-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathobfits_diag_nc.py
278 lines (269 loc) · 12.6 KB
/
obfits_diag_nc.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
from netCDF4 import Dataset
import numpy as np
import os, sys, dateutils
msg = 'date1 date2 datapath runid hem outfile'
if len(sys.argv) < 6:
print(msg)
raise SystemExit
date1 = sys.argv[1] # date range
date2 = sys.argv[2]
datapath = sys.argv[3] # path to diag files.
runid = sys.argv[4] # suffix for diag file
hem = sys.argv[5] # NH,TR,SH,GL
outfile = sys.argv[6] # profile stats saved here
sondesonly = False # use only 120,132,220,221,232 (sondes,pibals,drops)
noair = False
aironly = False
latbound = 30
# if sondesonly False, aircraft, pibals and surface data included also
dates = dateutils.daterange(date1,date2,6)
#if '2019092712' in dates: dates.remove('2019092712')
deltap = 50.; pbot = 975
nlevs = 23
levs = np.zeros(nlevs, np.float)
levs1 = np.zeros(nlevs, np.float)
levs2 = np.zeros(nlevs, np.float)
levs[0:18] = pbot - deltap*np.arange(18)
levs1[0:18] = levs[0:18] + 0.5*deltap
levs2[0:18] = levs[0:18] - 0.5*deltap
levs1[18] = levs2[17]
levs2[18] = 70.; levs1[19] = 70.
levs2[19] = 50.; levs1[20] = 50.
levs2[20] = 30.; levs1[21] = 30.
levs2[21] = 10.; levs1[22] = 10.
levs2[22] = 0.
levs1[0] = 1200.
pbins = np.zeros(nlevs+1,np.float)
pbins[0:nlevs] = levs1; pbins[nlevs] = levs2[-1]
for nlev in range(18,nlevs):
levs[nlev] = 0.5*(levs1[nlev]+levs2[nlev])
rms_wind = np.zeros(len(levs),np.float)
rms_temp = np.zeros(len(levs),np.float)
rms_humid = np.zeros(len(levs),np.float)
bias_temp = np.zeros(len(levs),np.float)
bias_humid = np.zeros(len(levs),np.float)
count_temp = np.zeros(len(levs),np.int)
count_humid = np.zeros(len(levs),np.int)
count_wind = np.zeros(len(levs),np.int)
rms_wind_meantot = []
rms_temp_meantot = []
rms_humid_meantot = []
ncout = Dataset(outfile+'.nc','w')
plevs = ncout.createDimension('plevs',len(levs))
times = ncout.createDimension('time',len(dates))
plevs = ncout.createVariable('plevs',np.float32,'plevs')
plevs_up = ncout.createVariable('plevs_up',np.float32,'plevs')
plevs_down = ncout.createVariable('plevs_down',np.float32,'plevs')
plevs_up[:] = levs2
plevs_down[:] = levs1
plevs[:]=levs
plevs.units = 'hPa'
times = ncout.createVariable('times',np.float64,'time')
times.units = 'hours since 01-01-01'
omf_wnd = ncout.createVariable('omf_rmswind',np.float32, ('time','plevs'))
omf_temp = ncout.createVariable('omf_rmstemp',np.float32, ('time','plevs'))
omf_tempb = ncout.createVariable('omf_biastemp',np.float32, ('time','plevs'))
temp_obcounts = ncout.createVariable('temp_obcounts',np.int32, ('time','plevs'))
wind_obcounts = ncout.createVariable('wind_obcounts',np.int32, ('time','plevs'))
counts_humid=0
#datapath2='/scratch3/BMC/gsienkf/whitaker/t1534_t574'
ndate = 0
for date in dates:
times[ndate] = dateutils.datetohrs(date)
if runid != '':
obsfile_uv = os.path.join(datapath,'%s/diag_conv_uv_ges.%s_%s.nc4' % (date,date,runid))
