-
Notifications
You must be signed in to change notification settings - Fork 15
/
analyze_mcmc.py
382 lines (329 loc) · 16.8 KB
/
analyze_mcmc.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
""" Analyze simulation output - mass change, runoff, etc. """
# Built-in libraries
from collections import OrderedDict
import datetime
import glob
import os
import pickle
# External libraries
import cartopy
import matplotlib as mpl
import matplotlib.pyplot as plt
from matplotlib.pyplot import MaxNLocator
from matplotlib.lines import Line2D
import matplotlib.patches as mpatches
from matplotlib.ticker import MultipleLocator
from matplotlib.ticker import EngFormatter
from matplotlib.ticker import StrMethodFormatter
from mpl_toolkits.axes_grid1.axes_divider import make_axes_locatable
import numpy as np
import pandas as pd
from scipy.stats import median_abs_deviation
from scipy.stats import linregress
from scipy.ndimage import uniform_filter
import scipy
import xarray as xr
# Local libraries
#import class_climate
#import class_mbdata
import pygem.pygem_input as pygem_prms
#import pygemfxns_gcmbiasadj as gcmbiasadj
import pygem.pygem_modelsetup as modelsetup
# Script options
option_compare_mcmc_emulator_fullsim = True
#%% ===== COMPARE MCMC FROM EMULATOR AND FULL SIMULATIONS =====
if option_compare_mcmc_emulator_fullsim:
overwrite = False
cal_fp = '/Users/drounce/Documents/HiMAT/Output/calibration-fullsim/'
plot_count_as_percent = False
glac_nos_pkl_fn = cal_fp + 'cal_fullsim_glac_nos.pkl'
if not os.path.exists(glac_nos_pkl_fn) or overwrite:
glac_nos = []
for fp in os.listdir(cal_fp):
if not fp.startswith('.'):
for i in os.listdir(cal_fp + fp):
if i.endswith('-modelprms_dict.pkl'):
glac_nos.append(i.split('-')[0])
glac_nos = sorted(glac_nos)
# Limit to 2500
nglaciers = 2500
if len(glac_nos) > nglaciers:
# Random indices
rand_idxs_all = np.arange(len(glac_nos))
np.random.shuffle(rand_idxs_all)
rand_idxs = rand_idxs_all[:nglaciers]
glac_nos = [glac_nos[x] for x in rand_idxs]
glac_nos = sorted(glac_nos)
# Save for reproducability
with open(glac_nos_pkl_fn, 'wb') as f:
pickle.dump(glac_nos, f)
else:
with open(glac_nos_pkl_fn, 'rb') as f:
glac_nos = pickle.load(f)
main_glac_rgi = modelsetup.selectglaciersrgitable(glac_no=glac_nos)
mb_obs_err = []
emulator_mb_med = []
emulator_mb_mad = []
fullsim_mb_med = []
fullsim_mb_mad = []
emulator_tbias_med = []
emulator_tbias_mad = []
fullsim_tbias_med = []
fullsim_tbias_mad = []
emulator_kp_med = []
emulator_kp_mad = []
fullsim_kp_med = []
fullsim_kp_mad = []
emulator_ddfsnow_med = []
emulator_ddfsnow_mad = []
fullsim_ddfsnow_med = []
fullsim_ddfsnow_mad = []
for glac_no in glac_nos:
modelprms_fn = glac_no + '-modelprms_dict.pkl'
modelprms_fp = cal_fp + glac_no.split('.')[0].zfill(2) + '/'
