-
Notifications
You must be signed in to change notification settings - Fork 1
/
gridded_dataset_intercomparison.py
6547 lines (5498 loc) · 337 KB
/
gridded_dataset_intercomparison.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
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
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
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Thu Apr 25 09:25:08 2024
@author: jgiles
Script for comparing precipitation datasets at different scales.
"""
import matplotlib
import matplotlib.pyplot as plt
import cartopy
import cartopy.crs as ccrs
import cartopy.feature
from cartopy.mpl.ticker import LongitudeFormatter, LatitudeFormatter
import numpy as np
from numpy import ma
import pandas as pd
import xarray as xr
from datetime import datetime
import matplotlib.colors as mcolors
import os
from matplotlib import gridspec
import scipy.stats
from scipy import signal
import gc
from dask.diagnostics import ProgressBar
import cordex as cx
import wradlib as wrl
import glob
import regionmask as rm
from cdo import Cdo
import xesmf as xe
import hvplot.xarray
import holoviews as hv
import copy
import time
import warnings
#ignore by message
# warnings.filterwarnings("ignore", message="Default reduction dimension will be changed to the grouped dimension")
# warnings.filterwarnings("ignore", message="More than 20 figures have been opened")
# warnings.filterwarnings("ignore", message="Setting the 'color' property will override the edgecolor or facecolor properties.")
# warnings.filterwarnings("ignore", category=FutureWarning)
try:
os.chdir('/home/jgiles/')
except FileNotFoundError:
None
try:
from Scripts.python.radar_processing_scripts import utils
from Scripts.python.radar_processing_scripts import radarmet
# from Scripts.python.radar_processing_scripts import colormap_generator
except ModuleNotFoundError:
import utils
import radarmet
# import colormap_generator
proj = ccrs.PlateCarree(central_longitude=0.0)
proj = ccrs.Mercator(central_longitude=0.0)
# Set rotated pole
# Euro-CORDEX rotated pole coordinates RotPole (198.0; 39.25)
rp = ccrs.RotatedPole(pole_longitude=198.0,
pole_latitude=39.25,
globe=ccrs.Globe(semimajor_axis=6370000,
semiminor_axis=6370000))
# Define a function to reduce the name of the datasets to the minimum
def reduce_dsname(dsname):
if "_" in dsname:
dsname0 = dsname.split("_")[0]
else:
dsname0 = dsname
reduced_names=["IMERG-V07B", "IMERG-V06B", "CMORPH", "TSMP-old", "TSMP-DETECT", "ERA5",
"RADKLIM", "RADOLAN", "EURADCLIM", "GPCC", "GPROF", "HYRAS", "E-OBS",
"CPC", "GSMaP"]
for reddsname in reduced_names:
if reddsname in dsname0:
return reddsname
warnmsg = ("No reduced name found for "+dsname+". Returning the same input name")
warnings.warn(warnmsg)
return dsname
colors = {
"IMERG-V07B-monthly": "#FF6347", "IMERG-V07B-30min": "#FF6347", "IMERG-V07B": "#FF6347", # Tomato
"IMERG-V06B-monthly": "crimson", "IMERG-V06B-30min": "crimson", "IMERG-V06B": "crimson", # crimson
"CMORPH-daily": "#A52A2A", "CMORPH": "#A52A2A", # Brown
"TSMP-old": "#4682B4", # SteelBlue
"TSMP-DETECT-Baseline": "#1E90FF", "TSMP-DETECT": "#1E90FF", # DodgerBlue
"ERA5-monthly": "#8A2BE2", "ERA5-hourly": "#8A2BE2", "ERA5": "#8A2BE2", # BlueViolet
"RADKLIM": "#006400", # DarkGreen
"RADOLAN": "#228B22", # ForestGreen
"EURADCLIM": "#32CD32", # LimeGreen
"GPCC-monthly": "black", "GPCC": "black", # Black
"GPROF": "#FF1493", # DeepPink
"GSMaP": "#FF7BA6", # PalePink
"HYRAS": "#FFD700", # Gold
"E-OBS": "#FFA500", # Orange
"CPC": "#FF8C00", # DarkOrange
}
def hex_to_rgba(hex_color, alpha=0.5):
# Convert HEX to RGB
rgb = mcolors.hex2color(hex_color)
# Add the alpha (transparency) value
rgba = rgb + (alpha,)
return rgba
#%% YEARLY analysis
#%%% Load yearly datasets
loadpath_yearly = "/automount/agradar/jgiles/gridded_data/yearly/"
paths_yearly = {
"IMERG-V07B-monthly": loadpath_yearly+"IMERG-V07B-monthly/IMERG-V07B-monthly_precipitation_yearlysum_2000-2023.nc",
"IMERG-V06B-monthly": loadpath_yearly+"IMERG-V06B-monthly/IMERG-V06B-monthly_precipitation_yearlysum_2000-2021.nc",
# "IMERG-V07B-30min": loadpath_yearly+,
# "IMERG-V06B-30min": loadpath_yearly+,
"CMORPH-daily": loadpath_yearly+"CMORPH-daily/CMORPH-daily_precipitation_yearlysum_1998-2023.nc",
"TSMP-old": loadpath_yearly+"TSMP-old/TSMP-old_precipitation_yearlysum_2000-2021.nc",
"TSMP-DETECT-Baseline": loadpath_yearly+"TSMP-DETECT-Baseline/TSMP-DETECT-Baseline_precipitation_yearlysum_2000-2022.nc",
"ERA5-monthly": loadpath_yearly+"ERA5-monthly/ERA5-monthly_precipitation_yearlysum_1979-2020.nc",
# "ERA5-hourly": loadpath_yearly+,
"RADKLIM": loadpath_yearly+"RADKLIM/RADKLIM_precipitation_yearlysum_2001-2022.nc",
"RADOLAN": loadpath_yearly+"RADOLAN/RADOLAN_precipitation_yearlysum_2006-2022.nc",
"EURADCLIM": loadpath_yearly+"EURADCLIM/EURADCLIM_precipitation_yearlysum_2013-2020.nc",
"GPCC-monthly": loadpath_yearly+"GPCC-monthly/GPCC-monthly_precipitation_yearlysum_1991-2020.nc",
# "GPCC-daily": loadpath_yearly+"GPCC-daily/GPCC-daily_precipitation_yearlysum_2000-2020.nc",
"GPROF": loadpath_yearly+"GPROF/GPROF_precipitation_yearlysum_2014-2023.nc",
"HYRAS": loadpath_yearly+"HYRAS/HYRAS_precipitation_yearlysum_1931-2020.nc",
"E-OBS": loadpath_yearly+"E-OBS/E-OBS_precipitation_yearlysum_1950-2023.nc",
"CPC": loadpath_yearly+"CPC/CPC_precipitation_yearlysum_1979-2024.nc",
"GSMaP": loadpath_yearly+"GSMaP/GSMaP_precipitation_yearlysum_*.nc",
}
data_yearlysum = {}
# reload the datasets
print("Loading yearly datasets ...")
