-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathmodel_DALEC.py
302 lines (277 loc) · 17.3 KB
/
model_DALEC.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
import numpy
import time
#from sklearn.neural_network import MLPRegressor
#import pickle
class MyModel(object):
def __init__(self):
self.name = 'DALEC'
self.npfts = 4
self.parms = {'gdd_min': numpy.zeros([self.npfts], numpy.float)+20., \
'ndays_on': numpy.zeros([self.npfts], numpy.float)+30., \
'tsmin': numpy.zeros([self.npfts], numpy.float)-2.0, \
'leaffall': numpy.zeros([self.npfts], numpy.float)+0.3, \
'flnr': numpy.zeros([self.npfts], numpy.float)+0.08, \
'nue': numpy.zeros([self.npfts], numpy.float)+15.0, \
'lma': numpy.zeros([self.npfts], numpy.float)+80.0, \
'leafcn': numpy.zeros([self.npfts], numpy.float)+25.0, \
'rg_frac': numpy.zeros([self.npfts], numpy.float)+0.3, \
'aleaf': numpy.zeros([self.npfts], numpy.float)+0.30, \
'astem': numpy.zeros([self.npfts], numpy.float)+0.50, \
'troot': numpy.zeros([self.npfts], numpy.float)+5.0, \
'tstem': numpy.zeros([self.npfts], numpy.float)+50.0, \
'evergreen': numpy.zeros([self.npfts], numpy.int)+0, \
'br_mr': numpy.zeros([1], numpy.float)+2.52e-6, \
'q10_mr': numpy.zeros([1], numpy.float)+2.0, \
'q10_hr': numpy.zeros([1], numpy.float)+2.0, \
'br_lit': numpy.zeros([1], numpy.float)+3.0, \
'br_som': numpy.zeros([1], numpy.float)+500.0, \
'dr': numpy.zeros([1], numpy.float)+0.001}
self.pdef = self.parms.copy()
self.statevars = ['leafc_stor_pft', 'xsmr_pft', 'gdd_pft', 'stemc_pft','rootc_pft','litrc','somc','leafc_pft','lai_pft']
self.fluxvars = ['mr_pft', 'gr_pft', 'gpp_pft', 'npp_pft', 'hr', 'nee', 'nee_lc_light']
#SPRUCE-specific parameter values
self.parms['evergreen'][0] = 1.0 #Black Spruce
self.parms['evergreen'][3] = 1.0 #Moss
self.parms['lma'][0] = 142.
self.parms['lma'][1] = 35.
self.parms['lma'][2] = 50.
self.parms['lma'][3] = 140.
#self.parms['tstem'][2] = 5.0
self.parms['tstem'][3] = 1.0
self.parms['troot'][3] = 1.0
self.parms['astem'][3] = 0.05
#self.parms['aleaf'][2] = 0.7
self.parms['aleaf'][3] = 0.7
#self.parms['flnr'][0] = 0.08
#self.parms['flnr'][1] = 0.15
#self.parms['flnr'][2] = 0.15
#self.parms['flnr'][3] = 0.15
#get neural network
#pkl_filename = './GPP_model_NN/bestmodel_daily.pkl'
#with open(pkl_filename, 'rb') as file:
# self.nnmodel = pickle.load(file)
#nsamples=20000
#self.nparms_nn = 14 #15
#ptrain_orig = (numpy.loadtxt('./GPP_model_NN/ptrain_daily.dat'))[0:nsamples,:]
#self.pmin_nn = numpy.zeros([self.nparms_nn], numpy.float)
#self.pmax_nn = numpy.zeros([self.nparms_nn], numpy.float)
#for i in range(0,self.nparms_nn):
# self.pmin_nn[i] = min(ptrain_orig[:,i])
# self.pmax_nn[i] = max(ptrain_orig[:,i])
def run_dalec(self, parms, ad_cycles=0, final_cycles=0, trans_startyear=-1, pftwt=-1):
self.states={}
self.fluxes={}
if (pftwt == -1):
pftwt = numpy.ones([self.npfts], numpy.float)/self.npfts
