-
Notifications
You must be signed in to change notification settings - Fork 0
/
SPECS_forecast_v4.py
891 lines (735 loc) · 45 KB
/
SPECS_forecast_v4.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
# -*- coding: utf-8 -*-
#!/usr/bin/env python
from __future__ import unicode_literals
import os, sys, glob, re, pickle, time
import numpy as np
import numpy.ma as ma
import scipy
import matplotlib
import matplotlib.pyplot as plt
from mpl_toolkits.basemap import Basemap, shiftgrid
from netCDF4 import Dataset
from scipy import stats
from scipy.stats import linregress,pearsonr
from sklearn import linear_model
from sklearn.preprocessing import Imputer
import urllib2
import zipfile
from SPECS_forecast_v2_tools import *
from cdo import *
cdo = Cdo()
from pyresample import geometry,image, kd_tree
import datetime
import warnings
warnings.filterwarnings("ignore", category=RuntimeWarning)
# TODO
# Check climatology period for different datasets.. jonathan had 1981-2010
# Check if there are higher resolution datafiles when choosing higher resolution..
dt = datetime.date.today()
date_list = [dt.year, dt.month, dt.day]
start0 = time.time()
predictands = ["GCEcom","20CRslp","GPCCcom"]
predictands = ["20CRslp","GPCCcom"]
#predictands = ["GISTEMP"]
#predictands = ['GCEcom']
bd = '/nobackup/users/krikken/SPESv2/'
bdid = bd+'inputdata/'
bdp = bd+'plots/'
bdnc = bd+'ncfiles/'
# Load these predictors, this does not mean that these are neceserally used.. see predictorz for those
predictors_1d = ['CO2EQ','NINO34','PDO','AMO','IOD']
predictors_3d = ['PERS','CPREC']
predictors = ['CO2EQ','NINO34','PDO','AMO','IOD','CPREC','PERS','PERS_TREND']
# Select method how to run
# NAMELIST
## Resolution, currently only 25 or 50 is supported..
resolution = 25 # 10, 25 or 50
## Redo full hindcast period and remove original nc output file?
overwrite = True
## Redo a specific month / year?
overwrite_m = False # Overwrite only the month specified with overw_m and overw_y
overw_m = 5 # Jan = 1, Feb = 2.. etc
overw_y = 2017
UPDATE = False
## Save a figure with the correlation between predictors and predictand
PLOT_PREDCOR = True
##
VALIDATION = True # Validates and makes figures of predicted values
DYN_MONAVG = False # Include the dynamical monthly averaging in the predictors
MLR_PRED = True # Include the trend of the last 3 month as predictor
FORECAST = True # Do forecast for given date?
HINDCAST = True # Validate forecast using leave n-out cross validation?
CROSVAL = True
CAUSAL = False
cv_years = 3 # Leave n out cross validation
## Validation period is 1961 - current
ens_size = 51
styear = 1901 # Use data from this year until current
stvalyear = 1961 # Start validation from this year until previous year
endyear = dt.year
endmonth = dt.month-1 # -1 as numpy arrays start with 0
tot_time = (dt.year - styear) * 12 + endmonth
# Defining some arrays used for writing labels and loading data
months = ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec']
monthzz = 'JFMAMJJASONDJFMAMJJASOND'
print 'Data from Jan '+str(styear)+' up to '+str(months[dt.month-2])+' '+str(dt.year)
print 'Predictands = ',predictands
print 'Predictors = ',predictors
print 'Horizontal resolution is ',str(resolution/10.),' degrees'
## Predefine arrays and make lats,lons according to specified resolution
## All url for data downloads
if resolution == 10:
targetgrid = 'griddes10.txt' # grid description file used for regridding
predadata = np.zeros((len(predictands),tot_time,180,360))
latx = 180; lonx = 360
latz = np.arange(89.5,-90.,-1.)
lonz = np.arange(-179.5,180.,1.)
elif resolution == 25:
targetgrid = 'griddes25.txt' # grid description file used for regridding
predadata = np.zeros((len(predictands),tot_time,72,144))
latx = 72; lonx = 144
latz = np.arange(88.75,-90.,-2.5)
lonz = np.arange(-178.75,180.,2.5)
elif resolution == 50:
targetgrid = 'griddes50.txt' # grid description file used for regridding
predadata = np.zeros((len(predictands),tot_time,36,82))
latx = 36; lonx = 82
latz = np.arange(87.5,-90.,-5)
lonz = np.arange(-177.5,180.,5)
# ************************************************************************
# Read in predictand data for fitting
# ************************************************************************
start1 = time.time()
print '-- Read in predictand data for fitting --'
predictorz = [] # Predefine empty array, fill with specified predictors for predictand
#predictorz_1d = []
#predictorz_3d = []
for p,predictand in enumerate(predictands):
if predictand == 'GISTEMP':
predictorz.append(['CO2EQ','NINO34','PDO','AMO','IOD','CPREC','PERS','PERS_TREND'])
url_gistemp = "https://data.giss.nasa.gov/pub/gistemp/gistemp1200_ERSSTv5.nc.gz"
#if check_timestap(url_gisstemp):
# loaddata(url_gisstemp,bd='inputdata/')
try: gistemp = cdo.selyear(str(styear)+'/2100',input = '-remapbil,targetgrid/griddes'+str(resolution)+'.txt inputdata/gistemp1200_ERSSTv5.nc',returnMaArray = 'tempanomaly')
except IOError:
if check_timestamp(url_gistemp) and UPDATE:
loaddata(url_gistemp,bd='inputdata/')
gistemp = cdo.selyear(str(styear)+'/2100',input = '-remapbil,targetgrid/griddes'+str(resolution)+'.txt inputdata/gistemp1200_ERSSTv5.nc',returnMaArray = 'temperature_anomaly')
mask = np.sum(gistemp.mask,axis=0) > 500
gistemp_nm = np.array(gistemp)
gistemp_nm[np.tile(mask,(gistemp.shape[0],1,1))]=np.nan # Put nans where too little data...
