forked from erinwalker1/vert_res_attribution
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Fig5_zonal_winds_maps.py
371 lines (293 loc) · 17.9 KB
/
Fig5_zonal_winds_maps.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
# -*- coding: utf-8 -*-
"""
Created on Tue Feb 23 16:00:10 2021
Updated on Wed Jul 21 11:55:13 2021
@author: Erin Walker
"""
import matplotlib.pyplot as plt
import numpy as np
import xarray as xr
import cartopy.crs as ccrs
from cartopy.util import add_cyclic_point
from scipy.stats import ks_2samp
"""
P = Present
F = 3Deg
C = Control
DA = Double All
DT = Double Tropopause
Q = Quadruple All
NS = No Stratopshere
"""
filepath = "" # Filepath to ensemble mean files
p_file_c = xr.open_dataset(filepath + "daily_ensmean_pres_c.nc",chunks={'time':30})
f_file_c = xr.open_dataset(filepath + "daily_ensmean_3deg_c.nc")
p_file_da = xr.open_dataset(filepath + "daily_ensmean_pres_da.nc",chunks={'time':30})
f_file_da = xr.open_dataset(filepath + "daily_ensmean_3deg_da.nc")
p_file_dt = xr.open_dataset(filepath + "daily_ensmean_pres_dt.nc",chunks={'time':30})
f_file_dt = xr.open_dataset(filepath + "daily_ensmean_3deg_dt.nc")
p_file_ns = xr.open_dataset(filepath + "daily_ensmean_pres_ns.nc",chunks={'time':30})
f_file_ns = xr.open_dataset(filepath + "daily_ensmean_3deg_ns.nc")
p_file_q = xr.open_dataset(filepath + "daily_ensmean_pres_q.nc",chunks={'time':30})
f_file_q = xr.open_dataset(filepath + "daily_ensmean_3deg_q.nc")
pval_filepath = "" # Filepath to the pvals
c_pval_10 = xr.open_dataset(pval_filepath + "c_pval_10hpa_uwinds.nc")
c_pval_500 = xr.open_dataset(pval_filepath + "c_pval_500hpa_uwinds1.nc")
c_pval_850 = xr.open_dataset(pval_filepath + "c_pval_850hpa_uwinds1.nc")
da_pval_10 = xr.open_dataset(pval_filepath + "da_pval_10hpa.nc")
da_pval_500 = xr.open_dataset(pval_filepath + "da_pval_500hpa.nc")
da_pval_850 = xr.open_dataset(pval_filepath + "da_pval_850hpa.nc")
dt_pval_10 = xr.open_dataset(pval_filepath + "dt_pval_10hpa.nc")
dt_pval_500 = xr.open_dataset(pval_filepath + "dt_pval_500hpa.nc")
dt_pval_850 = xr.open_dataset(pval_filepath + "dt_pval_850hpa.nc")
q_pval_10 = xr.open_dataset(pval_filepath + "q_pval_10hpa.nc")
q_pval_500 = xr.open_dataset(pval_filepath + "q_pval_500hpa.nc")
q_pval_850 = xr.open_dataset(pval_filepath + "q_pval_850hpa.nc")
ns_pval_10 = xr.open_dataset(pval_filepath + "ns_pval_10hpa.nc")
ns_pval_500 = xr.open_dataset(pval_filepath + "ns_pval_500hpa.nc")
ns_pval_850 = xr.open_dataset(pval_filepath + "ns_pval_850hpa.nc")
###############################################
## Zonal mean Wind Profile Difference ##
###############################################
#Select Northern Hemisphere and DJF
p_uwind_z_c = p_file_c.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
f_uwind_z_c = f_file_c.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
p_uwind_z_da = p_file_da.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
f_uwind_z_da = f_file_da.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
p_uwind_z_dt = p_file_dt.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
f_uwind_z_dt = f_file_dt.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
p_uwind_z_ns = p_file_ns.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
f_uwind_z_ns = f_file_ns.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
p_uwind_z_q = p_file_q.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
f_uwind_z_q = f_file_q.ucomp.sel(lat=slice(0,90),time=slice('0001-12-01','0002-02-30')).mean('time')
c_pval_10 = c_pval_10.__xarray_dataarray_variable__
c_pval_500 = c_pval_500.pvals
c_pval_850 = c_pval_850.pvals
da_pval_10 = da_pval_10.pval
da_pval_500 = da_pval_500.pval
da_pval_850 = da_pval_850.pval
dt_pval_10 = dt_pval_10.pval
dt_pval_500 = dt_pval_500.pval
dt_pval_850 = dt_pval_850.pval
q_pval_10 = q_pval_10.pval
q_pval_500 = q_pval_500.pval
q_pval_850 = q_pval_850.pval
ns_pval_10 = ns_pval_10.pval
ns_pval_500 = ns_pval_500.pval
ns_pval_850 = ns_pval_850.pval
#Check shapes are the same
print(p_uwind_z_c.shape, f_uwind_z_c.shape,
p_uwind_z_da.shape, f_uwind_z_da.shape,
p_uwind_z_dt.shape, f_uwind_z_dt.shape,
p_uwind_z_ns.shape, f_uwind_z_ns.shape,
p_uwind_z_q.shape, f_uwind_z_q.shape)
