\n",
"\n",
" ## Arguments\n",
" 1. TAW: `xr.DataArray`\n",
" water capacity of root layer for current day\n",
" 2. TDW: `xr.DataArray`\n",
" water capacity of deep layer for current day\n",
" 3. Zr: `xr.DataArray`\n",
" root layer height for current day\n",
" 4. TAW0: `xr.DataArray`\n",
" water capacity of root layer for previous day\n",
" 5. TDW0: `xr.DataArray`\n",
" water capacity of deep layer for previous day\n",
" 6. Dr0: `xr.DataArray`\n",
" depletion of the root layer for previous day\n",
" 7. Dd0: `xr.DataArray`\n",
" depletion of the deep laye for previous day\n",
" 8. Zr0: `xr.DataArray`\n",
" root layer height for previous day\n",
"\n",
" ## Returns\n",
" 1. output: `xr.DataArray`\n",
" updated depletion for the root layer\n",
" \"\"\"\n",
" \n",
" # Temporary variables to make calculation easier to read\n",
" tmp1 = xr_maximum(Dr0 + Dd0 * (TAW - TAW0) / TDW0, 0)\n",
" tmp2 = xr_minimum(Dr0 + Dd0 * (TAW - TAW0) / TDW0, TDW)\n",
"\n",
" # Return updated Dr\n",
" return xr.where(Zr > Zr0, tmp1, tmp2)\n",
"\n",
"\n",
"def update_Dd(TAW: xr.DataArray, TDW: xr.DataArray, Zr: xr.DataArray, TAW0: xr.DataArray, TDW0: xr.DataArray, Dd0: xr.DataArray, Zr0: xr.DataArray) -> xr.DataArray:\n",
" \"\"\"\n",
" Return the updated depletion for the deep layer\n",
"\n",
" ## Arguments\n",
" 1. TAW: `xr.DataArray`\n",
" water capacity of root layer for current day\n",
" 2. TDW: `xr.DataArray`\n",
" water capacity of deep layer for current day\n",
" 3. TAW0: `xr.DataArray`\n",
" water capacity of root layer for previous day\n",
" 5. TDW0: `xr.DataArray`\n",
" water capacity of deep layer for previous day\n",
" 6. Dd0: `xr.DataArray`\n",
" depletion of the deep laye for previous day\n",
" 7. Zr0: `xr.DataArray`\n",
" root layer height for previous day\n",
"\n",
" ## Returns\n",
" 1. output: `xr.DataArray`\n",
" updated depletion for the deep layer\n",
" \"\"\"\n",
" \n",
" # Temporary variables to make calculation easier to read\n",
" tmp1 = xr_maximum(Dd0 - Dd0 * (TAW - TAW0) / TDW0, 0)\n",
" tmp2 = xr_minimum(Dd0 - Dd0 * (TAW - TAW0) / TDW0, TDW)\n",
" \n",
" # Return updated Dd\n",
" return xr.where(Zr > Zr0, tmp1, tmp2)\n",
"\n",
"\n",
"def run_samir(json_config_file: str, csv_param_file: str, ndvi_cube_path: str, weather_cube_path: str, soil_params_path: str, land_cover_path: str, chunk_size: dict, save_path: str, max_GB: int = 2) -> None:\n",
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
" \n",
" # warnings.simplefilter(\"error\", category = RuntimeWarning())\n",
" warnings.filterwarnings(\"ignore\", message=\"invalid value encountered in cast\")\n",
" warnings.filterwarnings(\"ignore\", message=\"invalid value encountered in divide\")\n",
" np.errstate(all = 'raise')\n",
" gc.disable()\n",
" \n",
" #============ General parameters ============#\n",
" config_params = config(json_config_file)\n",