gistemp_nm[gistemp_nm==32767] = np.nan
gistemp_anom = anom(gistemp_nm,1980,2010,1901)
gistemp_anom
predadata[p,:] = gistemp_anom
elif predictand == 'GCEcom':
## These predictors are selelected for GCEcom in the first predictor selection step
predictorz.append(['CO2EQ','NINO34','PDO','AMO','IOD','CPREC','PERS','PERS_TREND'])
#predictorz_1d.append(['CO2EQ','NINO34','PDO','AMO','IOD'])
#predictorz_3d.append(['PERS','CPREC'])
url_ersstv5 = "http://climexp.knmi.nl/NCDCData/ersstv5.nc"
if check_timestamp(url_ersstv5) and UPDATE:
# Load ERSST, data starts from 1854
#ersst,ersst_lat,ersst_lon,ersst_time = loaddata(url_ersstv5,var="sst",bd='inputdata/')
loaddata(url_ersstv5,bd='inputdata/')
#ersst_rg = cdo.remapbil('targetgrid/griddes'+str(resolution)+'.txt',input = '-selyear,1948/2100 inputdata/ersstv5.nc',returnMaArray = 'sst').squeeze()
# TODO > if smaller resolution, load GHCN_CAMS 0.5 resolution iso regridding 2.5 resolution
#ersst_rg[ersst_rg==-1.8] = np.nan
#
url_ghcn_cams = "ftp://ftp.cpc.ncep.noaa.gov/wd51yf/GHCN_CAMS/ghcn_cams_1948_cur_2.5.grb"
if check_timestamp(url_ghcn_cams) and UPDATE:
loaddata(url_ghcn_cams,var="var11",bd='inputdata/')
# merge data, first regrid (bilinear) to targetgrid, then use fillmiss to fill in the coastal data with bilinear interpolation using 4 nearest neighbours.. this also fills the complete antarctic which has to be corrected manually.
gcecom = cdo.fillmiss(input = '-mergegrid -addc,273.15 -selyear,1948/2100 -remapbil,targetgrid/griddes'+str(resolution)+'.txt -setmissval,-999 inputdata/ersstv5.nc -remapbil,targetgrid/griddes'+str(resolution)+'.txt -setmissval,-999 inputdata/ghcn_cams_1948_cur_2.5.nc',returnMaArray = 'sst').squeeze()
#gcecom[:,60:,:][ersst_rg.mask[:gcecom.shape[0],60:,:]] = np.nan
mask = np.sum((gcecom == 271.35),axis=0)>100.
mask[61:,:] = True # Mask antarctic and sea ice of southern ocean
#gcecom[np.tile(mask,(gcecom.shape[0],1,1))] = np.nan
#gcecom[gcecom==271.35] = np.nan # Set all values below ice to nan
#loaddata("http://www-users.york.ac.uk/~kdc3/papers/coverage2013/had4_krig_v2_0_0.nc.gz",ret=False)
try: hadcrucw = cdo.selyear(str(styear)+'/2100',input = '-remapbil,targetgrid/griddes'+str(resolution)+'.txt inputdata/had4_krig_v2_0_0.nc',returnMaArray = 'temperature_anomaly')
except IOError:
if check_timestamp("http://www-users.york.ac.uk/~kdc3/papers/coverage2013/had4_krig_v2_0_0.nc.gz") and UPDATE:
loaddata("http://www-users.york.ac.uk/~kdc3/papers/coverage2013/had4_krig_v2_0_0.nc.gz",bd='inputdata/')
hadcrucw = cdo.selyear(str(styear)+'/2100',input = '-remapbil,targetgrid/griddes'+str(resolution)+'.txt inputdata/had4_krig_v2_0_0.nc',returnMaArray = 'temperature_anomaly')
hadcrucw_anom = anom(hadcrucw,1980,2010,1901)
gcecom_anom = anom(gcecom,1980,2010,1948)
#np.save('pickle/clim_gcecom.npy',clim_gcecom)
com = np.concatenate((hadcrucw_anom[:564+480,:],gcecom_anom[480:]),axis=0)
gcecom[:,mask] = np.nan
clim_gcecom = clim(gcecom,1980,2010,1948,keepdims=True)
# Where no or limited data fill with nans
#com[np.tile(mask,(com.shape[0],1,1))] = np.nan
com[:,mask] = np.nan
#clim_gcecom[:,mask] = np.nan
predadata[p,:] = com
elif predictand == 'HadCRU4CW':
# These predictors are selelected for HadCRU4CW in the first predictor selection step
predictorz.append(['CO2EQ','NINO34','PDO','AMO','IOD','PERS','CPREC'])
#predictorz_1d.append(['CO2EQ','NINO34','PDO','AMO','IOD'])
#predictorz_3d.append(['PERS','CPREC'])
# TAS anomalies relative to 1961-1990 climatology
if check_timestamp("http://www-users.york.ac.uk/~kdc3/papers/coverage2013/had4_krig_v2_0_0.nc.gz") and UPDATE:
loaddata("http://www-users.york.ac.uk/~kdc3/papers/coverage2013/had4_krig_v2_0_0.nc.gz",bd='inputdata/')
hadcrucw = cdo.selyear(str(styear)+'/2100',input = '-remapbil,targetgrid/griddes'+str(resolution)+'.txt inputdata/had4_krig_v2_0_0.nc',returnMaArray = 'temperature_anomaly')
#HadSAT,latHadSAT,lonHad SAT = loaddata("predsys/tas_Amon_HadCRUT4_1850-2013_CW_v2_anomaly.nc","temperature_anomaly")
#predadata[p,:] = hadcrucw[(styear-1850)*12:(endyear-1850)*12+endmonth,:,:]
predadata[p,:hadcrucw.shape[0],:] = hadcrucw
elif predictand == 'GPCCcom':
# These predictors are selelected for GPCCcom in the first predictor selection step
predictorz.append(['CO2EQ','NINO34','AMO','IOD','PERS'])
#predictorz_1d.append(['CO2EQ','NINO34',,'AMO','IOD'])
#predictorz_3d.append(['PERS'])
# Load GPCC precip data, starts in 1901 to current
url_gpcccom = "http://climexp.knmi.nl/GPCCData/gpcc_10_combined.nc"
if check_timestamp(url_gpcccom) and UPDATE:
loaddata(url_gpcccom,bd='inputdata/')
gpccprec = cdo.remapbil('targetgrid/griddes'+str(resolution)+'.txt',input = 'inputdata/gpcc_10_combined.nc',returnMaArray = 'prcp')
gpccprec = gpccprec[(styear-1901)*12:(endyear-1901)*12+endmonth,:,:]
gpccprec[gpccprec<=-1000.] = np.nan
clim_gpcc = clim(gpccprec,1980,2010,styear)
predadata[p,:] = anom(gpccprec,1980,2010,styear)
elif predictand == '20CRslp':
# These predictors are selelected for 20CRslp in the first predictor selection step
predictorz.append(['CO2EQ','NINO34','PDO','AMO','IOD','CPREC','PERS','PERS_TREND'])