#Difference of 3deg from present
diff_ucomp_c = f_uwind_z_c - p_uwind_z_c
diff_ucomp_da = f_uwind_z_da - p_uwind_z_da
diff_ucomp_dt = f_uwind_z_dt - p_uwind_z_dt
diff_ucomp_ns = f_uwind_z_ns - p_uwind_z_ns
diff_ucomp_q = f_uwind_z_q - p_uwind_z_q
#Difference of difference from control
diff_ucomp_c_da = diff_ucomp_da - diff_ucomp_c
diff_ucomp_c_dt = diff_ucomp_dt - diff_ucomp_c
diff_ucomp_c_ns = diff_ucomp_ns - diff_ucomp_c
diff_ucomp_c_q = diff_ucomp_q - diff_ucomp_c
#Add cyclic point to data
def add_cyclic(data):
lon = data.lon
lon_idx = data.dims.index('lon')
new, lons = add_cyclic_point(data.values,coord=lon, axis=lon_idx)
new_data = xr.DataArray(new,coords=[data.pfull,data.lat,lons],dims=('pfull','lat','lon'))
return new_data
diff_ucomp_c = add_cyclic(diff_ucomp_c)
diff_ucomp_c_da = add_cyclic(diff_ucomp_c_da)
diff_ucomp_c_dt = add_cyclic(diff_ucomp_c_dt)
diff_ucomp_c_ns = add_cyclic(diff_ucomp_c_ns)
diff_ucomp_c_q = add_cyclic(diff_ucomp_c_q)
diff_ucomp_da = add_cyclic(diff_ucomp_da)
diff_ucomp_dt = add_cyclic(diff_ucomp_dt)
diff_ucomp_ns = add_cyclic(diff_ucomp_ns)
diff_ucomp_q = add_cyclic(diff_ucomp_q)
#Add cyclic point for pvals
def add_cyclic(data):
lon = data.lon
lon_idx = data.dims.index('lon')
new, lons = add_cyclic_point(data.values,coord=lon, axis=lon_idx)
new_data = xr.DataArray(new,coords=[data.lat,lons],dims=('lat','lon'))
return new_data
c_pval_10 = add_cyclic(c_pval_10)
c_pval_500 = add_cyclic(c_pval_500)
c_pval_850 = add_cyclic(c_pval_850)
da_pval_10 = add_cyclic(da_pval_10)
da_pval_500 = add_cyclic(da_pval_500)
da_pval_850 = add_cyclic(da_pval_850)
dt_pval_10 = add_cyclic(dt_pval_10)
dt_pval_500 = add_cyclic(dt_pval_500)
dt_pval_850 = add_cyclic(dt_pval_850)
q_pval_10 = add_cyclic(q_pval_10)
q_pval_500 = add_cyclic(q_pval_500)
q_pval_850 = add_cyclic(q_pval_850)
ns_pval_10 = add_cyclic(ns_pval_10)
ns_pval_500 = add_cyclic(ns_pval_500)
ns_pval_850 = add_cyclic(ns_pval_850)
def store_sig_points(pval):
lon = list()
lat = list()
for i in range(0,pval.shape[0]):
for j in range(0,pval.shape[1]):
if pval[i,j] <= 0.05:
lon.append(pval[i,j].lon.values.item())
lat.append(pval[i,j].lat.values.item())
return lon,lat
da_10_pval_lon, da_10_pval_lat = store_sig_points(da_pval_10)
dt_10_pval_lon, dt_10_pval_lat = store_sig_points(dt_pval_10)
q_10_pval_lon, q_10_pval_lat = store_sig_points(q_pval_10)
ns_10_pval_lon, ns_10_pval_lat = store_sig_points(ns_pval_10)
da_500_pval_lon, da_500_pval_lat = store_sig_points(da_pval_500)
dt_500_pval_lon, dt_500_pval_lat = store_sig_points(dt_pval_500)
q_500_pval_lon, q_500_pval_lat = store_sig_points(q_pval_500)
ns_500_pval_lon, ns_500_pval_lat = store_sig_points(ns_pval_500)
da_850_pval_lon, da_850_pval_lat = store_sig_points(da_pval_850)
dt_850_pval_lon, dt_850_pval_lat = store_sig_points(dt_pval_850)
q_850_pval_lon, q_850_pval_lat = store_sig_points(q_pval_850)
ns_850_pval_lon, ns_850_pval_lat = store_sig_points(ns_pval_850)
#Masking where is not significant
#mask_c = diff_ucomp_c.sel(pfull=10,method='nearest').where(c_pval_10>0)
#mask_da = diff_ucomp_da.sel(pfull=10,method='nearest').where(da_pval_10>0)
#mask_new = mask_da - mask_c
#mask_new.plot.contourf(levels=21)
#Maps 3deg minus present
fig = plt.figure(figsize=(13,6))
ax1 = plt.subplot(3,5,1,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax2 = plt.subplot(3,5,2,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax3 = plt.subplot(3,5,3,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax4 = plt.subplot(3,5,4,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax5 = plt.subplot(3,5,5,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax6 = plt.subplot(3,5,6,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax7 = plt.subplot(3,5,7,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax8 = plt.subplot(3,5,8,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax9 = plt.subplot(3,5,9,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax10 = plt.subplot(3,5,10,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax11 = plt.subplot(3,5,11,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax12 = plt.subplot(3,5,12,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax13 = plt.subplot(3,5,13,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax14 = plt.subplot(3,5,14,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax15 = plt.subplot(3,5,15,projection=ccrs.Orthographic(central_longitude=-30, central_latitude=50))