for dsname in paths_yearly.keys():
print("... "+dsname)
if "*" in paths_yearly[dsname]:
data_yearlysum[dsname] = xr.open_mfdataset(paths_yearly[dsname])
else:
data_yearlysum[dsname] = xr.open_dataset(paths_yearly[dsname])
# Special tweaks
# RADOLAN GRID AND CRS
if "RADOLAN" in data_yearlysum.keys():
lonlat_radolan = wrl.georef.rect.get_radolan_grid(900,900, wgs84=True) # these are the left lower edges of each bin
data_yearlysum["RADOLAN"] = data_yearlysum["RADOLAN"].assign_coords({"lon":(("y", "x"), lonlat_radolan[:,:,0]), "lat":(("y", "x"), lonlat_radolan[:,:,1])})
data_yearlysum["RADOLAN"] = data_yearlysum["RADOLAN"].assign(crs=data_yearlysum['RADKLIM'].crs[0])
data_yearlysum["RADOLAN"].attrs["grid_mapping"] = "crs"
data_yearlysum["RADOLAN"].lon.attrs = data_yearlysum["RADKLIM"].lon.attrs
data_yearlysum["RADOLAN"].lat.attrs = data_yearlysum["RADKLIM"].lat.attrs
# EURADCLIM coords
if "EURADCLIM" in data_yearlysum.keys():
data_yearlysum["EURADCLIM"] = data_yearlysum["EURADCLIM"].set_coords(("lon", "lat"))
# Shift HYRAS and EURADCLIM timeaxis
if "EURADCLIM" in data_yearlysum.keys():
data_yearlysum["EURADCLIM"] = data_yearlysum["EURADCLIM"].resample({"time":"YS"}).first()
if "HYRAS" in data_yearlysum.keys():
data_yearlysum["HYRAS"] = data_yearlysum["HYRAS"].resample({"time":"YS"}).mean()
# Convert all non datetime axes (cf Julian calendars) into datetime
for dsname in paths_yearly.keys():
try:
data_yearlysum[dsname]["time"] = data_yearlysum[dsname].indexes['time'].to_datetimeindex()
except:
pass
# Special selections for incomplete extreme years
# IMERG
if "IMERG-V07B-monthly" in data_yearlysum.keys():
data_yearlysum["IMERG-V07B-monthly"] = data_yearlysum["IMERG-V07B-monthly"].loc[{"time":slice("2001", "2022")}]
if "IMERG-V06B-monthly" in data_yearlysum.keys():
data_yearlysum["IMERG-V06B-monthly"] = data_yearlysum["IMERG-V06B-monthly"].loc[{"time":slice("2001", "2020")}]
# CMORPH
if "CMORPH-daily" in data_yearlysum.keys():
data_yearlysum["CMORPH-daily"] = data_yearlysum["CMORPH-daily"].loc[{"time":slice("1998", "2022")}]
# GPROF
if "GPROF" in data_yearlysum.keys():
data_yearlysum["GPROF"] = data_yearlysum["GPROF"].loc[{"time":slice("2015", "2022")}]
# CPC
if "CPC" in data_yearlysum.keys():
data_yearlysum["CPC"] = data_yearlysum["CPC"].loc[{"time":slice("1979", "2023")}]
var_names = ["TOT_PREC", "precipitation", "pr", "surfacePrecipitation", "precip", "Precip",
"RW", "RR", "tp", "cmorph", "rr", "hourlyPrecipRateGC"]
dsignore = [] #['CMORPH-daily', 'RADKLIM', 'RADOLAN', 'EURADCLIM', 'GPROF', 'HYRAS', "IMERG-V06B-monthly", "ERA5-monthly"] # datasets to ignore in the plotting
dsref = ["GPCC-monthly"] # dataset to take as reference (black and bold curve)
#%%% Regional averages
#%%%% Calculate area means (regional averages)
data_to_avg = data_yearlysum # select which data to average (yearly, monthly, daily...)