#determine number of transient cycles
if trans_startyear < 0:
trans_startyear = self.start_year
nyr_cycle = (self.end_year - self.start_year)+1
trans_cycles = int(numpy.ceil((self.end_year - trans_startyear+1)/float(nyr_cycle)))
self.model_startyear = self.end_year-nyr_cycle*trans_cycles+1
for v in self.statevars:
if ('_pft' in v):
self.states[v] = numpy.zeros([self.npfts, int(self.nobs*trans_cycles)+1], numpy.float)
else:
self.states[v] = numpy.zeros([int(self.nobs*trans_cycles)+1], numpy.float)
for v in self.fluxvars:
if ('_pft' in v):
self.fluxes[v] = numpy.zeros([self.npfts, int(self.nobs*trans_cycles)+1], numpy.float)
else:
self.fluxes[v] = numpy.zeros([int(self.nobs*trans_cycles)+1], numpy.float)
#initial values
for p in range(0,self.npfts):
if (self.parms['evergreen'][p] == 0):
self.states['leafc_stor_pft'][p,0] = 300.0
else:
self.states['leafc_pft'][p,0] = 100.0
#Model parameters
gdd_min = self.parms['gdd_min']
ndays_on = self.parms['ndays_on']
tsmin = self.parms['tsmin']
leaffall= self.parms['leaffall']
nue = self.parms['nue']
rg_frac = self.parms['rg_frac']
br_mr = self.parms['br_mr']
q10_mr = self.parms['q10_mr']
aleaf = self.parms['aleaf']
astem = self.parms['astem']
troot = 1/(self.parms['troot']*365.0)
tstem_base = 1/(self.parms['tstem']*365.0)
tstem = tstem_base.copy()
q10_hr = self.parms['q10_hr']
br_lit = 1/(self.parms['br_lit']*365.0)
br_som_base = 1/(self.parms['br_som']*365.0)
dr = self.parms['dr']
lma = self.parms['lma']
leafcn = self.parms['leafcn']
#met_thistimestep_norm=numpy.zeros([1,self.nparms_nn], numpy.float)
#Run the model
ad_factor_stem = numpy.zeros([self.npfts], numpy.float)
ad_factor_som = max(self.parms['br_som']/5, 1.0)
for p in range(0,self.npfts):
ad_factor_stem[p] = max(self.parms['tstem'][p]/5, 1.0)
for s in range(0,ad_cycles+final_cycles+trans_cycles):
if (s > 0 and s <= ad_cycles+final_cycles):
for var in self.states:
if ('_pft' in var):
for p in range(0,self.npfts):
self.states[var][p,0] = self.states[var][p,self.nobs]
else:
self.states[var][0] = self.states[var][self.nobs]
if (s < ad_cycles):
br_som = br_som_base*ad_factor_som
for p in range(0,self.npfts):
tstem[p] = tstem_base[p]*ad_factor_stem[p]
elif (s == ad_cycles):
br_som = br_som_base
for p in range(0,self.npfts):
tstem[p] = tstem_base[p]
if (ad_cycles > 0):
for p in range(0,self.npfts):
self.states['stemc_pft'][p,0] = sum(self.states['stemc_pft'][p,0:self.nobs])/self.nobs*ad_factor_stem[p]
self.states['somc'][0] = sum(self.states['somc'][0:self.nobs])/self.nobs*ad_factor_som
this_trans_cycle = s - ad_cycles-final_cycles
#set PFT-level temporary variables
leafout_fromstor = numpy.zeros([self.npfts], numpy.float)
ndays_leafout = numpy.zeros([self.npfts], numpy.float)
for tf in range(0,self.nobs):
v = max(this_trans_cycle,0)*self.nobs + tf
model_year = max(self.model_startyear, self.model_startyear+(this_trans_cycle*self.nobs+tf)/365.)