obsfile_t = os.path.join(datapath,'%s/diag_conv_t_ges.%s_%s.nc4' % (date,date,runid))
obsfile_q = os.path.join(datapath,'%s/diag_conv_q_ges.%s_%s.nc4' % (date,date,runid))
else:
obsfile_uv = os.path.join(datapath,'%s/diag_conv_uv_ges.%s.nc4' % (date,date))
obsfile_t = os.path.join(datapath,'%s/diag_conv_t_ges.%s.nc4' % (date,date))
obsfile_q = os.path.join(datapath,'%s/diag_conv_q_ges.%s.nc4' % (date,date))
nc_uv = Dataset(obsfile_uv); nc_t = Dataset(obsfile_t); nc_q = Dataset(obsfile_q)
uv_code = nc_uv['Observation_Type'][:]
t_code = nc_t['Observation_Type'][:]
q_code = nc_q['Observation_Type'][:]
uv_used = nc_uv['Analysis_Use_Flag'][:]
t_used = nc_t['Analysis_Use_Flag'][:]
q_used = nc_q['Analysis_Use_Flag'][:]
uv_oberrinv = nc_uv['Errinv_Final'][:]
t_oberrinv = nc_t['Errinv_Final'][:]
q_oberrinv = nc_q['Errinv_Final'][:]
uv_press = nc_uv['Pressure'][:]
t_press = nc_t['Pressure'][:]
q_press = nc_q['Pressure'][:]
uv_lon = nc_uv['Longitude'][:]
t_lon = nc_t['Longitude'][:]
q_lon = nc_q['Longitude'][:]
uv_lat = nc_uv['Latitude'][:]
t_lat = nc_t['Latitude'][:]
q_lat = nc_q['Latitude'][:]
omf_u = nc_uv['u_Obs_Minus_Forecast_unadjusted'][:]
omf_v = nc_uv['v_Obs_Minus_Forecast_unadjusted'][:]
#omf_t = nc_t['Obs_Minus_Forecast_unadjusted'][:]
#omf_q = nc_q['Obs_Minus_Forecast_unadjusted'][:]
omf_t = nc_t['Obs_Minus_Forecast_adjusted'][:]
omf_q = nc_q['Obs_Minus_Forecast_adjusted'][:]
qsges = nc_q['Forecast_Saturation_Spec_Hum'][:]
if sondesonly:
insitu_wind = np.logical_or(uv_code == 220, # sondes
uv_code == 232) # drops
insitu_wind = np.logical_or(insitu_wind, uv_code == 221) # pibals
elif aironly:
insitu_wind = np.logical_and(uv_code >= 230, uv_code <= 235)
elif noair:
insitu_wind = np.logical_and(uv_code >= 280, uv_code <= 282) #sfc
# sones, pibals
insitu_wind = np.logical_or(insitu_wind,\
np.logical_or(uv_code == 220, uv_code == 221))
insitu_wind = np.logical_or(insitu_wind, uv_code == 232) # drops
else:
insitu_wind = np.logical_and(uv_code >= 280, uv_code <= 282) #sfc
# sones, pibals
insitu_wind = np.logical_or(insitu_wind,\
np.logical_or(uv_code == 220, uv_code == 221))
#print 'sondes,pibals',np.logical_and(uv_code >= 220, uv_code <= 221).sum()
# aircraft, drops
insitu_wind = np.logical_or(insitu_wind,\
np.logical_and(uv_code >= 230, uv_code <= 235))
#print 'aircraft,drops',np.logical_and(uv_code >= 230, uv_code <= 235).sum()
#print 'drops',(uv_code == 232).sum()
if sondesonly:
insitu_temp = np.logical_or(t_code == 120, # sondes
t_code == 132) # drops
insitu_q = np.logical_or(q_code == 120, # sondes
q_code == 132) # drops
elif aironly:
insitu_temp = np.logical_and(t_code >= 130, t_code <= 135)
insitu_q = np.logical_and(q_code >= 130, q_code <= 135)
elif noair:
insitu_temp = np.logical_or(t_code == 120, # sondes
t_code == 132) # drops
insitu_q = np.logical_or(q_code == 120, # sondes
q_code == 132) # drops
insitu_temp = np.logical_or(insitu_temp,np.logical_and(t_code >= 180, t_code <= 182)) #sfc