region =["Portugal", "Spain", "France", "United Kingdom", "Ireland",
"Belgium", "Netherlands", "Luxembourg", "Germany", "Switzerland",
"Austria", "Poland", "Denmark", "Slovenia", "Liechtenstein", "Andorra",
"Monaco", "Czechia", "Slovakia", "Hungary", "Romania"]#"land"
region = "Germany"
region_name = "Germany" # "Europe_EURADCLIM" # name for plots
mask = utils.get_regionmask(region)
TSMP_nudge_margin = 13 # number of gridpoints to mask out the relaxation zone at the margins
# TSMP-case: we make a specific mask to cut out the edge of the european domain + country
dsname = "TSMP-DETECT-Baseline"
mask_TSMP_nudge = False
if dsname in data_to_avg.keys():
mask_TSMP_nudge = True # This will be used later as a trigger for this extra mask
lon_bot = data_to_avg[dsname].lon[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][0].lon.values
lat_bot = data_to_avg[dsname].lat[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][0].lat.values
lon_top = data_to_avg[dsname].lon[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][-1].lon.values
lat_top = data_to_avg[dsname].lat[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][-1].lat.values
lon_right = data_to_avg[dsname].lon[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][:,-1].lon.values
lat_right = data_to_avg[dsname].lat[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][:,-1].lat.values
lon_left = data_to_avg[dsname].lon[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][:,0].lon.values
lat_left = data_to_avg[dsname].lat[TSMP_nudge_margin:-TSMP_nudge_margin,TSMP_nudge_margin:-TSMP_nudge_margin][:,0].lat.values
lon_tsmp_edge = np.concatenate((lon_bot, lon_right, lon_top[::-1], lon_left[::-1]))
lat_tsmp_edge = np.concatenate((lat_bot, lat_right, lat_top[::-1], lat_left[::-1]))
lonlat_tsmp_edge = list(zip(lon_tsmp_edge, lat_tsmp_edge))
TSMP_no_nudge = rm.Regions([ lonlat_tsmp_edge ], names=["TSMP_no_nudge"], abbrevs=["TSMP_NE"], name="TSMP")
# I did not find a way to directly combine this custom region with a predefined country region. I will
# have to just apply the masks consecutively
data_avgreg = {}
# Means over region
print("Calculating means over "+region_name)
to_add = {} # dictionary to add rotated versions
for dsname in data_to_avg.keys():
print("... "+dsname)
if dsname in ["RADOLAN", "RADKLIM", "HYRAS", "EURADCLIM"]:
# these datasets come in equal-pixel-sized grids, so we only need to apply the average over the region
mask0 = mask.mask(data_to_avg[dsname])
if mask_TSMP_nudge: mask0 = TSMP_no_nudge.mask(data_to_avg[dsname]).where(mask0.notnull())
data_avgreg[dsname] = data_to_avg[dsname].where(mask0.notnull()).mean(("x", "y")).compute()
if dsname in ["IMERG-V07B-monthly", "IMERG-V06B-monthly", "CMORPH-daily", "ERA5-monthly",
"GPCC-monthly", "GPCC-daily", "GPROF", "E-OBS", "CPC", "GSMaP"]:
# these datasets come in regular lat-lon grids, so we need to average over the region considering the area weights
variables_to_include = [vv for vv in data_to_avg[dsname].data_vars \
if "lonv" not in data_to_avg[dsname][vv].dims \
if "latv" not in data_to_avg[dsname][vv].dims \
if "nv" not in data_to_avg[dsname][vv].dims]
mask0 = mask.mask(data_to_avg[dsname])
if mask_TSMP_nudge: mask0 = TSMP_no_nudge.mask(data_to_avg[dsname]).where(mask0.notnull())
if dsname in ["ERA5-monthly", "E-OBS"]:
data_avgreg[dsname] = utils.calc_spatial_mean(data_to_avg[dsname][variables_to_include].where(mask0.notnull()),
lon_name="longitude", lat_name="latitude").compute()
else:
data_avgreg[dsname] = utils.calc_spatial_mean(data_to_avg[dsname][variables_to_include].where(mask0.notnull()),
lon_name="lon", lat_name="lat").compute()
if dsname in ["TSMP-DETECT-Baseline", "TSMP-old"]:
# we need to unrotate the TSMP grid and then average over the region considering the area weights
grid_out = xe.util.cf_grid_2d(-49.75,70.65,0.1,19.85,74.65,0.1) # manually recreate the EURregLonLat01deg grid
regridder = xe.Regridder(data_to_avg[dsname].cf.add_bounds(["lon", "lat"]), grid_out, "conservative")
to_add[dsname+"-EURregLonLat01deg"] = regridder(data_to_avg[dsname])
mask0 = mask.mask(to_add[dsname+"-EURregLonLat01deg"])
if mask_TSMP_nudge: mask0 = TSMP_no_nudge.mask(to_add[dsname+"-EURregLonLat01deg"]).where(mask0.notnull())
data_avgreg[dsname] = utils.calc_spatial_mean(to_add[dsname+"-EURregLonLat01deg"].where(mask0.notnull()),
lon_name="lon", lat_name="lat").compute()
# add the rotated datasets to the original dictionary
data_to_avg = {**data_to_avg, **to_add}
data_yearlysum = data_to_avg.copy()
#%%%% Simple map plot
dsname = "GPCC-monthly"
vname = "precip"
mask = utils.get_regionmask(region)
mask0 = mask.mask(data_yearlysum[dsname])
dropna = True
if mask_TSMP_nudge:
mask0 = TSMP_no_nudge.mask(data_yearlysum[dsname]).where(mask0.notnull())
dropna=False
f, ax1 = plt.subplots(1, 1, figsize=(8, 4), subplot_kw=dict(projection=proj))
plot = data_yearlysum[dsname][vname][0].where(mask0.notnull(), drop=dropna).plot(x="lon", y="lat", cmap="Blues", vmin=0, vmax=1000,
subplot_kws={"projection":proj}, transform=ccrs.PlateCarree(),
cbar_kwargs={'label': "mm", 'shrink':0.88})
if mask_TSMP_nudge: ax1.set_extent([-43.4, 63.65, 22.6, 71.15], crs=ccrs.PlateCarree())
plot.axes.coastlines(alpha=0.7)
plot.axes.gridlines(draw_labels={"bottom": "x", "left": "y"}, visible=False)
plot.axes.add_feature(cartopy.feature.BORDERS, linestyle='-', linewidth=1, alpha=0.4) #countries
plt.title(dsname)
#%%%% Simple map plot (for number of stations per gridcell)
dsname = "GPCC-monthly"
vname = "numgauge"