if (int(model_year) == 1974 and self.forcings['doy'][tf] == 1):
#Strip cut harvest the trees
self.states['stemc_pft'][0:2,v] = 0.01 * self.states['stemc_pft'][0:2,v]
self.states['leafc_pft'][0:2,v] = 0.01 * self.states['leafc_pft'][0:2,v]
self.states['leafc_stor_pft'][0:2,v] = 0.01 * self.states['leafc_stor_pft'][0:2,v]
self.states['rootc_pft'][0:2,v] = 0.01 * self.states['rootc_pft'][0:2,v]
self.states['xsmr_pft'][0:2,v] = 0.01 * self.states['xsmr_pft'][0:2,v]
#leaf, root to litter
self.states['litrc'][v] = self.states['litrc'][v] + pftwt[0] * (90*self.states['leafc_pft'][0,v] + \
99*self.states['rootc_pft'][0,v])
self.states['litrc'][v] = self.states['litrc'][v] + pftwt[1] * (90*self.states['leafc_stor_pft'][1,v] + \
99*self.states['rootc_pft'][1,v])
#set PFT-level temporary variables
leafc_on = numpy.zeros([self.npfts], numpy.float)
leafc_off = numpy.zeros([self.npfts], numpy.float)
leaf_litter = 0
stem_litter = 0
root_litter = 0
vegc = 0
vegc_last = 0
for p in range(0,self.npfts):
a = [nue[p], 0.0156935, 4.22273, 208.868, 0.0453194, 0.37836, 7.19298, 0.011136, \
2.1001, 0.789798]
#Phenology
self.states['gdd_pft'][p,v+1] = (self.forcings['doy'][tf] > 1) * (self.states['gdd_pft'][p,v] + \
max(0.5*(self.forcings['tmax'][tf]+self.forcings['tmin'][tf])-10.0, 0.0))
if (self.parms['evergreen'][p] == 0):
if (self.forcings['doy'][tf] < 200):
if (self.states['gdd_pft'][p,v+1] > gdd_min[p] and self.states['gdd_pft'][p,v] < gdd_min[p]):
leafout_fromstor[p] = self.states['leafc_stor_pft'][p,v] / 2.0
ndays_leafout[p] = ndays_on[p]
if (ndays_leafout[p] > 0):
leafc_on[p] = leafout_fromstor[p] / ndays_on[p]
ndays_leafout[p] = ndays_leafout[p] - 1
elif (self.forcings['tmin'][tf] < tsmin[p] and self.states['lai_pft'][p,v] > 0):
leafc_off[p] = min(leaffall[p]*leafout_fromstor[p], self.states['leafc_pft'][p,v])
else:
leafc_off[p] = self.states['leafc_pft'][p,v] / (3.0 * 365)
#if (p == 1):
# print self.forcings['doy'][tf], self.states['lai_pft'][p,v], ndays_leafout[p], leafc_off[p]
# time.sleep(0.03)
#Calculate GPP flux
if (self.states['lai_pft'][p,v] > 1e-3):
rtot = 1.0
psid = -2.0
leafn = lma[p]/leafcn[p]
if (self.forcings['tmax'][tf]+self.forcings['tmin'][tf])/2 > 0:
gs = abs(psid)**a[9]/((a[5]*rtot+(self.forcings['tmax'][tf]-self.forcings['tmin'][tf])))
pp = max(self.states['lai_pft'][p,v], 0.5)*leafn/gs*a[0]*numpy.exp(a[7]*self.forcings['tmax'][tf])
qq = a[2]-a[3]
#internal co2 concentration
ci = 0.5*(self.forcings['cair'][tf]+qq-pp+((self.forcings['cair'][tf]+qq-pp)**2-4.* \
(self.forcings['cair'][tf]*qq-pp*a[2]))**0.5)
e0 = a[6]*max(self.states['lai_pft'][p,v],0.5)**2/(max(self.states['lai_pft'][p,v],0.5)**2+a[8])