with open(modelprms_fp + modelprms_fn, 'rb') as f:
modelprms_dict = pickle.load(f)
mb_obs_err.append(modelprms_dict['emulator']['mb_obs_mwea_err'][0])
emulator_mb_med.append(np.median(modelprms_dict['MCMC']['mb_mwea']['chain_0']))
emulator_mb_mad.append(median_abs_deviation(modelprms_dict['MCMC']['mb_mwea']['chain_0']))
fullsim_mb_med.append(np.median(modelprms_dict['MCMC_fullsim']['mb_mwea']['chain_0']))
fullsim_mb_mad.append(median_abs_deviation(modelprms_dict['MCMC_fullsim']['mb_mwea']['chain_0']))
emulator_tbias_med.append(np.median(modelprms_dict['MCMC']['tbias']['chain_0']))
emulator_tbias_mad.append(median_abs_deviation(modelprms_dict['MCMC']['tbias']['chain_0']))
fullsim_tbias_med.append(np.median(modelprms_dict['MCMC_fullsim']['tbias']['chain_0']))
fullsim_tbias_mad.append(median_abs_deviation(modelprms_dict['MCMC_fullsim']['tbias']['chain_0']))
emulator_kp_med.append(np.median(modelprms_dict['MCMC']['kp']['chain_0']))
emulator_kp_mad.append(median_abs_deviation(modelprms_dict['MCMC']['kp']['chain_0']))
fullsim_kp_med.append(np.median(modelprms_dict['MCMC_fullsim']['kp']['chain_0']))
fullsim_kp_mad.append(median_abs_deviation(modelprms_dict['MCMC_fullsim']['kp']['chain_0']))
emulator_ddfsnow_med.append(np.median(modelprms_dict['MCMC']['ddfsnow']['chain_0']))
emulator_ddfsnow_mad.append(median_abs_deviation(modelprms_dict['MCMC']['ddfsnow']['chain_0']))
fullsim_ddfsnow_med.append(np.median(modelprms_dict['MCMC_fullsim']['ddfsnow']['chain_0']))
fullsim_ddfsnow_mad.append(median_abs_deviation(modelprms_dict['MCMC_fullsim']['ddfsnow']['chain_0']))
mb_obs_err = np.array(mb_obs_err)
emulator_mb_med = np.array(emulator_mb_med)
emulator_mb_mad = np.array(emulator_mb_mad)
fullsim_mb_med = np.array(fullsim_mb_med)
fullsim_mb_mad = np.array(fullsim_mb_mad)
emulator_tbias_med = np.array(emulator_tbias_med)
emulator_tbias_mad = np.array(emulator_tbias_mad)
fullsim_tbias_med = np.array(fullsim_tbias_med)
fullsim_tbias_mad = np.array(fullsim_tbias_mad)
emulator_kp_med = np.array(emulator_kp_med)
emulator_kp_mad = np.array(emulator_kp_mad)
fullsim_kp_med = np.array(fullsim_kp_med)
fullsim_kp_mad = np.array(fullsim_kp_mad)
emulator_ddfsnow_med = np.array(emulator_ddfsnow_med)
emulator_ddfsnow_mad = np.array(emulator_ddfsnow_mad)
fullsim_ddfsnow_med = np.array(fullsim_ddfsnow_med)
fullsim_ddfsnow_mad = np.array(fullsim_ddfsnow_mad)
dif_mb_med = emulator_mb_med - fullsim_mb_med
dif_mb_mad = emulator_mb_mad - fullsim_mb_mad
dif_mb_med_norm = (emulator_mb_med - fullsim_mb_med) / mb_obs_err
dif_tbias_med = emulator_tbias_med - fullsim_tbias_med
dif_tbias_mad = emulator_tbias_mad - fullsim_tbias_mad
dif_kp_med = emulator_kp_med - fullsim_kp_med
dif_kp_mad = emulator_kp_mad - fullsim_kp_mad
dif_ddfsnow_med = emulator_ddfsnow_med - fullsim_ddfsnow_med