" calculation_variables_t2 = ['diff_rei', 'diff_rep', 'diff_dr' , 'Dd', 'De', 'Dei', 'Dep', 'Dr', 'FCov', 'Irrig', 'Kcb', 'Kei', 'Kep', 'Ks', 'Kti', 'Ktp', 'RUE', 'TAW', 'TDW', 'TEW', 'Tei', 'Tep', 'W', 'Zr', 'fewi', 'fewp']\n",
" calculation_variables_t1 = ['Dr', 'Dd', 'TAW', 'TDW', 'Zr']\n",
" \n",
" #============ Manage inputs ============#\n",
" # NDVI\n",
" ndvi_cube = xr.open_dataset(ndvi_cube_path, chunks = chunk_size).astype('f4')\n",
" \n",
" # # Create a daily DateTimeIndex for the desired date range\n",
" # daily_index = pd.date_range(start = config_params.start_date, end = config_params.end_date, freq = 'D')\n",
"\n",
" # # Resample the dataset to a daily frequency and reindex with the new DateTimeIndex\n",
" # ndvi_cube = ndvi_cube.resample(time = '1D').asfreq().reindex(time = daily_index)\n",
"\n",
" # # Interpolate the dataset along the time dimension to fill nan values\n",
" # ndvi_cube = ndvi_cube.interpolate_na(dim = 'time', method = 'linear', fill_value = 'extrapolate').astype('u1')\n",
" \n",
" # Weather\n",
" weather_cube = xr.open_dataset(weather_cube_path, chunks = chunk_size).astype('f4')\n",
" \n",
" # Soil\n",
" soil_params = xr.open_dataset(soil_params_path, chunks = chunk_size).astype('f4')\n",
" \n",
" # SAMIR Parameters\n",
" param_dataset, scale_factor = rasterize_samir_parameters(csv_param_file, ndvi_cube.drop_vars(['ndvi', 'time']), land_cover_path, chunk_size = chunk_size)\n",
" \n",
" # SAMIR Variables\n",
" variables_t1, variables_t2 = setup_time_loop(calculation_variables_t1, calculation_variables_t2, ndvi_cube.drop_vars(['ndvi', 'time']))\n",
" \n",
" # Manage loading of data based on disk size of inputs\n",
" if ndvi_cube.nbytes < max_GB * (1024)**3:\n",
" ndvi_cube.load()\n",
" \n",
" if weather_cube.nbytes < max_GB * (1024)**3:\n",
" weather_cube.load()\n",
"\n",
" #============ Prepare outputs ============#\n",
" model_outputs = prepare_outputs(ndvi_cube.drop_vars(['ndvi']))\n",
" \n",
" #============ Prepare time iterations ============#\n",
" dates = ndvi_cube.time.values\n",
" \n",
" #============ Create aliases for better readability ============#\n",
" \n",
" # Variables for current day\n",
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
" diff_rei = variables_t2.diff_rei\n",
" diff_rep = variables_t2.diff_rep\n",
" diff_dr = variables_t2.diff_dr\n",
" Dd = variables_t2.Dd\n",
" De = variables_t2.De\n",
" Dei = variables_t2.Dei\n",
" Dep = variables_t2.Dep\n",
" Dr = variables_t2.Dr\n",
" FCov = variables_t2.FCov\n",
" Irrig = variables_t2.Irrig\n",
" Kcb = variables_t2.Kcb\n",
" Kei = variables_t2.Kei\n",
" Kep = variables_t2.Kep\n",
" Ks = variables_t2.Ks\n",
" Kti = variables_t2.Kti\n",
" Ktp = variables_t2.Ktp\n",
" RUE = variables_t2.RUE\n",
" TAW = variables_t2.TAW\n",
" TDW = variables_t2.TDW\n",
" TEW = variables_t2.TEW\n",