savepath = "/automount/agradar/jgiles/images/gridded_datasets_intercomparison/maps/Turkey/"
for yy in np.arange(2000,2021):
ysel = str(yy)
mask = utils.get_regionmask(region)
mask0 = mask.mask(data_yearlysum[dsname])
dropna = True
if mask_TSMP_nudge:
mask0 = TSMP_no_nudge.mask(data_yearlysum[dsname]).where(mask0.notnull())
dropna=False
f, ax1 = plt.subplots(1, 1, figsize=(8, 4), subplot_kw=dict(projection=proj))
cmap1 = copy.copy(plt.cm.Blues)
cmap1.set_under("lightgray")
plot = (data_yearlysum[dsname][vname].sel(time=ysel)/12).where(mask0.notnull(), drop=dropna).plot(x="lon", y="lat",
levels=3, cmap=cmap1, vmin=1, vmax=3,
subplot_kws={"projection":proj}, transform=ccrs.PlateCarree(),
cbar_kwargs={'label': "", 'shrink':0.88})
if mask_TSMP_nudge: plot.axes.set_extent([-43.4, 63.65, 22.6, 71.15], crs=ccrs.PlateCarree())
plot.axes.coastlines(alpha=0.7)
plot.axes.gridlines(draw_labels={"bottom": "x", "left": "y"}, visible=False)
plot.axes.add_feature(cartopy.feature.BORDERS, linestyle='-', linewidth=1, alpha=0.4) #countries
plt.title(dsname+" number of stations per gridcell "+ysel)
# save figure
savepath_yy = savepath+ysel+"/"
if not os.path.exists(savepath_yy):
os.makedirs(savepath_yy)
filename = "numgauge_"+region_name+"_"+dsname+"_"+ysel+".png"
plt.savefig(savepath_yy+filename, bbox_inches="tight")
# plt.show()
#%%%% Interannual variability area-means plot
# make a list with the names of the precipitation variables
var_names = ["TOT_PREC", "precipitation", "pr", "surfacePrecipitation", "precip", "Precip",
"RW", "RR", "tp", "cmorph", "rr", "hourlyPrecipRateGC"]
dsignore = [] #['CMORPH-daily', 'RADKLIM', 'RADOLAN', 'EURADCLIM', 'GPROF', 'HYRAS', "IMERG-V06B-monthly", "ERA5-monthly"] # datasets to ignore in the plotting
dsref = ["GPCC-monthly"] # dataset to take as reference (black and bold curve)
dstoplot = ['IMERG-V07B-monthly', 'IMERG-V06B-monthly', 'CMORPH-daily', 'GPROF', "GSMaP",
'RADKLIM', 'RADOLAN', 'EURADCLIM',
'TSMP-old', 'TSMP-DETECT-Baseline',
'ERA5-monthly',
'GPCC-monthly', 'HYRAS', 'E-OBS', 'CPC']
for dsname in dstoplot:
if dsname in dsignore:
continue
plotted = False
for vv in var_names:
if vv in data_avgreg[dsname].data_vars:
color=None
marker=None
if dsname in dsref:
color = "black"
marker = "o"
else:
color = colors[dsname]
try:
plt.plot(data_avgreg[dsname]['time'], data_avgreg[dsname][vv],
label=reduce_dsname(dsname), c=color, marker=marker)
except TypeError:
# try to change the time coord to datetime format
plt.plot(data_avgreg[dsname].indexes['time'].to_datetimeindex(), data_avgreg[dsname][vv],
label=reduce_dsname(dsname), c=color, marker=marker)
plotted = True
if not plotted:
raise Warning("Nothing plotted for "+dsname)
plt.legend(ncols=3, fontsize=7)
plt.title("Area-mean annual total precip "+region_name+" [mm]")
plt.xlim(datetime(2000,1,1), datetime(2020,1,1))
# plt.xlim(2000, 2020)
plt.grid()
#%%%% Interannual variability area-means plot (interactive html)
var_names = ["TOT_PREC", "precipitation", "pr", "surfacePrecipitation", "precip", "Precip",
"RW", "RR", "tp", "cmorph", "rr", "hourlyPrecipRateGC"]
dsignore = [] # datasets to ignore in the plotting
dsref = ["GPCC-monthly"] # dataset to take as reference (black and bold curve)
hvplots = []
for dsname in data_avgreg.keys():
if dsname in dsignore:
continue
plotted = False
for vv in var_names:
if vv in data_avgreg[dsname].data_vars:
if dsname in dsref:
color = "black"
lw = 4
hvplots.append(
data_avgreg[dsname][vv].hvplot.line(x='time', label=dsname).opts(color=color, line_width=lw, show_legend=True, muted_alpha=0)
)
else:
hvplots.append(
data_avgreg[dsname][vv].hvplot.line(x='time', label=dsname).opts(show_legend=True, muted_alpha=0)
)
plotted = True
if not plotted:
raise Warning("Nothing plotted for "+dsname)
layout = hvplots[0]
for nplot in np.arange(1, len(hvplots)):
layout = layout * hvplots[nplot]
layout.opts(title="Area-mean annual total precip "+region_name+" [mm]", xlabel="Time", show_grid=True, legend_position='right',
height=600, width=1200)
# Save to HTML file
hv.save(layout, "/automount/agradar/jgiles/images/gridded_datasets_intercomparison/interannual/"+region_name+"/lineplots/area_mean_annual_total_precip_"+region_name+".html")
#%%%% Plot the period from each dataset
# make a list with the names of the precipitation variables
var_names = ["TOT_PREC", "precipitation", "pr", "surfacePrecipitation", "precip", "Precip",
"RW", "RR", "tp", "cmorph", "rr", "hourlyPrecipRateGC"]
dsignore = [] #['CMORPH-daily', 'RADKLIM', 'RADOLAN', 'EURADCLIM', 'GPROF', 'HYRAS', "IMERG-V06B-monthly", "ERA5-monthly"] # datasets to ignore in the plotting
dsref = ["GPCC-monthly"] # dataset to take as reference (black and bold curve)
yticks = []
yticklabels = []
for dsn,dsname in enumerate(data_avgreg.keys()):
if dsname in dsignore:
continue
plotted = False
for vv in var_names:
if vv in data_avgreg[dsname].data_vars:
color=None
marker=None
if dsname in dsref:
color = "black"
marker = "o"
else:
color = colors[dsname]
try:
plt.plot(data_avgreg[dsname]['time'], data_avgreg[dsname][vv]*0+dsn+1, label=dsname, c=color, marker=marker)
except TypeError:
# try to change the time coord to datetime format
plt.plot(data_avgreg[dsname].indexes['time'].to_datetimeindex(), data_avgreg[dsname][vv]*0+dsn+1, label=dsname, c=color, marker=marker)
plotted = True
yticks.append(dsn+1)
yticklabels.append(dsname)
if not plotted:
raise Warning("Nothing plotted for "+dsname)
plt.title("Period from each dataset")
plt.gca().set_xticks([str(xx) for xx in np.arange(1980, 2024)], minor=True)
plt.xlim(datetime(1980,1,1), datetime(2024,1,1))