#predictorz_1d.append(['CO2EQ','NINO34','PDO','AMO','IOD'])
#predictorz_3d.append(['PERS','CPREC'])
# Load 20CR (1851-2011)
if check_timestamp('http://climexp.knmi.nl/20C/prmsl.mon.mean.nc') and UPDATE:
loaddata('http://climexp.knmi.nl/20C/prmsl.mon.mean.nc',bd='inputdata/')
slp20cr = cdo.remapbil('targetgrid/griddes'+str(resolution)+'.txt',input = '-selvar,prmsl inputdata/prmsl.mon.mean.nc',returnMaArray = 'prmsl')[(styear-1851)*12:(1948-1851)*12,:,:]
## Load NCEP-NCAR reanalysis SLP (1948-current)
if check_timestamp('http://climexp.knmi.nl/NCEPNCAR40/slp.mon.mean.nc') and UPDATE:
loaddata('http://climexp.knmi.nl/NCEPNCAR40/slp.mon.mean.nc',bd='inputdata/')
slp21cr = cdo.remapbil('targetgrid/griddes'+str(resolution)+'.txt',input = 'inputdata/slp.mon.mean.nc',returnMaArray = 'slp')
## Combine both datasets, slp20cr 1901-1947, ncepncar 1948-current and remove climatology
predadata[p,:] = anom(np.concatenate((slp20cr/100.,slp21cr),axis=0),1980,2010,styear)
else:
print 'predictand not yet known.. exiting!'
sys.exit()
print '-- Done reading in predictand data for fitting, time = ',str(np.int(time.time()-start1)),' seconds --'
# ************************************************************************
# Read in predictor data for fitting
# ************************************************************************
start1 = time.time()
print '-- Read in predictor data for fitting --'
#predodata_1d = np.zeros((len(predictors_1d),predadata.shape[1]))
#predodata_3d = np.zeros((len(predictors_3d),predadata.shape[1],latx,lonx))
#predodata = []
predodata = np.zeros((len(predictors),predadata.shape[1],latx,lonx))
# Load 3d predictors (time,lat,lon)
for i,pred in enumerate(predictors):
if pred == 'CO2EQ': # CO2EQ RCP45 [Years,data] - 1765-2500]
data = urllib2.urlopen("http://climexp.knmi.nl/CDIACData/RCP45_CO2EQ.dat")
co2eq = np.repeat(np.genfromtxt(data),12,axis=0) # [Years (1765:2500) : CO2 concentration]
co2eqs = co2eq[(styear-int(co2eq[0,0]))*12:(endyear-int(co2eq[0,0]))*12+endmonth,:]
#predodata[i,:] = np.rollaxis(np.tile(co2eqs[:,1],(latx,lonx,1)),2,0)
predodata[i,:] = co2eqs[:,1][:,np.newaxis,np.newaxis]
#predodata_1d[i,:] = co2eqs[:,1]
elif pred == 'NINO34': # NINO34 [years,data] - [1854 - currently]
data = urllib2.urlopen("http://climexp.knmi.nl/NCDCData/ersst_nino3.4a.dat")
nino34 = np.genfromtxt(data) # Maanden (jaar.maand) / nino34
nino34s = nino34[(styear-1854)*12:(endyear-1854)*12+endmonth,:]
#predodata[i,:] = np.rollaxis(np.tile(nino34s[:,1],(latx,lonx,1)),2,0)
predodata[i,:] = nino34s[:,1][:,np.newaxis,np.newaxis]
#predodata_1d[i,:] = nino34s[:,1]
elif pred == 'QBO': # QBO [years,months,data] - [1501-2300]
data = urllib2.urlopen("http://climexp.knmi.nl/data/iqbo_30.dat")
qbo = np.genfromtxt(data) # Maanden (jaar.maand) / qbo
qbos = qbo[(styear-int(qbo[0,0]))*12:(endyear-int(qbo[0,0]))*12+endmonth,:]
#predodata[i,:] = np.rollaxis(np.tile(qbos[:,2],(latx,lonx,1)),2,0)
#predodata_1d[i,:] = qbos[:,1]
predodata[i,:] = qbos[:,2][:,np.newaxis,np.newaxis]
elif pred == 'IOD': # IOD [years,jan,feb,...,dec] - [1854-2017]
data = urllib2.urlopen("http://climexp.knmi.nl/NCDCData/dmi_ersst.dat")
iod = np.genfromtxt(data) # Maanden (jaar.maand) / IOD
iodys = iod[:,0]
iodda = iod[:,1:].ravel()
iods = iodda[(styear-int(iodys[0]))*12:(endyear-int(iodys[0]))*12+endmonth]
#predodata[i,:] = np.rollaxis(np.tile(iods,(latx,lonx,1)),2,0)
#predodata_1d[i,:] = iods[:,1]
predodata[i,:] = iods[:,np.newaxis,np.newaxis]
elif pred == 'PDO': # PDO [years,jan,feb,...,dec]
data = urllib2.urlopen("http://climexp.knmi.nl/UWData/pdo_ersst.dat")
pdo = np.genfromtxt(data) # Maanden (jaar.maand) / PDO
pdoys = pdo[:,0]
pdoda = pdo[:,1:].ravel()
pdos = pdoda[(styear-int(pdoys[0]))*12:(endyear-int(pdoys[0]))*12+endmonth]
#predodata[i,:] = np.rollaxis(np.tile(pdos,(latx,lonx,1)),2,0)
#predodata_1d[i,:] = pdos[:,1]
predodata[i,:] = pdos[:,np.newaxis,np.newaxis]
elif pred == 'AMO': # AMO [years,jan,feb,...,dec] - [1854-2017]
#data = urllib2.urlopen("http://climexp.knmi.nl/data/iamo_ersst_ts.dat")
data = urllib2.urlopen("http://climexp.knmi.nl/NCDCData/amo_ersst_ts.dat")
amo = np.genfromtxt(data) # Maanden (jaar.maand) / AMO
amoys = amo[:,0]
amoda = amo[:,1:].ravel()
amos = amoda[(styear-int(amoys[0]))*12:(endyear-int(amoys[0]))*12+endmonth]
#predodata[i,:] = np.rollaxis(np.tile(amos,(latx,lonx,1)),2,0)
#predodata_1d[i,:] = amos[:,1]
predodata[i,:] = amos[:,np.newaxis,np.newaxis]
elif pred == 'SIE_EA': # Sea ice extent Eurasia
sic1 = cdo.selyear(str(styear)+'/2100',input = 'inputdata/sic_1850-2013.nc',returnMaArray = 'seaice_conc')
lonz = np.arange(-44.5,315.,1.) # Quicker than loading data from netcdf :-)
#lon = cdo.selvar('longitude',input = 'inputdata/G10010_SIBT1850_v1.1.nc',returnMaArray = 'longitude')
area = cdo.selvar('cell_area',input = 'inputdata/area_sic_1x1.nc',returnMaArray = 'cell_area')
sic2 = cdo.selyear('2014/2100',input='inputdata/conc_n.nc',returnMaArray = 'ice')[1:,:].squeeze()