" Tei = variables_t2.Tei\n",
" Tep = variables_t2.Tep\n",
" Zr = variables_t2.Zr\n",
" W = variables_t2.W\n",
" fewi = variables_t2.fewi\n",
" fewp = variables_t2.fewp\n",
" \n",
" # Variables for previous day\n",
" TAW0 = variables_t1.TAW\n",
" TDW0 = variables_t1.TDW\n",
" Dr0 = variables_t1.Dr\n",
" Dd0 = variables_t1.Dd\n",
" Zr0 = variables_t1.Zr\n",
" \n",
" # Parameters\n",
" # Parameters have an underscore (_) behind their name for recognition \n",
" DiffE_ = param_dataset.DiffE\n",
" DiffR_ = param_dataset.DiffR\n",
" FW_ = param_dataset.FW\n",
" Fc_stop_ = param_dataset.Fc_stop\n",
" FmaxFC_ = param_dataset.FmaxFC\n",
" Foffset_ = param_dataset.Foffset\n",
" Fslope_ = param_dataset.Fslope\n",
" Init_RU_ = param_dataset.Init_RU\n",
" Irrig_auto_ = param_dataset.Irrig_auto\n",
" Kcmax_ = param_dataset.Kcmax\n",
" KmaxKcb_ = param_dataset.KmaxKcb\n",
" Koffset_ = param_dataset.Koffset\n",
" Kslope_ = param_dataset.Kslope\n",
" Lame_max_ = param_dataset.Lame_max\n",
" REW_ = param_dataset.REW\n",
" Ze_ = param_dataset.Ze\n",
" Zsoil_ = param_dataset.Zsoil\n",
" maxZr_ = param_dataset.maxZr\n",
" minZr_ = param_dataset.minZr\n",
" p_ = param_dataset.p\n",
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
" \n",
" # scale factors\n",
" # Scale factors have the following name scheme : s_ + parameter_name\n",
" s_DiffE = scale_factor['DiffE']\n",
" s_DiffR = scale_factor['DiffR']\n",
" s_FW = scale_factor['FW']\n",
" s_Fc_stop = scale_factor['Fc_stop']\n",
" s_FmaxFC = scale_factor['FmaxFC']\n",
" s_Foffset = scale_factor['Foffset']\n",
" s_Fslope = scale_factor['Fslope']\n",
" s_Init_RU = scale_factor['Init_RU']\n",
" # s_Irrig_auto = scale_factor['Irrig_auto']\n",
" s_Kcmax = scale_factor['Kcmax']\n",
" s_KmaxKcb = scale_factor['KmaxKcb']\n",
" s_Koffset = scale_factor['Koffset']\n",
" s_Kslope = scale_factor['Kslope']\n",
" s_Lame_max = scale_factor['Lame_max']\n",
" s_REW = scale_factor['REW']\n",
" s_Ze = scale_factor['Ze']\n",
" s_Zsoil = scale_factor['Zsoil']\n",
" s_maxZr = scale_factor['maxZr']\n",
" s_minZr = scale_factor['minZr']\n",
" s_p = scale_factor['p']\n",
" \n",
" #============ First day initialization ============#\n",
" # Fraction cover\n",
" FCov = s_Fslope * Fslope_ * (ndvi_cube.ndvi.sel({'time': dates[0]})/255) + s_Foffset * Foffset_\n",
" FCov = xr_minimum(xr_maximum(FCov, 0), s_Fc_stop * Fc_stop_)\n",
" \n",
" # Root depth upate\n",
" Zr = s_minZr * minZr_ + (FCov / (s_FmaxFC * FmaxFC_)) * s_maxZr * (maxZr_ - minZr_)\n",
" \n",
" # Water capacities\n",
" TEW = (soil_params.FC - soil_params.WP/2) * s_Ze * Ze_\n",
" RUE = (soil_params.FC - soil_params.WP) * s_Ze * Ze_\n",
" TAW = (soil_params.FC - soil_params.WP) * Zr\n",
" TDW = (soil_params.FC - soil_params.WP) * (s_Zsoil * Zsoil_ - Zr) # Zd = Zsoil - Zr\n",
" \n",
" # Depletions\n",