plt.yticks(yticks, yticklabels) # set yticks and labels
plt.grid()
#%%% BIAS and ERRORS
#%%%% Bias (absolute and relative) calculation from regional averages
var_names = ["TOT_PREC", "precipitation", "pr", "surfacePrecipitation", "precip", "Precip",
"RW", "RR", "tp", "cmorph", "rr", "hourlyPrecipRateGC"]
dsignore = [] # datasets to ignore in the plotting
dsref = ["GPCC-monthly"] # dataset to take as reference
data_to_bias = data_avgreg
data_bias = {}
data_bias_relative = {}
for dsname in data_to_bias.keys():
if dsname in dsignore+dsref:
continue
for vv in var_names:
if vv in data_to_bias[dsname].data_vars:
for vvref in var_names:
if vvref in data_to_bias[dsref[0]].data_vars:
data_bias[dsname] = data_to_bias[dsname][vv] - data_to_bias[dsref[0]][vvref]
data_bias_relative[dsname] = (data_to_bias[dsname][vv] - data_to_bias[dsref[0]][vvref])/data_to_bias[dsref[0]][vvref]*100
break
#%%%% Region-averages bias bar plot
# Calculate bar width based on the number of data arrays
bar_width = 0.8 / len(data_bias)
# Get time values
time_values_ref = data_to_bias[dsref[0]]['time']
#%%%%% Plotting each DataArray in the dictionary
plt.figure(figsize=(20, 6)) # Adjust figure size as needed
for idx, (key, value) in enumerate(data_bias.items()):
value_padded = value.broadcast_like(data_to_bias[dsref[0]])
time_values = value_padded['time']
bar_positions = np.arange(len(time_values)) + idx * bar_width
plt.bar(bar_positions, value_padded, width=bar_width, label=key, color=colors[key])
plt.xlabel('Time')
plt.ylabel(value.attrs['units'])
plt.title("Area-mean annual total precip BIAS with respect to "+dsref[0]+" "+region_name)
plt.xticks(np.arange(len(time_values)) + 0.4, time_values_ref.dt.year.values, rotation=45) # Rotate x-axis labels for better readability
plt.legend()
plt.grid(True)
plt.tight_layout()
plt.show()
#%%%%% Plotting each DataArray in the dictionary (same but for relative bias)
plt.figure(figsize=(20, 6)) # Adjust figure size as needed
for idx, (key, value) in enumerate(data_bias_relative.items()):
value_padded = value.broadcast_like(data_to_bias[dsref[0]])
time_values = value_padded['time']
bar_positions = np.arange(len(time_values)) + idx * bar_width
plt.bar(bar_positions, value_padded, width=bar_width, label=key, color=colors[key])
plt.xlabel('Time')
plt.ylabel("%")
plt.title("Area-mean annual total precip RELATIVE BIAS with respect to "+dsref[0]+" "+region_name)
plt.xticks(np.arange(len(time_values)) + 0.4, time_values_ref.dt.year.values, rotation=45) # Rotate x-axis labels for better readability
plt.legend()
plt.grid(True)
plt.tight_layout()
plt.show()
#%%%% Relative bias and errors calculation (at gridpoint level, not with the area means)
# First we need to transform EURADCLIM, RADKLIM, RADOLAN and HYRAS to regular grids
# We use the DETECT 1 km grid for this
to_add = {} # dictionary to add regridded versions
for dsname in ["EURADCLIM", "RADOLAN", "HYRAS", "RADKLIM"]:
if dsname not in data_yearlysum: continue
print("Regridding "+dsname+" ...")
grid_out = xe.util.cf_grid_2d(-49.746,70.655,0.01,19.854,74.654,0.01) # manually recreate the EURregLonLat001deg grid
grid_out = xe.util.grid_2d(-49.746,70.655,0.01,19.854,74.654,0.01) # manually recreate the EURregLonLat001deg grid
# # I tried to use dask for the weight generation to avoid memory crash
# # but I did not manage to make it work: https://xesmf.readthedocs.io/en/latest/notebooks/Dask.html
# grid_out = grid_out.chunk({"x": 50, "y": 50, "x_b": 50, "y_b": 50,})
# # we then try parallel regridding: slower but less memory-intensive (this takes forever)
# regridder = xe.Regridder(data_yearlysum[dsname].cf.add_bounds(["lon", "lat"]),
# grid_out,
# "conservative", parallel=True)
# to_add[dsname+"-EURregLonLat001deg"] = regridder(data_to_avg[dsname])
# regridder.to_netcdf() # we save the weights
# # to reuse the weigths:
# xe.Regridder(data_yearlysum[dsname].cf.add_bounds(["lon", "lat"]),
# grid_out,
# "conservative", parallel=True, weights="/path/to/weights") #!!! Can I use CDO weights here?
# Cdo().gencon()
# cdo gencon,/automount/ags/jgiles/IMERG_V06B/global_monthly/griddes.txt -setgrid,/automount/agradar/jgiles/TSMP/griddes_mod.txt /automount/agradar/jgiles/TSMP/postprocessed/TSMP_TOT_PREC_yearlysum_2001-2020.nc weights_to_IMERG.nc
# Instead, just regrid to the reference dataset grid (this is fast)
regridder = xe.Regridder(data_yearlysum[dsname].cf.add_bounds(["lon", "lat"]), data_yearlysum[dsref[0]], "conservative")
to_add[dsname+"_"+dsref[0]+"-grid"] = regridder(data_yearlysum[dsname])
# add the regridded datasets to the original dictionary
data_yearlysum = {**data_yearlysum, **to_add}
# Compute the biases
dsignore = ["EURADCLIM", "RADOLAN", "HYRAS", "RADKLIM", 'TSMP-old', 'TSMP-DETECT-Baseline'] # datasets to ignore (because we want the regridded version)
data_to_bias = copy.copy(data_yearlysum)
to_add = {} # dictionary to add regridded versions
data_bias_map = {} # maps of yearly biases
data_bias_relative_map = {} # maps of yearly relative biases
data_abs_error_map = {} # maps of yearly absolute errors
data_bias_relative_gp = {} # yearly relative biases on a gridpoint basis (sum of gridpoint biases divided by sum of reference-data values)
data_mean_abs_error_gp = {} # yearly MAE on a gridpoint basis (sum of gridpoint abs errors divided by number of data values)
data_norm_mean_abs_error_gp = {} # yearly NMAE on a gridpoint basis (sum of gridpoint abs errors divided by sum of reference-data values)
for dsname in data_to_bias.keys():
if dsname in dsignore+dsref:
continue
print("Processing "+dsname+" ...")