ax1.coastlines()
ax2.coastlines()
ax3.coastlines()
ax4.coastlines()
ax5.coastlines()
ax6.coastlines()
ax7.coastlines()
ax8.coastlines()
ax9.coastlines()
ax10.coastlines()
ax11.coastlines()
ax12.coastlines()
ax13.coastlines()
ax14.coastlines()
ax15.coastlines()
ax1.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax2.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax3.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax4.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax5.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax6.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax7.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax8.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax9.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax10.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax11.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax12.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax13.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax14.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
ax15.set_extent((0, 500000, 20, 500000),crs=ccrs.PlateCarree())
c1 = diff_ucomp_c.sel(pfull=10,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax1,transform=ccrs.PlateCarree(),vmin=-15,vmax=15,cmap='RdBu_r')
c2 = diff_ucomp_c.sel(pfull=500,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax6,transform=ccrs.PlateCarree(),vmin=-5,vmax=5,cmap='RdBu_r')
c3 = diff_ucomp_c.sel(pfull=850,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax11,transform=ccrs.PlateCarree(),vmin=-4,vmax=4,cmap='RdBu_r')
c4 = diff_ucomp_c_da.sel(pfull=10,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax2,transform=ccrs.PlateCarree(),vmin=-7,vmax=7,cmap='RdBu_r')
c5 = diff_ucomp_c_da.sel(pfull=500,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax7,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
c6 = diff_ucomp_c_da.sel(pfull=850,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax12,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
diff_ucomp_c_q.sel(pfull=10,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax3,transform=ccrs.PlateCarree(),vmin=-7,vmax=7,cmap='RdBu_r')
diff_ucomp_c_q.sel(pfull=500,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax8,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
diff_ucomp_c_q.sel(pfull=850,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax13,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
diff_ucomp_c_dt.sel(pfull=10,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax4,transform=ccrs.PlateCarree(),vmin=-7,vmax=7,cmap='RdBu_r')
diff_ucomp_c_dt.sel(pfull=500,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax9,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
diff_ucomp_c_dt.sel(pfull=850,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax14,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
diff_ucomp_c_ns.sel(pfull=10,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax5,transform=ccrs.PlateCarree(),vmin=-7,vmax=7,cmap='RdBu_r')
diff_ucomp_c_ns.sel(pfull=500,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax10,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
diff_ucomp_c_ns.sel(pfull=850,method='nearest').plot.contourf(levels=21,add_colorbar=False,ax=ax15,transform=ccrs.PlateCarree(),vmin=-3,vmax=3,cmap='RdBu_r')
ax2.scatter(x=da_10_pval_lon,y=da_10_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax3.scatter(q_10_pval_lon,q_10_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax4.scatter(dt_10_pval_lon,dt_10_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax5.scatter(ns_10_pval_lon,ns_10_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax7.scatter(da_500_pval_lon,da_500_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax8.scatter(q_500_pval_lon,q_500_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax9.scatter(dt_500_pval_lon,dt_500_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax10.scatter(ns_500_pval_lon,ns_500_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax12.scatter(da_850_pval_lon,da_850_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax13.scatter(q_850_pval_lon,q_850_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax14.scatter(dt_850_pval_lon,dt_850_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