# Calculat extent, so assume more than 15% as fully ice covered..
sic2[sic2.mask] = 0.
sic2[sic2<0.15] = 0.
sic2[sic2>0.15] = 1.
sic1[sic1<15.] = 0.
sic1[sic1>=15.] = 1.
sic = np.concatenate((sic1,sic2),axis=0)
sia = sic * area[None,:]
sie_ea = np.nansum(np.nansum(np.roll(sia,-1)[:,:,:180],axis=1),axis=1) / 1.e12
sie_am = np.nansum(np.nansum(np.roll(sia,-1)[:,:,180:],axis=1),axis=1) / 1.e12
sie_ea_anom = anom(sie_ea,1980,2010,1901)
sie_am_anom = anom(sie_am,1980,2010,1901)
predodata[i,:] = sie_ea_anom[:,np.newaxis,np.newaxis]
#else:
#print 'predictor: '+pred+' not known.. exiting!'
#sys.exit()
# Load 3d predictors (time,lat,lon)
#for i,pred in enumerate(predictors_3d):
elif pred == 'LSST': # Local SST
print 'LSST not operational'
sys.exit()
elif pred == 'PERS':
# Fill array later
predodata[i,:] = np.nan
elif pred == 'PERS_TREND':
# Fill array later
predodata[i,:] = np.nan
elif pred == 'CPREC': # Cum precip [time,lat,lon] - 1901 -current
if 'GPCCcom' in predictands:
predodata[i,:] = predadata[predictands.index('GPCCcom'),:]
else:
gpccprec,gpccprec_lats,gpccprec_lons,gpccprec_time = loaddata("http://climexp.knmi.nl/GPCCData/gpcc_10_combined.nc",var="prcp",bd='inputdata/')
if resolution != 10:
gpccprec = cdo.remapbil('targetgrid/griddes'+str(resolution)+'.txt',input = 'inputdata/gpcc_10_combined.nc',returnMaArray = 'prcp')
gpccprec = gpccprec[(styear-1901)*12:(endyear-1901)*12+endmonth,:,:]
gpccprec[gpccprec<-1000.] = np.nan
predodata[i,:] = anom(gpccprec,1980,2010,styear)
else:
print 'predictor: '+pred+' not known.. exiting!'
sys.exit()
print '-- Done reading in predictor data for fitting, time = ',str(np.int(time.time()-start1)),' seconds --'
# Normalize predodata..
#predodata = (predodata - np.nanmean(predodata,axis=1)[:,np.newaxis,:,:]) / np.nanstd(predodata,axis=1)[:,np.newaxis,:,:]
#sys.exit()
# *************************************************************************
# Now start the predictor selection and the forecasting / hindcasting loop
# *************************************************************************
for p,predictand in enumerate(predictands):
# Fill persistence predictor with predictand data
predodata[predictors.index('PERS'),:] = predadata[p,:]
if predictand != 'GPCCcom':
predodata[predictors.index('PERS_TREND'),:] = predadata[p,:]
# Only use predictors that are selected using 1st step in selection process
ps = []
for pr in predictorz[p]:
ps.append(predictors.index(pr))
predodata2 = predodata[ps,:]
#ps = []
#for pr in predictorz_1d[p]:
# ps.append(predictors_1d.index(pr))
#predodata2_1d = predodata_1d[ps,:]
print 'Predictand: ';predictand
print 'Predictors: ',predictorz[p]
#print '3d predictors: ',predictorz_3d[p]
# Try to collect last saved forecast month
try:
datanc = Dataset(bdnc+'pred_v2_'+predictand+'.nc')
timenc = datanc.variables['time']
year_nc = num2date(timenc[:][-1],timenc.units).year
month_nc = num2date(timenc[:][-1],timenc.units).month
#dr = date_range(num2date(timenc[:][-1],timenc.units),datetime.datetime.today())
datanc.close()
print 'last forecast month is: '+str(month_nc)
mon_range = range(month_nc,endmonth+1)
print mon_range
except IOError: # If file does not exist do full hindcast
month_nc = 0
mon_range = range(month_nc,12)
print 'no previous output, do full hindcast!'
if FORECAST and not HINDCAST: # Only redo forecast loop
mon_range = range(12)
if overwrite:
month_nc = 0
mon_range = range(month_nc,12)
if os.path.isfile(bdnc+'pred_v2_'+predictand+'.nc') and HINDCAST:
os.rename(bdnc+'pred_v2_'+predictand+'.nc',bdnc+'pred_v2_'+predictand+'_moved.nc')
if os.path.isfile(bdnc+'fit_data/rg_v2_'+predictand+'.nc'):
os.rename(bdnc+'fit_data/rg_v2_'+predictand+'.nc',bdnc+'rg_v2_'+predictand+'_moved.nc')
if overwrite_m:
mon_range = [overw_m]
#else:
# mon_range = range(month_nc,12)
# Exit loop if already up to date..
#if len(dr)==0 and not overwrite and not overwrite_m:
# print 'already up to date for ',predictand
# continue
#for m,mon in enumerate(months):
for m in mon_range:
if m < endmonth+1:
years = np.arange(1902,dt.year+1)
else:
years = np.arange(1902,dt.year)
mo = np.array([m-3,m-2,m-1]) # Select months to calculate seasonal average predictors
ma = np.array([m+1,m+2,m+3]) # Select months to calculate seasonal average predictors
print 'prediction month = ',months[m]
print 'predictor season = ',np.asarray(months+months)[mo]
print 'predictor season = ',np.asarray(months+months)[ma]
#print years
# Seasonalize data.. m is the month the forecast is made, predo is 3 months prior to m and preda 3 months after m
#predo_seas,preda_seas = seazon(predodata2,predadata[p,:],m)
predo_seas3,preda_seas = seazon(predodata2,predadata[p,:],m)
if DYN_MONAVG:
predo_seas1 = seazon(predodata2,predadata[p,:],m,month_avg=1)[0]
predo_seas5 = seazon(predodata2,predadata[p,:],m,month_avg=5)[0]
if MLR_PRED:
predo_trend = seazon_trend(predodata2,m) # Predictor trend over last 3 months
predo_prad = seazon_prad(predodata2,m) # Predictor data at future timestep (i.e. same time step as predictand)
# Put PERS_TREND at its correct location in the predictor data
if predictand != 'GPCCcom':
predo_seas3[predictorz[p].index('PERS_TREND'),:] = predo_trend[predictorz[p].index('PERS_TREND'),:]
#sys.exit()
if FORECAST:
print 'Forecasting mode'
train = np.ones((predo_seas3.shape[1]),dtype='bool')
n=1
train[-1] = False
test = ~train
if predictand == 'GPCCcom':
train[:49] = False
predo_tr = predo_seas3[:,train,:,:]
predo_te = predo_seas3[:,test,:,:]
if MLR_PRED:
predo_trend_tr = predo_trend[:,train,:,:]
predo_trend_te = predo_trend[:,test,:,:]
else: predo_trend_tr = []; predo_trend_te = []
year = years[test]
print 'test years: ',year
try: preda_tr = preda_seas[train,:]
except IndexError: preda_tr = preda_seas[train[:-1],:]
if MLR_PRED:
try: predo_prad_tr = predo_prad[:,train,:]
except IndexError: predo_prad_tr = predo_prad[:,train[:-1],:]
else: predo_prad_tr = []
t0 = time.time()
regr_loop(predo_tr,predo_te,preda_tr,year,m,False,FORECAST,DYN_MONAVG,MLR_PRED,n,ens_size,latx,lonx,bdnc,predictand,predictorz[p],resolution,latz,lonz,predo_trend_tr=predo_trend_tr,predo_prad_tr=predo_prad_tr,predo_trend_te=predo_trend_te,stvalyear=stvalyear,PLOT_PREDCOR=PLOT_PREDCOR)
print 'time regr_loop = ',time.time()-t0
if HINDCAST and CAUSAL:
print 'Hindcasting mode, causal ',str(stvalyear),'-current'
for y in range(stvalyear,years[-1]):
train = np.zeros((predo_seas3.shape[1]),dtype='bool')
train[:np.argmin(np.abs(years-y))] = True
test = np.zeros((predo_seas3.shape[1]),dtype='bool')
test[np.argmin(np.abs(years-y))] = True
#print y
year = years[test]
print 'train data is 1901 ... ',years[train][-1]
print 'test data is : ',year
predo_tr = predo_seas3[:,train,:,:]
predo_te = predo_seas3[:,test,:,:]
#preda_tr = preda_seas[train,:,:]
try: preda_tr = preda_seas[train,:]
except IndexError: preda_tr = preda_seas[train[:-1],:]
try: preda_te = preda_seas[test,:,:]
except IndexError: preda_te = preda_seas[test[:-1],:]
print predo_tr.shape,preda_tr.shape
if predo_tr.shape[1] != preda_tr.shape[0]:
sys.exit()
if MLR_PRED:
try: predo_prad_tr = predo_prad[:,train,:]
except IndexError: predo_prad_tr = predo_prad[:,train[:-1],:]
try: predo_prad_te = predo_prad[:,test,:]
except IndexError: predo_prad_te = predo_prad[:,test[:-1],:]
else: predo_prad_te = []; predo_prad_tr = []
n=1
regr_loop(predo_tr,predo_te,preda_tr,year,m,HINDCAST,False,DYN_MONAVG,MLR_PRED,n,ens_size,latx,lonx,bdnc,predictand,predictorz[p],resolution,latz,lonz,preda_te=preda_te,predo_trend_tr=predo_trend_tr,predo_prad_tr=predo_prad_tr,predo_trend_te=predo_trend_te,predo_prad_te=predo_prad_te,stvalyear=stvalyear)
elif HINDCAST and CROSVAL:
n = cv_years
print 'Hindcasting mode, leave ',str(n),' out cross-validation'
if predo_seas3.shape[1] % n > 0:
#cvl_range = range(predo_seas3.shape[1]/n+1)
cvl_range = range((years[-1]-stvalyear)/n+1)
else:
#cvl_range = range(predo_seas3.shape[1]/n)
cvl_range = range((years[-1]-stvalyear)/n)
for cvl in cvl_range:
samesize = predo_seas3.shape[1] == preda_seas.shape[0]
train = np.ones((predo_seas3.shape[1]),dtype='bool')
stvalyear_idx = np.argmin(np.abs(years-stvalyear))
train[cvl*n+stvalyear_idx:cvl*n+n+stvalyear_idx] = False
test = ~train