for vv in var_names:
if vv in data_to_bias[dsname].data_vars:
for vvref in var_names:
if vvref in data_to_bias[dsref[0]].data_vars:
if dsref[0]+"-grid" not in dsname: # if no regridded already, do it now
if "longitude" in data_to_bias[dsname].coords or "latitude" in data_to_bias[dsname].coords:
# if the names of the coords are longitude and latitude, change them to lon, lat
data_to_bias[dsname] = data_to_bias[dsname].rename({"longitude":"lon", "latitude":"lat"})
if dsname in ["IMERG-V07B-monthly", "IMERG-V06B-monthly"]:
# we need to remove the default defined bounds or the regridding will fail
data_to_bias[dsname] = data_to_bias[dsname].drop_vars(["lon_bnds", "lat_bnds"])
del(data_to_bias[dsname].lon.attrs["bounds"])
del(data_to_bias[dsname].lat.attrs["bounds"])
if dsname in ["CMORPH-daily"]:
# we need to remove the default defined bounds or the regridding will fail
data_to_bias[dsname] = data_to_bias[dsname].drop_vars(["lon_bounds", "lat_bounds"])
del(data_to_bias[dsname].lon.attrs["bounds"])
del(data_to_bias[dsname].lat.attrs["bounds"])
if dsname in ["GPROF", "TSMP-old-EURregLonLat01deg", "TSMP-DETECT-Baseline-EURregLonLat01deg"]:
# we need to remove the default defined bounds or the regridding will fail
del(data_to_bias[dsname].lon.attrs["bounds"])
del(data_to_bias[dsname].lat.attrs["bounds"])
# regridder = xe.Regridder(data_to_bias[dsname].cf.add_bounds(["lon", "lat"]), data_to_bias[dsref[0]], "conservative")
regridder = xe.Regridder(data_to_bias[dsname], data_to_bias[dsref[0]], "conservative")
data0 = regridder(data_to_bias[dsname][vv])
to_add[dsname+"_"+dsref[0]+"-grid"] = regridder(data_to_bias[dsname][vv])
else:
data0 = data_to_bias[dsname][vv]
data0 = data0.where(data0>0)
dataref = data_to_bias[dsref[0]][vvref]
mask0 = mask.mask(data_to_bias[dsref[0]])
if mask_TSMP_nudge: mask0 = TSMP_no_nudge.mask(data_to_bias[dsref[0]]).where(mask0.notnull())
data_bias_map[dsname] = ( data0 - dataref ).compute()
data_bias_relative_map[dsname] = ( data_bias_map[dsname] / dataref ).compute() *100
data_bias_map_masked = data_bias_map[dsname].where(mask0.notnull())
data_bias_relative_map_masked = data_bias_relative_map[dsname].where(mask0.notnull())
data_abs_error_map[dsname] = abs(data_bias_map[dsname])
data_abs_error_map_masked = data_abs_error_map[dsname].where(mask0.notnull())
data_bias_relative_gp[dsname] = utils.calc_spatial_integral(data_bias_map_masked,
lon_name="lon", lat_name="lat").compute() / \
utils.calc_spatial_integral(dataref.where(mask0.notnull()),
lon_name="lon", lat_name="lat").compute() *100
data_norm_mean_abs_error_gp[dsname] = utils.calc_spatial_integral(data_abs_error_map_masked,
lon_name="lon", lat_name="lat").compute() / \
utils.calc_spatial_integral(dataref.where(mask0.notnull()),
lon_name="lon", lat_name="lat").compute() *100
data_mean_abs_error_gp[dsname] = utils.calc_spatial_mean(data_abs_error_map_masked,
lon_name="lon", lat_name="lat").compute()
break
# add the regridded datasets to the original dictionary
data_yearlysum = {**data_yearlysum, **to_add}
#%%%% Relative bias and error plots
#%%%%% Simple map plot
# region = "Germany" #"land"
to_plot = data_bias_map
dsname = "TSMP-DETECT-Baseline-EURregLonLat01deg"
title = "BIAS"
yearsel = "2016"
cbarlabel = "mm" # mm
vmin = -250
vmax = 250
lonlat_slice = [slice(-43.4,63.65), slice(22.6, 71.15)]
mask = utils.get_regionmask(region)
mask0 = mask.mask(to_plot[dsname])
dropna = True
if mask_TSMP_nudge:
mask0 = TSMP_no_nudge.mask(to_plot[dsname]).where(mask0.notnull())
dropna=False
f, ax1 = plt.subplots(1, 1, figsize=(8, 4), subplot_kw=dict(projection=proj))
plot = to_plot[dsname].loc[{"time":yearsel}].where(mask0.notnull(), drop=dropna).loc[{"lon":lonlat_slice[0],
"lat":lonlat_slice[1]}].plot(x="lon",
y="lat",
cmap="RdBu_r",
vmin=vmin, vmax=vmax,
subplot_kws={"projection":proj}, transform=ccrs.PlateCarree(),
cbar_kwargs={'label': cbarlabel, 'shrink':0.88})
if mask_TSMP_nudge: plot.axes.set_extent([-43.4, 63.65, 22.6, 71.15], crs=ccrs.PlateCarree())
plot.axes.coastlines(alpha=0.7)
plot.axes.gridlines(draw_labels={"bottom": "x", "left": "y"}, visible=False)
plot.axes.add_feature(cartopy.feature.BORDERS, linestyle='-', linewidth=1, alpha=0.4) #countries
plt.title(title+" "+yearsel+"\n"+dsname+"\n "+region_name+" Ref.: "+dsref[0])
#%%%%% Simple map plot (loop)
# Like previous but for saving all plots
# region = "Germany" #"land"
savepath = "/automount/agradar/jgiles/images/gridded_datasets_intercomparison/maps/annual/"+region_name+"/"
period = np.arange(2000,2024)
to_plot_dict = [
(data_bias_map, "BIAS", "mm", -250, 250),
(data_bias_relative_map, "RELATIVE BIAS", "%", -75, 75),
]
lonlat_slice = [slice(-43.4,63.65), slice(22.6, 71.15)]
for to_plot, title, cbarlabel, vmin, vmax in to_plot_dict:
print("Plotting "+title)
for dsname in to_plot.keys():
print("... "+dsname)
dsname_short = dsname.split("_")[0]
mask = utils.get_regionmask(region)
mask0 = mask.mask(to_plot[dsname])
dropna = True
if mask_TSMP_nudge:
mask0 = TSMP_no_nudge.mask(to_plot[dsname]).where(mask0.notnull())
dropna=False
for yearsel in period:
try:
plt.close()
# f, ax1 = plt.subplots(1, 1, figsize=(8, 4), subplot_kw=dict(projection=proj))
plot = to_plot[dsname].loc[{"time":str(yearsel)}].where(mask0.notnull(),
drop=dropna).loc[{"lon":lonlat_slice[0],
"lat":lonlat_slice[1]}].plot(x="lon",
y="lat",
cmap="RdBu_r",
vmin=vmin, vmax=vmax,
subplot_kws={"projection":proj}, transform=ccrs.PlateCarree(),