insitu_q = np.logical_or(insitu_q,np.logical_and(q_code >= 180, q_code <= 182)) #sfc
else:
insitu_temp = np.logical_and(t_code >= 180, t_code <= 182) #sfc
insitu_temp = np.logical_or(insitu_temp, t_code==120) # sondes
# aircraft, drops
insitu_temp = np.logical_or(insitu_temp,\
np.logical_and(t_code >= 130, t_code <= 135))
insitu_q = np.logical_and(q_code >= 180, q_code <= 182) #sfc
insitu_q = np.logical_or(insitu_q, q_code==120) # sondes
# aircraft, drops
insitu_q = np.logical_or(insitu_q,\
np.logical_and(q_code >= 130, q_code <= 135))
# consider this of if used flag is 1, inverse oberr is < 1.e-5 and a valid pressure level is included
#uv_used = np.logical_and(uv_used == 1, uv_oberrinv > 1.e-5)
uv_used = uv_used == 1
uv_used = np.logical_and(uv_used, np.isfinite(uv_press))
insitu_wind = np.logical_and(insitu_wind,uv_used)
#t_used = np.logical_and(t_used == 1, t_oberrinv > 1.e-5)
t_used = t_used == 1
t_used = np.logical_and(t_used, np.isfinite(t_press))
insitu_temp = np.logical_and(insitu_temp,t_used)
#q_used = np.logical_and(q_used == 1, q_oberrinv > 1.e-5)
q_used = q_used == 1
q_used = np.logical_and(q_used, np.isfinite(q_press))
insitu_q = np.logical_and(insitu_q,q_used)
if hem == 'NH':
uv_latcond = uv_lat > latbound
t_latcond = t_lat > latbound
q_latcond = q_lat > latbound
elif hem == 'SH':
uv_latcond = uv_lat < -latbound
t_latcond = t_lat < -latbound
q_latcond = q_lat < -latbound
elif hem == 'TR':
uv_latcond = np.logical_and(uv_lat <= latbound,uv_lat >= -latbound)
t_latcond = np.logical_and(t_lat <= latbound,t_lat >= -latbound)
q_latcond = np.logical_and(q_lat <= latbound,q_lat >= -latbound)
if hem in ['NH','TR','SH']:
indxuv = np.logical_and(insitu_wind,uv_latcond)
indxt = np.logical_and(insitu_temp,t_latcond)
indxq = np.logical_and(insitu_q,q_latcond)
else:
indxuv = insitu_wind; indxt = insitu_temp; indxq = insitu_q
omf_u = omf_u[indxuv]
omf_v = omf_v[indxuv]
omf_t = omf_t[indxt]
omf_q = 100.*omf_q[indxq]*qsges[indxq] # convert to g/kg?
press_u = uv_press[indxuv]
press_t = t_press[indxt]
press_q = q_press[indxq]
# compute innovation stats for temperature.
pindx = np.digitize(press_t,pbins)-1
# check on pindx calculation
#for n in range(len(press_t)):
# ip = pindx[n]
# p = press_t[n]
# if not (p < levs1[ip] and p >= levs2[ip]):
# print p, levs2[ip], levs1[ip]
# raise IndexError('wind p mismatch')
rms_temp += np.bincount(pindx,minlength=nlevs,weights=omf_t**2)
counts, bin_edges = np.histogram(press_t,pbins[::-1])
omf_tempb[ndate] = np.bincount(pindx,minlength=nlevs,weights=omf_t)/counts[::-1]
omf_temp[ndate] = np.bincount(pindx,minlength=nlevs,weights=omf_t**2)/counts[::-1]
temp_obcounts[ndate] = counts[::-1]
bias_temp += np.bincount(pindx,minlength=nlevs,weights=omf_t)
nobs_temp = np.where(counts[::-1] == -1, 0, counts[::-1]).sum()
count_temp += counts[::-1]
rms_temp_mean = np.sqrt(np.bincount(pindx,minlength=nlevs,weights=omf_t**2)/counts[::-1])[0:18].mean()