dif_ddfsnow_mad = emulator_ddfsnow_mad - fullsim_ddfsnow_mad
#%%
# ===== PRIOR VS POSTERIOR FOR EACH GLACIER =====
print('abs max dif mb_med:', np.max(np.absolute(dif_mb_med)))
print('abs max dif mb_mad:', np.max(np.absolute(dif_mb_mad)))
print('abs max dif mb_med_norm:', np.max(np.absolute(dif_mb_med_norm)))
print('abs max dif tbias_med:', np.max(np.absolute(dif_tbias_med)))
print('abs max dif tbias_mad:', np.max(np.absolute(dif_tbias_mad)))
print('abs max dif kp_med:', np.max(np.absolute(dif_kp_med)))
print('abs max dif kp_mad:', np.max(np.absolute(dif_kp_mad)))
print('abs max dif ddfsnow_med:', np.max(np.absolute(dif_ddfsnow_med)))
print('abs max dif ddfsnow_mad:', np.max(np.absolute(dif_ddfsnow_mad)))
# Bin spacing (note: offset them, so centered on 0)
bdict = {}
bdict['mb_mwea-Median'] = np.arange(-0.25, 0.26, 0.01) - 0.005
bdict['tbias-Median'] = np.arange(-1, 1.025, 0.05) - 0.025
bdict['kp-Median'] = np.arange(-1, 1.025, 0.05) - 0.025
bdict['ddfsnow-Median'] = np.arange(-1, 1.025, 0.05) - 0.025
bdict['mb_mwea-Median Absolute Deviation'] = np.arange(-0.25, 0.26, 0.01) - 0.005
bdict['tbias-Median Absolute Deviation'] = np.arange(-1, 1.025, 0.05) - 0.025
bdict['kp-Median Absolute Deviation'] = np.arange(-1, 1.025, 0.05) - 0.025
bdict['ddfsnow-Median Absolute Deviation'] = np.arange(-1, 1.025, 0.05) - 0.025
vn_label_dict = {'mb_mwea':'Mass Balance (m w.e. $\mathregular{yr^{-1}}$)',
'kp':'Precipitation Factor (-)',
'tbias':'Temperature Bias ($\mathregular{^{\circ}C}$)',
'ddfsnow':'f$_{snow}$ (mm w.e. $\mathregular{d^{-1}}$ $\mathregular{^{\circ}C^{-1}}$)'}
variables = ['mb_mwea', 'tbias', 'kp', 'ddfsnow']
estimators = ['Median', 'Median Absolute Deviation']
fig, ax = plt.subplots(len(variables), len(estimators), squeeze=False, sharex=False, sharey=False,
gridspec_kw = {'wspace':0.1, 'hspace':0.4})
letters = ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H']
ncount = 0
for nvar, vn in enumerate(variables):
print(nvar, vn)
if vn == 'mb_mwea':
em_med = emulator_mb_med
em_mad = emulator_mb_mad
fullsim_med = fullsim_mb_med
fullsim_mad = fullsim_mb_mad
elif vn == 'tbias':
em_med = emulator_tbias_med
em_mad = emulator_tbias_mad
fullsim_med = fullsim_tbias_med
fullsim_mad = fullsim_tbias_mad
elif vn == 'kp':
em_med = emulator_kp_med
em_mad = emulator_kp_mad
fullsim_med = fullsim_kp_med
fullsim_mad = fullsim_kp_mad
elif vn == 'ddfsnow':
em_med = emulator_ddfsnow_med * 1e3
em_mad = emulator_ddfsnow_mad * 1e3
fullsim_med = fullsim_ddfsnow_med * 1e3
fullsim_mad = fullsim_ddfsnow_mad * 1e3
dif_med = em_med - fullsim_med
dif_mad = em_mad - fullsim_mad
print(' dif_mean (min/max):', np.round(dif_med.min(),2), np.round(dif_med.max(),2))
print(' dif_std (min/max):', np.round(dif_mad.min(),2), np.round(dif_mad.max(),2))