" Dei = RUE * (1 - s_Init_RU * Init_RU_)\n",
" Dep = RUE * (1 - s_Init_RU * Init_RU_)\n",
" Dr = TAW * (1 - s_Init_RU * Init_RU_)\n",
" Dd = TDW * (1 - s_Init_RU * Init_RU_)\n",
" \n",
" # Irrigation ==============!!!!!\n",
" Irrig = xr_minimum(xr_maximum(Dr - weather_cube.tp.sel({'time': dates[0]}) / 1000, 0), s_Lame_max * Lame_max_) * Irrig_auto_\n",
" Irrig = xr.where(Dr > TAW * s_p * p_, Irrig, 0)\n",
" \n",
" # Kcb\n",
" Kcb = xr_minimum(s_Kslope * Kslope_ * (ndvi_cube.ndvi.sel({'time': dates[0]}) / 255) + s_Koffset * Koffset_, s_KmaxKcb * KmaxKcb_)\n",
" \n",
" # Update depletions with rainfall and/or irrigation \n",
" ## DP\n",
" model_outputs.DP.loc[{'time': dates[0]}] = -xr_minimum(Dd + xr_minimum(Dr - (weather_cube.tp.sel({'time': dates[0]}) / 1000) - Irrig, 0), 0)\n",
" Dei = xr_minimum(xr_maximum(Dei - (weather_cube.tp.sel({'time': dates[0]}) / 1000) - Irrig / (s_FW * FW_ / 100), 0), TEW)\n",
" Dep = xr_minimum(xr_maximum(Dep - (weather_cube.tp.sel({'time': dates[0]}) / 1000), 0), TEW)\n",
" \n",
" fewi = xr_minimum(1 - FCov, (s_FW * FW_ / 100))\n",
" fewp = 1 - FCov - fewi\n",
" \n",
" De = (Dei * fewi + Dep * fewp) / (fewi + fewp)\n",
" # variables_t1['De'] = xr.where(variables_t1['De'].isfinite(), variables_t1['De'], variables_t1['Dei'] * (s_FW * FW_ / 100) + variables_t1['Dep'] * (1 - (s_FW * FW_ / 100))) #================= find replacement for .isfinite() method !!!!!!!!!\n",
"\n",
" ## Dr\n",
" Dr = xr_minimum(xr_maximum(Dr - (weather_cube.tp.sel({'time': dates[0]}) / 1000) - Irrig, 0), TAW)\n",
" Dd = xr_minimum(xr_maximum(Dd + xr_minimum(Dr - (weather_cube.tp.sel({'time': dates[0]}) / 1000) - Irrig, 0), 0), TDW)\n",
" \n",
" # Diffusion coefficients\n",
" diff_rei = calculate_diff_re(TAW, Dr, Zr, RUE, Dei, FCov, Ze_, DiffE_, scale_factor)\n",
" diff_rep = calculate_diff_re(TAW, Dr, Zr, RUE, Dep, FCov, Ze_, DiffE_, scale_factor)\n",
" diff_dr = calculate_diff_dr(TAW, TDW, Dr, Zr, Dd, FCov, Zsoil_, DiffR_, scale_factor) \n",
" \n",
" # Weighing factor W\n",
" W = calculate_W(TEW, Dei, Dep, fewi, fewp)\n",
" \n",
" # Soil water contents\n",
" model_outputs.SWCe.loc[{'time': dates[0]}] = 1 - De/TEW\n",
" model_outputs.SWCr.loc[{'time': dates[0]}] = 1 - Dr/TAW\n",
" \n",
" # Water Stress coefficient\n",
" Ks = xr_minimum((TAW - Dr) / (TAW * (1 - s_p * p_)), 1)\n",
" \n",
" # Reduction coefficient for evaporation\n",
" Kei = xr_minimum(W * calculate_Kr(TEW, Dei, REW_, scale_factor) * (s_Kcmax * Kcmax_ - Kcb), fewi * s_Kcmax * Kcmax_)\n",
" Kep = xr_minimum((1 - W) * calculate_Kr(TEW, Dep, REW_, scale_factor) * (s_Kcmax * Kcmax_ - Kcb), fewp * s_Kcmax * Kcmax_)\n",
" \n",
" # Prepare coefficients for evapotranspiration\n",
" Kti = xr_minimum(((s_Ze * Ze_ / Zr)**6) * (1 - Dei / TEW) / xr_maximum(1 - Dr / TAW, 0.001), 1)\n",
" Ktp = xr_minimum(((s_Ze * Ze_ / Zr)**6) * (1 - Dep / TEW) / xr_maximum(1 - Dr / TAW, 0.001), 1)\n",