# compute innovation stats for humidity.
if sum(indxq) > 0:
pindx = np.digitize(press_q,pbins)-1
# check on pindx calculation
#for n in range(len(press_q)):
# ip = pindx[n]
# p = press_q[n]
# if not (p < levs1[ip] and p >= levs2[ip]):
# print p, levs2[ip], levs1[ip]
# raise IndexError('wind p mismatch')
rms_humid += np.bincount(pindx,minlength=nlevs,weights=omf_q**2)
bias_humid += np.bincount(pindx,minlength=nlevs,weights=omf_q)
counts, bin_edges = np.histogram(press_q,pbins[::-1])
counts = np.where(counts == 0, -1, counts)
nobs_humid = np.where(counts[::-1] == -1, 0, counts[::-1]).sum()
count_humid += counts[::-1]
rms_humid_mean = np.sqrt(np.bincount(pindx,minlength=nlevs,weights=omf_q**2)/counts[::-1])[0:18].mean()
else:
counts_humid += 0
rms_humid_mean = np.zeros(rms_temp_mean.shape, rms_temp_mean.dtype)
# compute innovation stats for wind.
pindx = np.digitize(press_u,pbins)-1
# check on pindx calculation
#for n in range(len(press_u)):
# ip = pindx[n]
# p = press_u[n]
# if not (p < levs1[ip] and p >= levs2[ip]):
# print p, levs2[ip], levs1[ip]
# raise IndexError('wind p mismatch')
rms_wind += np.bincount(pindx,minlength=nlevs,weights=np.sqrt(omf_u**2+omf_v**2))
counts, bin_edges = np.histogram(press_u,pbins[::-1])
omf_wnd[ndate] = np.bincount(pindx,minlength=nlevs,weights=np.sqrt(omf_u**2+omf_v**2))/counts[::-1]
wind_obcounts[ndate] = counts[::-1]
nobs_uv = np.where(counts[::-1] == -1, 0, counts[::-1]).sum()
count_wind += counts[::-1]
rms_wind_mean = (np.bincount(pindx,minlength=nlevs,weights=np.sqrt(omf_u**2+omf_v**2))/counts[::-1])[0:18].mean()
rms_wind_meantot.append(rms_wind_mean)
rms_temp_meantot.append(rms_temp_mean)
rms_humid_meantot.append(rms_humid_mean)
print(date, nobs_uv, rms_wind_mean, nobs_temp, rms_temp_mean, nobs_humid, rms_humid_mean)
ndate += 1
ncout.close()
rms_wind = rms_wind/count_wind
rms_temp = np.sqrt(rms_temp/count_temp)
bias_temp = bias_temp/count_temp
rms_humid = np.sqrt(rms_humid/count_humid)
bias_humid = bias_humid/count_humid
rms_wind_mean = np.nanmean(np.array(rms_wind_meantot))
rms_temp_mean = np.nanmean(np.array(rms_temp_meantot))
rms_humid_mean = np.nanmean(np.array(rms_humid_meantot))
fout = open(outfile,'w')
fout.write('# %s %s %s %s-%s\n' % (datapath,runid,hem,date1,date2))
fout.write('# press wind_count wind_rms temp_count temp_rms temp_bias humid_rmsx1000 humid_biasx1000\n')
fout.write('# 1000-0 %10i %7.4f %10i %7.4f %7.4f %10i %7.4f %7.4f\n' % (count_wind.sum(), rms_wind.mean(), count_temp.sum(), rms_temp.mean(), bias_temp.mean(), count_humid.sum(), 1000*rms_humid.mean(), 1000*bias_humid.mean()))
for n,p in enumerate(levs):
fout.write('%8.2f %10i %7.4f %10i %7.4f %7.4f %10i %7.4f %7.4f\n' % (p,count_wind[n],rms_wind[n],count_temp[n],rms_temp[n],bias_temp[n],count_humid[n],1000*rms_humid[n],1000*bias_humid[n]))
fout.write('# %7.4f %7.4f %7.4f\n' % (rms_wind_mean,rms_temp_mean,1000.*rms_humid_mean))
fout.close()