for nest, estimator in enumerate(estimators):
if estimator == 'Median':
dif = dif_med
bcolor = 'lightgrey'
elif estimator == 'Median Absolute Deviation':
dif = dif_mad
bcolor = 'lightgrey'
# ===== Plot =====
# hist, bins = np.histogram(dif)
hist, bins = np.histogram(dif, bins=bdict[vn + '-' + estimator])
if plot_count_as_percent:
hist = hist * 100.0 / hist.sum()
y_label = 'Count (%)'
tdict = {}
glac_ylim = 40
tdict['Median'] = np.arange(0, glac_ylim + 1, 10)
tdict['Median Absolute Deviation'] = np.arange(0, glac_ylim + 1, 10)
else:
y_label = 'Count'
tdict = {}
glac_ylim = 750
tdict['Median'] = np.arange(0, glac_ylim + 1, 100)
tdict['Median Absolute Deviation'] = np.arange(0, glac_ylim + 1, 100)
bins_centered = bins[1:] + (bins[0] - bins[1]) / 2
# plot histogram
ax[nvar,nest].bar(x=bins_centered, height=hist, width=(bins[1]-bins[0]), align='center',
edgecolor='black', color=bcolor, alpha=0.5)
ax[nvar,nest].set_yticks(tdict[estimator])
ax[nvar,nest].set_ylim(0,glac_ylim)
# axis labels
ax[nvar,nest].set_xlabel(vn_label_dict[vn], fontsize=10, labelpad=1)
if nvar == 0:
ax[nvar,nest].set_title('$\Delta$ ' + estimator, fontsize=12)
if nest == 1:
ax[nvar,nest].set_yticks([])
print(' ', estimator, '% near 0:', np.round(hist[np.where(bins > 0)[0][0] - 1]))
letter = letters[ncount]
ncount += 1
ax[nvar,nest].text(0.98, 0.98, letter, size=10, horizontalalignment='right',
verticalalignment='top', transform=ax[nvar,nest].transAxes, weight='bold')
# Save figure
fig.set_size_inches(6.5,8)
figure_fn = 'mcmc_emulator_vs_fullsim_hist.png'
figure_fp = cal_fp + '../figures/'
if not os.path.exists(figure_fp):
os.makedirs(figure_fp)
fig.savefig(figure_fp + figure_fn, bbox_inches='tight', dpi=300)
#%%
fig, ax = plt.subplots(len(variables), len(estimators), squeeze=False, sharex=False, sharey=False,
gridspec_kw = {'wspace':0.3, 'hspace':0.4})
vn_label_dict = {'mb_mwea':'Mass Balance\n(m w.e. $\mathregular{yr^{-1}}$)',
'kp':'Precipitation Factor\n(-)',
'tbias':'Temperature Bias\n($\mathregular{^{\circ}C}$)',
'ddfsnow':'f$_{snow}$\n(mm w.e. $\mathregular{d^{-1}}$ $\mathregular{^{\circ}C^{-1}}$)'}
ncount = 0
for nvar, vn in enumerate(variables):
print(nvar, vn)
if vn == 'mb_mwea':
em_med = emulator_mb_med
em_mad = emulator_mb_mad
fullsim_med = fullsim_mb_med
fullsim_mad = fullsim_mb_mad
ymin = -0.8
ymax = 0.8
ytick_major = 0.2
ytick_minor = 0.1
elif vn == 'tbias':
em_med = emulator_tbias_med
em_mad = emulator_tbias_mad
fullsim_med = fullsim_tbias_med
fullsim_mad = fullsim_tbias_mad
ymin = -6
ymax = 6
ytick_major = 2
ytick_minor = 1
elif vn == 'kp':
em_med = emulator_kp_med
em_mad = emulator_kp_mad
fullsim_med = fullsim_kp_med
fullsim_mad = fullsim_kp_mad
ymin = -2.5
ymax = 2.5
ytick_major = 1
ytick_minor = 0.5
elif vn == 'ddfsnow':