" Tei = Kti * Ks * Kcb * weather_cube.ET0.sel({'time': dates[0]}) / 1000\n",
" Tep = Ktp * Ks * Kcb * weather_cube.ET0.sel({'time': dates[0]}) / 1000\n",
" Dei = xr.where(fewi > 0, xr_minimum(xr_maximum(Dei + (weather_cube.ET0.sel({'time': dates[0]}) / 1000) * Kei / fewi + Tei - diff_rei, 0), TEW), xr_minimum(xr_maximum(Dei + Tei - diff_rei, 0), TEW))\n",
" Dep = xr.where(fewp > 0, xr_minimum(xr_maximum(Dep + (weather_cube.ET0.sel({'time': dates[0]}) / 1000) * Kep / fewp + Tep - diff_rep, 0), TEW), xr_minimum(xr_maximum(Dep + Tep - diff_rep, 0), TEW))\n",
" \n",
" De = (Dei * fewi + Dep * fewp) / (fewi + fewp)\n",
" # De = xr.where(De.isfinite(), De, Dei * (s_FW * FW_ / 100) + Dep * (1 - (s_FW * FW_ / 100))) #================= find replacement for .isfinite() method !!!!!!!!!\n",
" \n",
" # Evaporation\n",
" model_outputs.E.loc[{'time': dates[0]}] = xr_maximum((Kei + Kep) * weather_cube.ET0.sel({'time': dates[0]}) / 1000, 0)\n",
" model_outputs.Tr.loc[{'time': dates[0]}] = Kcb * Ks * weather_cube.ET0.sel({'time': dates[0]}) / 1000\n",
" \n",
" # Potential evapotranspiration and evaporative fraction ??\n",
" \n",
" # Update depletions (root and deep zones) at the end of the day\n",
" Dr = xr_minimum(xr_maximum(Dr + model_outputs.E.loc[{'time': dates[0]}] + model_outputs.Tr.loc[{'time': dates[0]}] - diff_dr, 0), TAW)\n",
" Dd = xr_minimum(xr_maximum(Dd + diff_dr, 0), TDW)\n",
" \n",
" # Write outputs\n",
" model_outputs.Irr.loc[{'time': dates[0]}] = Irrig\n",
" \n",
" # Update variable_t1 values\n",
" for variable in calculation_variables_t1:\n",
" variables_t1[variable] = variables_t2[variable].copy(deep = True)\n",
" \n",
" #============ Time loop ============#\n",
" for i in range(1, len(dates)):\n",
" \n",
" # Update variables\n",
" ## Fraction cover\n",
" FCov = s_Fslope * Fslope_ * (ndvi_cube.ndvi.sel({'time': dates[i]})/255) + s_Foffset * Foffset_\n",
" FCov = xr_minimum(xr_maximum(FCov, 0), s_Fc_stop * Fc_stop_)\n",
" \n",
" ## Root depth upate\n",
" Zr = s_minZr * minZr_ + (FCov / (s_FmaxFC * FmaxFC_)) * s_maxZr * (maxZr_ - minZr_)\n",
" \n",
" # Water capacities\n",
" TAW = (soil_params.FC - soil_params.WP) * Zr\n",
" TDW = (soil_params.FC - soil_params.WP) * (s_Zsoil * Zsoil_ - Zr)\n",
" \n",
" # Update depletions\n",
" Dr = update_Dr(TAW, TDW, Zr, TAW0, TDW0, Dr0, Dd0, Zr0)\n",
" Dd = update_Dd(TAW, TDW, Zr, TAW0, TDW0, Dd0, Zr0)\n",
" \n",
" # Update param p\n",
" p_ = (xr_minimum(xr_maximum(s_p * p_ + 0.04 * (5 - weather_cube.ET0.sel({'time': dates[i-1]}) / 1000), 0.1), 0.8) * (1 / s_p)).round(0).astype('i2')\n",
" \n",
" # Irrigation ==============!!!!!\n",
" Irrig = xr_minimum(xr_maximum(Dr - weather_cube.tp.sel({'time': dates[i]}) / 1000, 0), s_Lame_max * Lame_max_) * Irrig_auto_\n",
" Irrig = xr.where(Dr > TAW * s_p * p_, Irrig, 0)\n",