cbar_kwargs={'label': cbarlabel, 'shrink':0.88})
if mask_TSMP_nudge: plot.axes.set_extent([-43.4, 63.65, 22.6, 71.15], crs=ccrs.PlateCarree())
plot.axes.coastlines(alpha=0.7)
plot.axes.gridlines(draw_labels={"bottom": "x", "left": "y"}, visible=False)
plot.axes.add_feature(cartopy.feature.BORDERS, linestyle='-', linewidth=1, alpha=0.4) #countries
plt.title(title+" "+str(yearsel)+"\n"+dsname_short+"\n "+region_name+" Ref.: "+dsref[0])
# save figure
savepath_yy = savepath+str(yearsel)+"/"
if not os.path.exists(savepath_yy):
os.makedirs(savepath_yy)
filename = "_".join([title.lower().replace(" ","_"), region_name, dsname_short,dsref[0],str(yearsel)])+".png"
plt.savefig(savepath_yy+filename, bbox_inches="tight")
plt.close()
except KeyError:
continue
#%%%%% Box plots of BIAS and ERRORS
# the box plots are made up of the yearly bias or error values, and the datasets are ordered according to their median
to_plot0 = data_bias_relative_gp.copy() # data_mean_abs_error_gp # data_bias_relative_gp # data_norm_mean_abs_error_gp
savepath = "/automount/agradar/jgiles/images/gridded_datasets_intercomparison/interannual/"+region_name+"/boxplots/relative_bias/"
savefilename = "boxplot_relative_bias_yearly"
title = "Relative bias (yearly values) "+region_name+". Ref.: "+dsref[0]
ylabel = "%" # % # mm
dsignore = [] # ['CMORPH-daily', 'GPROF', 'HYRAS_GPCC-monthly-grid', "E-OBS", "CPC"] #['CMORPH-daily', 'RADKLIM', 'RADOLAN', 'EURADCLIM', 'GPROF', 'HYRAS', "IMERG-V06B-monthly", "ERA5-monthly"] # datasets to ignore in the plotting
tsel = [
slice(None, None), # if I want to consider only certain period. Otherwise set to (None, None). Multiple options possible
slice("2001-01-01", "2020-01-01"),
slice("2006-01-01", "2020-01-01"),
slice("2013-01-01", "2020-01-01"),
slice("2015-01-01", "2020-01-01")
]
ignore_incomplete = True # flag for ignoring datasets that do not cover the complete period. Only works for specific periods (not for slice(None, None))
for tseln in tsel:
to_plot = to_plot0.copy()
savefilenamen = copy.deepcopy(savefilename)
titlen = copy.deepcopy(title)
if tseln.start is not None and tseln.stop is not None: # add specific period to title
titlen = titlen+". "+tseln.start+" - "+tseln.stop
savefilenamen = savefilenamen+"_"+tseln.start+"-"+tseln.stop
if ignore_incomplete:
for key in to_plot.copy().keys():
if not (to_plot[key].time[0].dt.date <= datetime.strptime(tseln.start, "%Y-%m-%d").date() and
to_plot[key].time[-1].dt.date >= datetime.strptime(tseln.stop, "%Y-%m-%d").date()):
del(to_plot[key])
# Initialize a figure and axis
plt.figure(figsize=(1.25*(len(to_plot.keys())-len(dsignore)), 6))
ax = plt.subplot(111)
# Create a list to hold the data arrays
plotted_arrays = []
plotted_arrays_lengths = []
# Order according to median:
to_plot = dict(sorted(to_plot.items(), key=lambda item: item[1].sel(time=tseln).median()))
# Iterate over the datasets in the dictionary
for key, value in to_plot.items():
if key not in dsignore:
# Plot a box plot for each dataset
value = value.sel(time=tseln)
plotted_arrays.append(value.values) # values of each box
plotted_arrays_lengths.append(len(value)) # number of values in each box
box = ax.boxplot(value.values, positions=[len(plotted_arrays)], widths=0.6,
patch_artist=True,
boxprops=dict(facecolor=hex_to_rgba(colors[reduce_dsname(key)], 0.2),
edgecolor=colors[reduce_dsname(key)], lw = 2), #'#b6d6e3'
medianprops=dict(color=colors[reduce_dsname(key)], lw=2.5), #"#20788e"
whiskerprops=dict(color=colors[reduce_dsname(key)], lw=2),
capprops=dict(color=colors[reduce_dsname(key)], lw=2),
flierprops=dict(markeredgecolor=colors[reduce_dsname(key)], lw=2)
)
# Set x-axis ticks and labels with dataset names
ax.set_xticks(range(1, len(plotted_arrays) + 1))
ax.set_xticklabels([reduce_dsname(dsname)
for dsname in
[ds for ds in to_plot.keys() if ds not in dsignore]
],
rotation=45, fontsize=15)
ax.xaxis.label.set_size(15) # change xlabel size
ax.yaxis.label.set_size(15) # change ylabel size
ax.tick_params(axis='x', labelsize=15) # change xtick label size
ax.tick_params(axis='y', labelsize=15) # change xtick label size
# dsnames=[dsname.split("_")[0] if "_" in dsname
# else "-".join(dsname.split("-")[:-1]) if "EURreg" in dsname
# else dsname
# for dsname in
# [ds for ds in to_plot.keys() if ds not in dsignore]
# ]
# for xtick, color in zip(ax.get_xticklabels(), [colors[dsname] for dsname in dsnames]):
# xtick.set_color(color)
# Make a secondary x axis to display the number of values in each box
ax2 = ax.secondary_xaxis('top')
ax2.xaxis.set_ticks_position("bottom")
ax2.xaxis.set_label_position("top")
ax2.set_xticks(range(1, len(plotted_arrays) + 1))
ax2.set_xticklabels(plotted_arrays_lengths)
ax2.set_xlabel('Number of years', fontsize= 15)
# Set labels and title
#ax.set_xlabel('')
ax.set_ylabel(ylabel)
ax.set_title(titlen, fontsize=20)
# plot a reference line at zero
plt.hlines(y=0, xmin=0, xmax=len(plotted_arrays)+1, colors='black', lw=2, zorder=0)
plt.xlim(0.5, len(plotted_arrays) + 0.5)
# Show the plot
plt.grid(True)
plt.tight_layout()
if not os.path.exists(savepath):
os.makedirs(savepath)
plt.savefig(savepath+savefilenamen+".png", bbox_inches="tight")
plt.show()
#%%%%% Taylor diagram
# The Taylor diagram can be done by computing the stats over all gridpoints and all timesteps (spatiotemporal)
# or only doing the stats over space or time separately (for these, either temporal or spatial averages must be done first)