cps = e0*self.forcings['rad'][tf]*gs*(self.forcings['cair'][tf]-ci)/ \
(e0*self.forcings['rad'][tf]+gs*(self.forcings['cair'][tf]-ci))
self.fluxes['gpp_pft'][p,v+1] = cps*(a[1]*self.forcings['dayl'][tf]+a[4])
#ACM not valid at LAI < 0.5. Scale linearly
if (self.states['lai_pft'][p,v] < 0.5):
self.fluxes['gpp_pft'][p,v+1] = self.fluxes['gpp_pft'][p,v+1]*self.states['lai_pft'][p,v]/0.5
else:
self.fluxes['gpp_pft'][p,v+1] = 0.0
else:
self.fluxes['gpp_pft'][p,v+1] = 0.0
#Use the Neural network trained with ELM data
#dayl_factor = (self.forcings['dayl'][tf]/max(self.forcings['dayl'][0:365]))**2.0
#flnr = self.parms['flnr'][p]
#if (v < 10):
# t10 = (self.forcings['tmax'][tf]+self.forcings['tmin'][tf])/2.0+273.15
#else:
# t10 = sum(self.forcings['tmax'][tf-10:tf]+self.forcings['tmin'][tf-10:tf])/20.0+273.15
##Use the NN trained on daily data
#slatop = 1.0/lma[p]
#met_thistimestep=[1.0, self.states['lai_pft'][p,v], self.states['lai_pft'][p,v]/4.0, self.forcings['tmax'][tf]+273.15, \
# self.forcings['tmin'][tf]+273.15, t10, self.forcings['rad'][tf]*1e6, 50.0, self.forcings['cair'][tf]/10.0, \
# dayl_factor, flnr, slatop, leafcn[p], 9.0]
#for i in range(0,self.nparms_nn): #normalize
# met_thistimestep_norm[0,i] = ( met_thistimestep[i] - self.pmin_nn[i] ) / \
# (self.pmax_nn[i] - self.pmin_nn[i])
#self.fluxes['gpp_pft'][p,v+1] = max(self.nnmodel.predict(met_thistimestep_norm), 0.0)
#Autotrophic repiration fluxes
trate = q10_mr**((0.5*(self.forcings['tmax'][tf]+self.forcings['tmin'][tf])-20)/10.0)
if (0.5*(self.forcings['tmax'][tf]+self.forcings['tmin'][tf]) < 0):
trate = 0.0 #no maintenance respiration if air temperature below freezing
self.fluxes['mr_pft'][p,v] = (self.states['leafc_pft'][p,v] / parms['leafcn'][p] + \
0.1*self.states['rootc_pft'][p,v] / 42.0)*br_mr*86400*trate
self.fluxes['gr_pft'][p,v] = max(rg_frac[p]*(self.fluxes['gpp_pft'][p,v] - \
self.fluxes['mr_pft'][p,v]), 0.0)
self.fluxes['npp_pft'][p,v] = self.fluxes['gpp_pft'][p,v] - self.fluxes['mr_pft'][p,v] - self.fluxes['gr_pft'][p,v]
if (p >= 2):
self.fluxes['nee_lc_light'][v] = self.fluxes['nee_lc_light'][v] - pftwt[p] * (self.fluxes['gpp_pft'][p,v]*24. \
/self.forcings['dayl'][tf] -self.fluxes['mr_pft'][p,v] - self.fluxes['gr_pft'][p,v])
self.fluxes['nee'][v] = self.fluxes['nee'][v] - pftwt[p] * self.fluxes['npp_pft'][p,v]
elif (p == 0):
self.fluxes['nee_lc_light'][v] = pftwt[p] * (0.1*self.states['rootc_pft'][p,v] / 42.0)*br_mr*86400*trate
self.fluxes['nee'][v] = -1.0 * pftwt[p] * self.fluxes['npp_pft'][p,v]
elif (p == 1):
self.fluxes['nee_lc_light'][v] = self.fluxes['nee_lc_light'][v] + \
pftwt[p] * (0.1*self.states['rootc_pft'][p,v] / 42.0)*br_mr*86400*trate