em_med = emulator_ddfsnow_med * 1e3
em_mad = emulator_ddfsnow_mad * 1e3
fullsim_med = fullsim_ddfsnow_med * 1e3
fullsim_mad = fullsim_ddfsnow_mad * 1e3
ymin = -2.5
ymax = 2.5
ytick_major = 1
ytick_minor = 0.5
dif_med = em_med - fullsim_med
dif_mad = em_mad - fullsim_mad
print(' dif_mean (min/max):', np.round(dif_med.min(),2), np.round(dif_med.max(),2))
print(' dif_std (min/max):', np.round(dif_mad.min(),2), np.round(dif_mad.max(),2))
for nest, estimator in enumerate(estimators):
if estimator == 'Median':
dif = dif_med
elif estimator == 'Median Absolute Deviation':
dif = dif_mad
# ===== Plot =====
max_size = 300
# plot histogram
ax[nvar,nest].scatter(main_glac_rgi.Area.values, dif, s=2, color='k', marker='o')
ax[nvar,nest].hlines(0, 0, main_glac_rgi.Area.max(), color='k', lw=0.5)
ax[nvar,nest].set_xlim(0,max_size)
ax[nvar,nest].set_ylim(ymin, ymax)
ax[nvar,nest].yaxis.set_major_locator(MultipleLocator(ytick_major))
ax[nvar,nest].yaxis.set_minor_locator(MultipleLocator(ytick_minor))
ax[nvar,nest].tick_params(direction='inout', right=False)
ax[nvar,nest].xaxis.set_major_locator(MultipleLocator(100))
ax[nvar,nest].xaxis.set_minor_locator(MultipleLocator(20))
ax[nvar,nest].tick_params(direction='inout', right=False)
# axis labels
if nest == 0:
ax[nvar,nest].set_ylabel(vn_label_dict[vn], fontsize=10, labelpad=1)
if nvar == 3:
ax[nvar,nest].set_xlabel('Area (km$^{2}$)')
if nvar == 0:
ax[nvar,nest].set_title('$\Delta$ ' + estimator, fontsize=12)
letter = letters[ncount]
ncount += 1
ax[nvar,nest].text(0.98, 0.98, letter, size=10, horizontalalignment='right',
verticalalignment='top', transform=ax[nvar,nest].transAxes, weight='bold')
# Save figure
fig.set_size_inches(6.5,8)
figure_fn = 'mcmc_emulator_vs_fullsim_scatter_vs_area_' + str(max_size) + 'km2.png'
fig.savefig(figure_fp + figure_fn, bbox_inches='tight', dpi=300)
#%%
print('95% mb_mwea_norm:', np.round(np.percentile(dif_mb_med_norm,2.5),2), np.round(np.percentile(dif_mb_med_norm,97.5),2))
print('95% mb_mwea:', np.round(np.percentile(dif_mb_med,2.5),3), np.round(np.percentile(dif_mb_med,97.5),3))
print('95% tbias:', np.round(np.percentile(dif_tbias_med,2.5),3), np.round(np.percentile(dif_tbias_med,97.5),3))
print('95% kp:', np.round(np.percentile(dif_kp_med,2.5),3), np.round(np.percentile(dif_kp_med,97.5),3))
print('95% ddfsnow:', np.round(np.percentile(dif_ddfsnow_med,2.5),5), np.round(np.percentile(dif_ddfsnow_med,97.5),5))
print('normalized dif mb_med by mb_obs_err:')
print('min/max:', np.round(np.min(dif_mb_med_norm),3), np.round(np.max(dif_mb_med_norm),3))
#%%
#import pickle
##fn = '/Users/drounce/Documents/HiMAT/PyGEM/Cal_fullsim_batch_0.pkl'
#fn = '/Users/drounce/Documents/HiMAT/Output/calibration-fullsim/01/1.00058-modelprms_dict.pkl'
#with open(fn, 'rb') as f:
# A = pickle.load(f)