ax15.scatter(ns_850_pval_lon,ns_850_pval_lat,color='k',s=2,marker='.',transform=ccrs.PlateCarree())
c_pval_10.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax1,transform=ccrs.PlateCarree())
c_pval_500.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax6,transform=ccrs.PlateCarree())
c_pval_850.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax11,transform=ccrs.PlateCarree())
#da_pval_10.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax2,transform=ccrs.PlateCarree())
#da_pval_500.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax7,transform=ccrs.PlateCarree())
#da_pval_850.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax12,transform=ccrs.PlateCarree())
#
#q_pval_10.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax3,transform=ccrs.PlateCarree())
#q_pval_500.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax8,transform=ccrs.PlateCarree())
#q_pval_850.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax13,transform=ccrs.PlateCarree())
#
#dt_pval_10.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax4,transform=ccrs.PlateCarree())
#dt_pval_500.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax9,transform=ccrs.PlateCarree())
#dt_pval_850.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax14,transform=ccrs.PlateCarree())
#
#ns_pval_10.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax5,transform=ccrs.PlateCarree())
#ns_pval_500.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax10,transform=ccrs.PlateCarree())
#ns_pval_850.plot.contourf(colors='none',hatches=['...'],alpha=0,add_colorbar=False,levels=2,ax=ax15,transform=ccrs.PlateCarree())
ax1.set_title('Control',fontsize =14)
ax2.set_title('Double All - Control',fontsize =14)
ax3.set_title('Quadruple - Control',fontsize =14)
ax4.set_title('Double Tropo - Control',fontsize =14)
ax5.set_title('No Strat - Control',fontsize =14)
ax6.set_title('')
ax7.set_title('')
ax8.set_title('')
ax9.set_title('')
ax10.set_title('')
ax11.set_title('')
ax12.set_title('')
ax13.set_title('')
ax14.set_title('')
ax15.set_title('')
ax1.annotate('10hPa',xy=(0,0), xytext=(-125, 0),textcoords='offset points', fontsize=14,size='large')
ax6.annotate('500hPa',xy=(0,0), xytext=(-125, 0),textcoords='offset points', fontsize=14,size='large')
ax11.annotate('850hPa',xy=(0,0), xytext=(-125, 0),textcoords='offset points', fontsize=14,size='large')
p0 = ax1.get_position().get_points().flatten()
p1 = ax6.get_position().get_points().flatten()
p2 = ax11.get_position().get_points().flatten()
p3 = ax2.get_position().get_points().flatten()
p4 = ax7.get_position().get_points().flatten()
p5 = ax12.get_position().get_points().flatten()
p6 = ax5.get_position().get_points().flatten()
p7 = ax10.get_position().get_points().flatten()
p8 = ax15.get_position().get_points().flatten()
plt.tight_layout()
#Set colorbars add_axes([left,bottom,width,height])
cbar_ax = fig.add_axes([p0[0]-0.05,0.68,0.16,0.03])
cb1 = plt.colorbar(c1, orientation='horizontal',cax=cbar_ax,label='')
cbar_ax = fig.add_axes([p1[0]-0.05,0.35,0.16,0.03])
cb2 = plt.colorbar(c2, orientation='horizontal',cax=cbar_ax,label='')
cbar_ax = fig.add_axes([p2[0]-0.05,0,0.16,0.03])
cb3 = plt.colorbar(c3, orientation='horizontal',cax=cbar_ax,label='ms$^{-1}$').set_label(label='ms$^{-1}$',size=12)
cbar_ax = fig.add_axes([p3[0],0.68,0.7,0.03])#1, 0.15, 0.05, 0.7])
cb1 = plt.colorbar(c4, orientation='horizontal',cax=cbar_ax,label='')
cbar_ax = fig.add_axes([p4[0],0.35,0.7,0.03])
cb2 = plt.colorbar(c5, orientation='horizontal',cax=cbar_ax,label='')
cbar_ax = fig.add_axes([p5[0],0,0.7,0.03])
cb3 = plt.colorbar(c6, orientation='horizontal',cax=cbar_ax,label='ms$^{-1}$').set_label(label='ms$^{-1}$',size=12)
plt.savefig("Fig5/zonal_global_uwinds_3deg_minus_present_minus_control_NA_maps_djf.png",bbox_inches = "tight")
plt.show()