self.fluxes['nee'][v] = self.fluxes['nee'][v] - pftwt[p] * self.fluxes['npp_pft'][p,v]
#Update PFT-level variabiles
npp_alloc = max(self.fluxes['npp_pft'][p,v], 0.0)
npp_alloc_to_xsmr = min(-1.0*self.states['xsmr_pft'][p,v]/30.0, npp_alloc)
self.states['xsmr_pft'][p,v+1] = self.states['xsmr_pft'][p,v] + min(self.fluxes['npp_pft'][p,v], 0.0) + npp_alloc_to_xsmr
npp_alloc = npp_alloc - npp_alloc_to_xsmr
leaf_litter = leaf_litter + pftwt[p] * leafc_off[p]
stem_litter = stem_litter + pftwt[p] * (tstem[p] * self.states['stemc_pft'][p,v])
root_litter = root_litter + pftwt[p] * (troot[p] * self.states['rootc_pft'][p,v])
if (self.parms['evergreen'][p] == 0):
self.states['leafc_stor_pft'][p,v+1] = self.states['leafc_stor_pft'][p,v] + npp_alloc*aleaf[p] - leafc_on[p]
self.states['leafc_pft'][p,v+1] = self.states['leafc_pft'][p,v] + leafc_on[p] - leafc_off[p]
else:
self.states['leafc_pft'][p,v+1] = self.states['leafc_pft'][p,v] + npp_alloc*aleaf[p] - leafc_off[p]
self.states['stemc_pft'][p,v+1] = self.states['stemc_pft'][p,v] - (tstem[p] * self.states['stemc_pft'][p,v]) + \
npp_alloc*astem[p]
self.states['rootc_pft'][p,v+1] = self.states['rootc_pft'][p,v] - (troot[p] * self.states['rootc_pft'][p,v]) + \
npp_alloc*(1.0-aleaf[p]-astem[p])
vegc = vegc + pftwt[p] * (self.states['leafc_pft'][p,v+1] + self.states['stemc_pft'][p,v+1] + self.states['rootc_pft'][p,v+1])
vegc_last = vegc_last + pftwt[p] * (self.states['leafc_pft'][p,v] + self.states['stemc_pft'][p,v] + self.states['rootc_pft'][p,v])
self.states['lai_pft'][p,v+1] = self.states['leafc_pft'][p,v] / lma[p]
#Update vegetation and litter pools (allocation, litterfall and decomp)
trate = q10_hr**((0.5*(self.forcings['tmax'][tf]+self.forcings['tmin'][tf])-10)/10.0)
self.states['litrc'][v+1] = self.states['litrc'][v] + leaf_litter + stem_litter + root_litter - \
dr*self.states['litrc'][v] - br_lit*self.states['litrc'][v]*trate
self.states['somc'][v+1] = self.states['somc'][v] + dr*self.states['litrc'][v] - \
br_som*self.states['somc'][v]*trate
self.fluxes['hr'][v] = br_lit*self.states['litrc'][v]*trate + br_som*self.states['somc'][v]*trate
self.fluxes['nee_lc_light'][v] = (self.fluxes['nee_lc_light'][v] + self.fluxes['hr'][v]) / self.forcings['dayl'][tf] #gC/m2/hr (daytime)
self.fluxes['nee'][v] = self.fluxes['nee'][v] + self.fluxes['hr'][v]
def generate_synthetic_obs(self, parms, err):
#generate synthetic observations from model with Gaussian error
self.obs = numpy.zeros([self.nobs], numpy.float)
self.obs_err = numpy.zeros([self.nobs], numpy.float)+err
self.run(parms)
for v in range(0,self.nobs):
self.obs[v] = self.fluxes[v]+numpy.random.normal(0,err,1)
self.issynthetic = True