train[-1]=False
test[-1]=False
#if predictand == 'GPCCcom':
# train[:48] = False
#if not samesize: test[-2] = False
if predictand == 'GPCCcom':
train[:49] = False
year = years[test]
print 'train data is: ',years[train][:5],' ... ',years[train][-5:]
print 'test data is : ',year
#print 'test years: ',year
n=np.sum(test)
if n == 0:
continue
predo_tr = predo_seas3[:,train,:,:]
predo_te = predo_seas3[:,test,:,:]
if MLR_PRED:
predo_trend_tr = predo_trend[:,train,:,:]
predo_trend_te = predo_trend[:,test,:,:]
else: predo_trend_tr = [];predo_trend_te = []
#preda_tr = preda_seas[train,:,:]
try: preda_tr = preda_seas[train,:]
except IndexError: preda_tr = preda_seas[train[:-1],:]
try: preda_te = preda_seas[test,:,:]
except IndexError: preda_te = preda_seas[test[:-1],:]
print predo_tr.shape,preda_tr.shape
if predo_tr.shape[1] != preda_tr.shape[0]:
sys.exit()
if MLR_PRED:
try: predo_prad_tr = predo_prad[:,train,:]
except IndexError: predo_prad_tr = predo_prad[:,train[:-1],:]
try: predo_prad_te = predo_prad[:,test,:]
except IndexError: predo_prad_te = predo_prad[:,test[:-1],:]
else: predo_prad_tr = []; predo_prad_te = []
t0 = time.time()
regr_loop(predo_tr,predo_te,preda_tr,year,m,HINDCAST,False,DYN_MONAVG,MLR_PRED,n,ens_size,latx,lonx,bdnc,predictand,predictorz[p],resolution,latz,lonz,preda_te=preda_te,predo_trend_tr=predo_trend_tr,predo_prad_tr=predo_prad_tr,predo_trend_te=predo_trend_te,predo_prad_te=predo_prad_te,stvalyear=stvalyear)
print 'regr_loop time: ',time.time()-t0
elif HINDCAST:
print 'either CROSVAL or CAUSAL should be set to true'
if VALIDATION:
# predictand = 'GCEcom'
print 'Start validation for the last year of all months'
for p,predictand in enumerate(predictands):
dataset=Dataset(bdnc+'pred_v2_'+predictand+'.nc')
timenz = dataset.variables['time']
year_nc = num2date(timenz[:][-1],timenz.units).year
month_nc = num2date(timenz[:][-1],timenz.units).month
#timen = num2date(timenz[:],timenz.units)
pre = dataset.variables[predictand+'_fc'][:]
obs = dataset.variables[predictand+'_obs'][:]
ref = dataset.variables[predictand+'_ref'][:]
co2 = dataset.variables[predictand+'_co2'][:]
dataset.close()
#mon_range = range(12)
#mon
#monthz,yearz = np.zeros(len(timen)),np.zeros(len(timen))
#for i,d in enumerate(timen):
# monthz[i] = d.month
# yearz[i] = d.year
if predictand == 'GCEcom':
var = 'Surface air temperature'
clevz = np.array((-2.,-1.,-0.5,-0.2,0.2,0.5,1.,2.))
cmap1 = matplotlib.colors.ListedColormap(['#000099','#3355ff','#66aaff','#77ffff','#ffffff','#ffff33','#ffaa00','#ff4400','#cc0022'])
cmap2 = matplotlib.colors.ListedColormap(['#3355ff','#66aaff','#77ffff','#ffffff','#ffff33','#ffaa00','#ff4400'])
cmap_under = '#000099'
cmap_over = '#cc0022'
elif predictand == 'GPCCcom':
var = 'Surface precipitation'
clevz = np.array((-200.,-100.,-50.,-20.,20.,50.,100.,200.))
cmap1 = matplotlib.colors.ListedColormap(['#993300','#cc8800','#ffcc00','#ffee99','#ffffff','#ccff66','#33ff00','#009933','#006666'])
cmap2 = matplotlib.colors.ListedColormap(['#cc8800','#ffcc00','#ffee99','#ffffff','#ccff66','#33ff00','#009933'])
cmap_under = '#993300'
cmap_over = '#006666'
elif predictand == '20CRslp':
var = 'Mean sea level pressure'
clevz=np.array((-4.,-2.,-1.,-0.5,0.5,1.,2.,4.))
cmap1 = matplotlib.colors.ListedColormap(['#000099','#3355ff','#66aaff','#77ffff','#ffffff','#ffff33','#ffaa00','#ff4400','#cc0022'])
cmap2 = matplotlib.colors.ListedColormap(['#3355ff','#66aaff','#77ffff','#ffffff','#ffff33','#ffaa00','#ff4400'])
cmap_under = '#000099'
cmap_over = '#cc0022'
#for m in (np.unique(timez[0,:])-1).astype(int):
for m in [month_nc-1]:
#for m in range(12):
mon = str(m+1)
if len(str(mon)) == 1: mon = '0'+mon
#year = str(int(yearz[m::12][-1]-y))
year = str(year_nc)
season = monthzz[m+1:m+4]
print 'validation for '+season+' '+year
bdpo = bdp+predictand+'/'+str(resolution)+'/'+year+mon+'/'
if not os.path.exists(bdpo):
os.makedirs(bdpo)
pref = pre[m::12,:][-1,:]
prem = pre[m::12,:][:-1,:]
obsm = obs[m::12,:][:-1,:]
refm = ref[m::12,:][:-1,:]
co2m = co2[m::12,:][:-1,:]
rmse_pred = f_rmse(prem,obsm,SS=True,ref=refm)
crps_pred_co2 = f_crps(prem,obsm,SS=True,ref=co2m)
crps_pred = f_crps(prem,obsm,SS=True,ref=refm)
corr_pred,corrp_pred = linregrez(np.nanmean(prem,axis=1),obsm,COR=True)
tercile = tercile_category(prem,pref)
tmp = np.nanmean(pref,axis=0)
posneg = tmp > 0.
above = 1.-(np.sum(pref>0,axis=0)/51.)
below = 1.-(np.sum(pref<0,axis=0)/51.)