#%%%%%% Compute stats
import skill_metrics as sm
# https://github.com/PeterRochford/SkillMetrics/blob/master/Examples/taylor10.py#L123
# I cannot use skill_metrics to calculate the stats because they do not filter out
# nan values (because of the masks) so the result is erroneous. They also do not handle weighted arrays.
mode = "" # if "spatial" then average in time and compute the diagram in space. Viceversa for "temporal"
dsref = ["GPCC-monthly"]
data_to_stat = data_yearlysum
# choose common period (only datasets that cover the whole period are included)
tslice = slice("2015","2020") # this covers all
# tslice = slice("2013","2020") # this excludes GPROF
# tslice = slice("2006","2020") # this excludes GPROF and EURADCLIM
# tslice = slice("2001","2020") # this excludes GPROF, EURADCLIM and RADOLAN
ccoef = dict()
crmsd = dict()
sdev = dict()
for vv in var_names: # get the name of the desired variable in the reference dataset
if vv in data_to_stat[dsref[0]]:
ref_var_name = vv
break
# Get reference dataset
ds_ref = data_to_stat[dsref[0]][ref_var_name]
# Get area weights
try:
weights = xr.DataArray(utils.grid_cell_areas(ds_ref.lon.values, ds_ref.lat.values),
coords=ds_ref.to_dataset()[["lat","lon"]].coords)
except AttributeError:
weights = xr.DataArray(utils.grid_cell_areas(ds_ref.lon.values, ds_ref.lat.values),
coords=ds_ref.to_dataset()[["latitude","longitude"]].coords)
# Get mask
mask = utils.get_regionmask(region)
mask_ref = mask.mask(ds_ref)
if mask_TSMP_nudge: mask_ref = TSMP_no_nudge.mask(ds_ref).where(mask_ref.notnull())
ds_ref = ds_ref.where(mask_ref.notnull())#.mean(tuple([cn for cn in ds_ref.coords if cn!="time"]))
# Normalize weights in the mask
weights = weights.where(mask_ref.notnull(), other=0.)/weights.where(mask_ref.notnull(), other=0.).sum()
for dsname in data_to_stat.keys(): # compute the stats
if dsref[0]+"-grid" in dsname or dsname==dsref[0]:
# get dataset
if type(data_to_stat[dsname]) is xr.DataArray:
ds_n = data_to_stat[dsname].where(mask_ref.notnull())
else:
for vv in var_names:
if vv in data_to_stat[dsname]:
ds_n = data_to_stat[dsname][vv].where(mask_ref.notnull())
break
# Subset period
tslice_array = ds_ref.sel(time=tslice).time
ds_ref_tsel = ds_ref.sel(time=tslice_array)
try:
ds_n_tsel = ds_n.sel(time=tslice_array)
except KeyError:
print(dsname+" ignored because it does not cover the selected time period")
continue
# Reduce in case mode is "spatial" or "temporal"
if mode=="spatial":
ds_ref_tsel = ds_ref_tsel.mean("time")
ds_n_tsel = ds_n_tsel.mean("time")
mode_name="Spatial"
elif mode=="temporal":
ds_ref_tsel = ds_ref_tsel.weighted(weights).mean([cn for cn in ds_ref_tsel.dims if cn!="time"])
ds_n_tsel = ds_n_tsel.weighted(weights).mean([cn for cn in ds_n_tsel.dims if cn!="time"])
mode_name="Temporal"
else:
mode_name="Spatiotemporal"
if mode=="temporal":
# Get Correlation Coefficient (ccoef)
ccoef[dsname] = xr.corr(ds_n_tsel, ds_ref_tsel).compute()
# Get Centered Root-Mean-Square-Deviation (CRMSD)
crmsd_0 = ( (ds_n_tsel - ds_n_tsel.mean() ) -
(ds_ref_tsel - ds_ref_tsel.mean()) )**2
crmsd_1 = crmsd_0.sum()/xr.ones_like(crmsd_0).where(crmsd_0.notnull()).sum()
crmsd[dsname] = np.sqrt(crmsd_1)
# Get Standard Deviation (SDEV)
sdev[dsname] = ds_n_tsel.std()
else:
# Get Correlation Coefficient (ccoef)
# could work like this but I have to update xarray to include the weights
# ccoef[dsname] = xr.corr(ds_n_tsel, ds_ref_tsel, weigths=weights )
ccoef[dsname] = xr.corr(ds_n_tsel*weights, ds_ref_tsel*weights).compute()
# Get Centered Root-Mean-Square-Deviation (CRMSD)
crmsd_0 = ( (ds_n_tsel - ds_n_tsel.mean() ) -
(ds_ref_tsel - ds_ref_tsel.mean()) )**2
crmsd_1 = crmsd_0.weighted(weights).sum()/xr.ones_like(crmsd_0).where(crmsd_0.notnull()).sum()
crmsd[dsname] = np.sqrt(crmsd_1)
# Get Standard Deviation (SDEV)
sdev[dsname] = ds_n_tsel.weighted(weights).std()
#%%%%%% Plot the diagram
'''
Specify individual marker label (key), label color, symbol, size, symbol face color,
symbol edge color
'''
# Define colors for each group
color_gauges = "k"
color_radar = "r"
color_satellite = "b"
color_reanalysis = "m"
color_model = "c"
# Define marker size
markersize = 7
MARKERS = {
"GPCC-monthly": {
"labelColor": "k",
"symbol": "+",
"size": markersize,
"faceColor": color_gauges,
"edgeColor": color_gauges,
},
"HYRAS": {
"labelColor": "k",
"symbol": "o",
"size": markersize,
"faceColor": color_gauges,
"edgeColor": color_gauges,
},
"E-OBS": {
"labelColor": "k",
"symbol": "D",
"size": markersize,
"faceColor": color_gauges,
"edgeColor": color_gauges,
},
"CPC": {
"labelColor": "k",
"symbol": "X",
"size": markersize,
"faceColor": color_gauges,
"edgeColor": color_gauges,
},
"EURADCLIM": {
"labelColor": "k",
"symbol": "^",
"size": markersize,
"faceColor": color_radar,
"edgeColor": color_radar,
},
"RADOLAN": {