" \n",
" # Kcb\n",
" Kcb = xr_minimum(s_Kslope * Kslope_ * (ndvi_cube.ndvi.sel({'time': dates[i]}) / 255) + s_Koffset * Koffset_, s_KmaxKcb * KmaxKcb_)\n",
" \n",
" # Update depletions with rainfall and/or irrigation \n",
" ## DP\n",
" model_outputs.DP.loc[{'time': dates[i]}] = -xr_minimum(Dd + xr_minimum(Dr - (weather_cube.tp.sel({'time': dates[i]}) / 1000) - Irrig, 0), 0)\n",
" Dei = xr_minimum(xr_maximum(Dei - (weather_cube.tp.sel({'time': dates[i]}) / 1000) - Irrig / (s_FW * FW_ / 100), 0), TEW)\n",
" Dep = xr_minimum(xr_maximum(Dep - (weather_cube.tp.sel({'time': dates[i]}) / 1000), 0), TEW)\n",
" \n",
" fewi = xr_minimum(1 - FCov, (s_FW * FW_ / 100))\n",
" fewp = 1 - FCov - fewi\n",
" \n",
" De = (Dei * fewi + Dep * fewp) / (fewi + fewp)\n",
" # variables_t1['De'] = xr.where(variables_t1['De'].isfinite(), variables_t1['De'], variables_t1['Dei'] * (s_FW * FW_ / 100) + variables_t1['Dep'] * (1 - (s_FW * FW_ / 100))) #================= find replacement for .isfinite() method !!!!!!!!!\n",
"\n",
" ## Dr\n",
" Dr = xr_minimum(xr_maximum(Dr - (weather_cube.tp.sel({'time': dates[i]}) / 1000) - Irrig, 0), TAW)\n",
" Dd = xr_minimum(xr_maximum(Dd + xr_minimum(Dr - (weather_cube.tp.sel({'time': dates[i]}) / 1000) - Irrig, 0), 0), TDW)\n",
" \n",
" # Diffusion coefficients\n",
" diff_rei = calculate_diff_re(TAW, Dr, Zr, RUE, Dei, FCov, Ze_, DiffE_, scale_factor)\n",
" diff_rep = calculate_diff_re(TAW, Dr, Zr, RUE, Dep, FCov, Ze_, DiffE_, scale_factor)\n",
" diff_dr = calculate_diff_dr(TAW, TDW, Dr, Zr, Dd, FCov, Zsoil_, DiffR_, scale_factor) \n",
" \n",
" # Weighing factor W\n",
" W = calculate_W(TEW, Dei, Dep, fewi, fewp)\n",
" \n",
" # Soil water contents\n",
" model_outputs.SWCe.loc[{'time': dates[i]}] = 1 - De/TEW\n",
" model_outputs.SWCr.loc[{'time': dates[i]}] = 1 - Dr/TAW\n",
" \n",
" # Water Stress coefficient\n",
" Ks = xr_minimum((TAW - Dr) / (TAW * (1 - s_p * p_)), 1)\n",
" \n",
" # Reduction coefficient for evaporation\n",
" Kei = xr_minimum(W * calculate_Kr(TEW, Dei, REW_, scale_factor) * (s_Kcmax * Kcmax_ - Kcb), fewi * s_Kcmax * Kcmax_)\n",
" Kep = xr_minimum((1 - W) * calculate_Kr(TEW, Dei, REW_, scale_factor) * (s_Kcmax * Kcmax_ - Kcb), fewp * s_Kcmax * Kcmax_)\n",
" \n",
" # Prepare coefficients for evapotranspiration\n",
" Kti = xr_minimum(((s_Ze * Ze_ / Zr)**6) * (1 - Dei / TEW) / xr_maximum(1 - Dr / TAW, 0.001), 1)\n",
" Ktp = xr_minimum(((s_Ze * Ze_ / Zr)**6) * (1 - Dep / TEW) / xr_maximum(1 - Dr / TAW, 0.001), 1)\n",
" Tei = Kti * Ks * Kcb * weather_cube.ET0.sel({'time': dates[i]}) / 1000\n",
" Tep = Ktp * Ks * Kcb * weather_cube.ET0.sel({'time': dates[i]}) / 1000\n",
" Dei = xr.where(fewi > 0, xr_minimum(xr_maximum(Dei + (weather_cube.ET0.sel({'time': dates[i]}) / 1000) * Kei / fewi + Tei - diff_rei, 0), TEW), xr_minimum(xr_maximum(Dei + Tei - diff_rei, 0), TEW))\n",