sig_ensmean = np.ones_like(crps_pred)
sig_ensmean[posneg] = above[posneg]
sig_ensmean[~posneg] = below[~posneg]
plot_climexp(rmse_pred,
'RMSESS hindcasts, climatology as reference (1961-current)',
'SPECS Empirical Seasonal Forecast: '+var+' ('+season+' '+year+')',
'Ensemble size: 51 | Forecast generation date: '+dt.strftime("%d/%m/%Y")+' | base time: '+months[m]+' '+year,
predictand = predictand,
fname=bdpo+predictand+'_rmsess_'+year+mon+'.png',
clevs = np.array((-0.5,-0.35,-0.2,-0.1,0.1,0.2,0.35,0.5)),
cmap=cmap2,
cmap_under = cmap_under,
cmap_over = cmap_over,
)
plot_climexp(crps_pred,
'CRPSS hindcasts, reference: climatology (1961-'+str(year_nc-1)+')',
'SPECS Empirical Seasonal Forecast: '+var+' ('+season+' '+year+')',
'Ensemble size: 51 | Forecast generation date: '+dt.strftime("%d/%m/%Y")+' | base time: '+months[m]+' '+year,
predictand = predictand,
cmap=cmap2,
cmap_under = cmap_under,
cmap_over = cmap_over,
fname=bdpo+predictand+'_crpss_'+year+mon+'.png',
clevs = np.array((-0.5,-0.35,-0.2,-0.1,0.1,0.2,0.35,0.5)),
)
plot_climexp(crps_pred_co2,
'CRPSS hindcasts, reference: hindcasts with only CO2 as predictor (1961-'+str(year_nc-1)+')',
'SPECS Empirical Seasonal Forecast: '+var+' ('+season+' '+year+')',
'Ensemble size: 51 | Forecast generation date: '+dt.strftime("%d/%m/%Y")+' | base time: '+months[m]+' '+year,
predictand = predictand,
cmap = cmap2,
fname=bdpo+predictand+'_crpss_detrended_clim_'+year+mon+'.png',
clevs = np.array((-0.5,-0.35,-0.2,-0.1,0.1,0.2,0.35,0.5)),
cmap_under = cmap_under,
cmap_over = cmap_over,
)
plot_climexp(np.nanmean(pref,axis=0),
'Ensemble mean anomaly (wrt 1980-2010)',
'SPECS Empirical Seasonal Forecast: '+var+' ('+season+' '+year+')',
'Ensemble size: 51 | Forecast generation date: '+dt.strftime("%d/%m/%Y")+' | Stippled where NOT significant at 10% level'+' | base time: '+months[m]+' '+year,
sig=sig_ensmean,
cmap=cmap2,
predictand = predictand,
cmap_under = cmap_under,
cmap_over = cmap_over,
fname=bdpo+predictand+'_ensmean_'+year+mon+'.png',
clevs = clevz,
MEAN=True,
)
plot_climexp(corr_pred,
'Correlation between hindcast anomaly and observations (1961-'+str(year_nc-1)+'',
'SPECS Empirical Seasonal Forecast: '+var+' ('+season+' '+year+')',
'Ensemble size: 51 | Forecast generation date: '+dt.strftime("%d/%m/%Y")+' | Stippled where signficant at 5% level'+' | base time: '+months[m]+' '+year,
sig = corrp_pred,
predictand = predictand,
fname=bdpo+predictand+'_correlation_'+year+mon+'.png',
clevs = np.arange(-1.,1.01,0.2),
)
plot_climexp(tercile,
'Probabilty (most likely tercile of '+var+'), based on 1961-'+str(year_nc-1)+' hindcasts',
'SPECS Empirical Seasonal Forecast: '+var+' ('+season+' '+year+')',
'Ensemble size: 51 | Forecast generation date: '+dt.strftime("%d/%m/%Y")+' | base time: '+months[m]+' '+year,
cmap = cmap1,
predictand = predictand,
fname=bdpo+predictand+'_tercile_'+year+mon+'.png',
clevs = np.array((-100,-70,-60,-50,-40,40,50,60,70,100)),
barticks = ['100%','70%','60%', '50%', '40%', '40%', '50%','60%', '70%', '100%'],
TERCILE=True,
)
#plt.annotate('<---- below lower tercile
import time
print 'Total time taken is: ',np.int((time.time()-start0)//60),' minutes and ',np.int((time.time()-start0)%60), 'seconds'
## Regression loop function
TEST = False
if TEST:
# Data from 1981 onwards..
# S5 data are note anomalies, hence either make them anomalies or add climatology to our data..
# First try to add climatology (1980-2010) to our data..
obs_seas = nans_like(predodata[p,-682:,:])
for m in range(12):
tmp = seazon_prad(predadata[p,-682:,:]+clim_gcecom[-682:,:],m)
obs_seas[m::12,:][:tmp.shape[0],:] = tmp
clim_seas = clim(obs_seas,1980,2010,1961,keepdims=True)
nc1 = Dataset('ncfiles/pred_v2_GCEcom_causal.nc') # Baseline forecast
fc1 = nc1.variables['GCEcom_fc'][:] #+ clim_seas[:,None,:,:]
ref = nc1.variables['GCEcom_ref'][:] #+ clim_seas[:,None,:,:]
obs = nc1.variables['GCEcom_obs'][:] #+ clim_seas
lons = nc1.variables['longitude'][:]
lats = nc1.variables['latitude'][:]
nc1.close()
nc2 = Dataset('ncfiles/pred_v2_GCEcom_l3o.nc') # Fit trend and mean of predictor on predictand
fc2 = nc2.variables['GCEcom_fc'][:] #+ clim_seas[:,None,:,:]
ref2 = nc2.variables['GCEcom_ref'][:] #+ clim_seas[:,None,:,:]
#obs2 = nc2.variables['GCEcom_obs'][:]
nc2.close()
nc3 = Dataset('ncfiles/pred_v2_GCEcom.nc') # Fit trend and mean of predictor on predictor (Seems physically better)
fc3 = nc3.variables['GCEcom_fc'][:] #+ clim_seas[:,None,:,:]
ref3 = nc3.variables['GCEcom_ref'][:]
#obs3 = nc3.variables['GCEcom_obs'][:]
nc3.close()
#nc4 = Dataset('ncfiles/pred_v2_GCEcom_l3o_mlr_noPERS.nc') # Use trend and mean as separate predictors
#fc4 = nc4.variables['GCEcom_fc'][:] #+ clim_seas[:,None,:,:]
#ref4 = nc4.variables['GCEcom_ref'][:]
#nc4.close()