" Dep = xr.where(fewp > 0, xr_minimum(xr_maximum(Dep + (weather_cube.ET0.sel({'time': dates[i]}) / 1000) * Kep / fewp + Tep - diff_rep, 0), TEW), xr_minimum(xr_maximum(Dep + Tep - diff_rep, 0), TEW))\n",
" \n",
" De = (Dei * fewi + Dep * fewp) / (fewi + fewp)\n",
" # De = xr.where(De.isfinite(), De, Dei * (s_FW * FW_ / 100) + Dep * (1 - (s_FW * FW_ / 100))) #================= find replacement for .isfinite() method !!!!!!!!!\n",
" \n",
" # Evaporation\n",
" model_outputs.E.loc[{'time': dates[i]}] = xr_maximum((Kei + Kep) * weather_cube.ET0.sel({'time': dates[i]}) / 1000, 0)\n",
" model_outputs.Tr.loc[{'time': dates[i]}] = Kcb * Ks * weather_cube.ET0.sel({'time': dates[i]}) / 1000\n",
" \n",
" # Potential evapotranspiration and evaporative fraction ??\n",
" \n",
" # Update depletions (root and deep zones) at the end of the day\n",
" Dr = xr_minimum(xr_maximum(Dr + model_outputs.E.loc[{'time': dates[i]}] + model_outputs.Tr.loc[{'time': dates[i]}] - diff_dr, 0), TAW)\n",
" Dd = xr_minimum(xr_maximum(Dd + diff_dr, 0), TDW)\n",
" \n",
" # Write outputs\n",
" model_outputs.Irr.loc[{'time': dates[i]}] = Irrig\n",
" \n",
" # Update variable_t1 values\n",
" for variable in calculation_variables_t1:\n",
" variables_t1[variable] = variables_t2[variable].copy(deep = True)\n",
" \n",
" print('day ', i+1, '/', len(dates), ' ', end = '\\r')\n",
" \n",
" # Scale the model_outputs variable to save in int16 format\n",
" model_outputs = model_outputs * 1000\n",
" \n",
" # Write encoding dict\n",
" encoding_dict = {}\n",
" for variable in list(model_outputs.keys()):\n",
" encod = {}\n",
" encod['dtype'] = 'i2'\n",
" encoding_dict[variable] = encod\n",
" \n",
" # Save model outputs to netcdf\n",
" model_outputs.to_netcdf(save_path, encoding = encoding_dict)\n",
" \n",
" return None"
]
},
{
"cell_type": "code",
"source": [
"client = Client()\n",
"client"
]
},
{
"cell_type": "code",
"source": [
"data_path = '/mnt/e/DATA/DEV_inputs_test'\n",
"\n",
"ndvi_path = data_path + os.sep + 'ndvi.nc'\n",
"weather_path = data_path + os.sep + 'weather.nc'\n",
"land_cover_path = data_path + os.sep + 'land_cover.nc'\n",
"json_config_file = '/home/auclairj/GIT/modspa-pixel/config/config_modspa.json'\n",
"param_file = '/home/auclairj/GIT/modspa-pixel/parameters/csv_files/params_samir_test.csv'\n",
"soil_path = data_path + os.sep + 'soil.nc'\n",
"save_path = data_path + os.sep + 'outputs.nc'\n",
"\n",
"chunk_size = {'x': -1, 'y': -1, 'time': -1}\n",
"\n",
"\n",
"run_samir(json_config_file, param_file, ndvi_path, weather_path, soil_path, land_cover_path, chunk_size, save_path)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"client.close()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "modspa_pixel",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.9"
},
"orig_nbformat": 4
},
"nbformat": 4,
"nbformat_minor": 2
}