#nc5 = Dataset('ncfiles/pred_v2_GCEcom_l3o_mlr_noPERS_ens.nc') # Use trend and mean as separate predictors
#fc5 = nc5.variables['GCEcom_fc'][:] #+ clim_seas[:,None,:,:]
#ref4 = nc4.variables['GCEcom_ref'][:]
#nc5.close()
#nc5 = Dataset('ncfiles/pred_GCEcom_dmavg.nc') # Use either 1, 3 or 5 monthly average for predictors
#fc5 = nc5.variables['GCEcom_fc'][:]
#nc5.close()
#nc6 = Dataset('ncfiles/pred_GCEcom_dmavg_trend.nc') # Use either 1, 3 or 5 monthly average for predictors
#fc6 = nc6.variables['GCEcom_fc'][:]
#nc6.close()
for m in [1,4,7,10]:
# Load ecmwf data
s5 = load_ecmwf2(var='t2m',m=m,anom=True)
#bias = np.nanmean(np.nanmean(s5 - clim_seas[m::12,None,:,:][20:-1,:],axis=0),axis=0)
#crps1 = f_crps(fc1[m::12,:][-37:-1,:],obs[m::12,:][-37:-1,:],SS=True,ref=ref[m::12,:][-37:-1,:])
#crps2 = f_crps(fc2[m::12,:][-37:-1,:],obs[m::12,:][-37:-1,:],SS=True,ref=ref2[m::12,:][-37:-1,:])
#crps3 = f_crps(fc3[m::12,:][-37:-1,:],obs[m::12,:][-37:-1,:],SS=True,ref=ref2[m::12,:][-37:-1,:])
#crps4 = f_crps(fc4[m::12,:][-37:-1,:],obs[m::12,:][-37:-1,:],SS=True,ref=ref3[m::12,:][-37:-1,:])
#crps5 = f_crps(fc5[m::12,:][:-1,:],obs[m::12,:][:-1,:],SS=True,ref=ref[m::12,:][:-1,:])
#crps6 = f_crps(fc6[m::12,:][:-1,:],obs[m::12,:][:-1,:],SS=True,ref=ref[m::12,:][:-1,:])
#crps_s5 = f_crps(s5,obs[m::12,:][-37:-1,:],SS=True,ref=ref2[m::12,:][-37:-1,:])
#crps1_2 = f_crps(fc1[m::12,:][:-1,:],obs[m::12,:][:-1,:],SS=True,ref=fc2[m::12,:][:-1,:])
#crps1_3 = f_crps(fc1[m::12,:][:-1,:],obs[m::12,:][:-1,:],SS=True,ref=fc3[m::12,:][:-1,:])
#crps5_4 = f_crps(fc5[m::12,:][:-1,:],obs[m::12,:][:-1,:],SS=True,ref=fc4[m::12,:][:-1,:])
crps3_s5 = f_crps(fc3[m::12,:][-37:-1,:],obs[m::12,:][-37:-1,:],SS=True,ref=s5[:])
#crpss5 = f_crps(s5,obs[m::12,:],SS=True,ref=ref3[m::12,:][-37:-1,:])
#crps2_s5 = f_crps(fc2[m::12,:][-37:-1,:],obs[m::12,:][-37:-1,:],SS=True,ref=s5[:])
#crps2_1 = f_crps(fc2[m::12,:][:-1,:],obs[m::12,:][59:-1,:],SS=True,ref=fc1[m::12,:][59:-1,:])
#crps3_2 = f_crps(fc3[m::12,:][:-1,:],obs[m::12,:][59:-1,:],SS=True,ref=fc2[m::12,:][:-1,:])
cmap2 = matplotlib.colors.ListedColormap(['#3355ff','#66aaff','#77ffff','#ffffff','#ffff33','#ffaa00','#ff4400'])
cmap_u = '#000099'
cmap_o = '#cc0022'
clev1 = np.array((-0.5,-0.35,-0.2,-0.1, 0.1, 0.2,0.35,0.5))
clev2 = np.array((-0.2,-0.1, -0.05,-0.025,0.025,0.05,0.1, 0.2))
#plotdata(crps1,lons=lons,lats=lats,clev=clev1,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='causal',fname=str(m)+'crps1.png',PLOT=False,extend=True)
#plotdata(crps2,lons=lons,lats=lats,clev=clev1,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='leave-3-out',fname=str(m)+'crps2.png',PLOT=False,extend=True)
#plotdata(crps3,lons=lons,lats=lats,clev=clev1,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='leave-3-out mlr',fname=str(m)+'crps3.png',PLOT=False,extend=True)
#plotdata(crps4,lons=lons,lats=lats,clev=clev1,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='l3o mlr noPERS',fname=str(m)+'crps4.png',PLOT=False,extend=True)
#plotdata(crps_s5,lons=lons,lats=lats,clev=clev1,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='s5',fname=str(m)+'crps_s5.png',PLOT=False,extend=True)
#plotdata(crps6,lons=lons,lats=lats,clev=clev1,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='dmavg_pred',fname=str(m)+'crps6.png',PLOT=False,extend=True)
#plotdata(crps2_1,lons=lons,lats=lats,clev=clev2,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='crps2_1',fname=str(m)+'crps1_2.png',PLOT=False,extend=True)
#plotdata(crps3_2,lons=lons,lats=lats,clev=clev2,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='crps3_2',fname=str(m)+'crps2_3.png',PLOT=False,extend=True)
#plotdata(crps2_s5,lons=lons,lats=lats,clev=clev2,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='crps2_s5',fname=str(m)+'crps2_s5.png',PLOT=False,extend=True)
plotdata(crps3_s5,lons=lons,lats=lats,clev=clev2,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='crps3_s5',fname=str(m)+'crps3_s5.png',PLOT=False,extend=True)
#plotdata(crps1_6,lons=lons,lats=lats,clev=clev2,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='crps1_6',fname=str(m)+'crps1_6.png',PLOT=False,extend=True)
#plotdata(crps5_4,lons=lons,lats=lats,clev=clev2,cmap=cmap2,cmap_u=cmap_u,cmap_o=cmap_o,title='crps5_4',fname=str(m)+'crps5_4.png',PLOT=False,extend=True)
for m in range(12):
po,pa = predo_seas3,preda_seas = seazon(predodata[:,:-1,:],predadata[0,:-1,:],m)
mask = get_boreal_mask()
if po.shape[1] != pa.shape[0]:
po = po[:,:-1,:]
sie_noco2 = remove_co2(po[8,:-1,:],po[0,:-1,:])
tas_noco2 = remove_co2(pa[:-1,:],po[0,:-1,:])
mask[:,:80]=False
tas_bor = np.nanmean(tas_noco2[:,mask],axis=1)
#plot_regr(sie_noco2[:,20,20],tas_bor)
print m
plotcor2d(tas_noco2,sie_noco2,CLICK_R=True)
#plotcor2d(tas_noco2,sie_noco2,CLICK_R=True)