-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathMILc_2.py
267 lines (196 loc) · 7.37 KB
/
MILc_2.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
# -*- coding: utf-8 -*-
"""
Created on Tue Mar 8 23:20:11 2016
@author: christian_massari, Luca Brocca
"""
import numpy as np
import scipy as sc
from scipy import misc
import pandas as pd
from datetime import datetime, timedelta
import matplotlib.pyplot as plt
import sys
#%%
def IUH_comp(gamma,Ab,dt,deltaT):
"""
%% % -------------------------------------------------------------------------------
#% Calculation of Geomorphological Instantaneous Unit Hydrograph
#% ---------------------------------------------------------------------------------
"""
Lag=(gamma*1.19*Ab**0.33)/deltaT
hp=0.8/Lag
data=np.loadtxt('IUH.txt')
t=data[:,0]*Lag
IUH_0=data[:,1]*hp
ti=np.arange(0,max(t),dt)
IUH=np.interp(ti,t,IUH_0)
return IUH
#%%
def IUH_NASH(n,gamma,Ab,dt,deltaT):
"""
% -------------------------------------------------------------------------------
% Calculation of Nash Instantaneous Unit Hydrograph
% -------------------------------------------------------------------------------
"""
K=(gamma*1.19*Ab**.33)/deltaT
time=np.arange(0,100,dt)
IUH=((time/K)**(n-1)*np.exp(-time/K)/sc.special.factorial(n-1)/K)
return IUH
#%%
class Perf:
def __init__(self,pd_ser):
self.pd_ser=pd_ser
self.descrition="This class calculates the agreement between two time series thorugh different performance scores"
self.author="Christian Massari"
def RMSE(self):
temp1=((self.pd_ser['S']-self.pd_ser['O'])**2)**0.5
return temp1.mean()
def NS(self):
temp1=((self.pd_ser['S']-self.pd_ser['O'])**2)
temp2=(self.pd_ser['O']-self.pd_ser['O'].mean())**2
NS=1-temp1.sum()/temp2.sum()
return NS
def ANSE(self):
temp1=((self.pd_ser['S']-self.pd_ser['O'])**2)
temp2=((self.pd_ser['O']-self.pd_ser['O'].mean())**2)
temp3=(self.pd_ser['O']+self.pd_ser['O'].mean())
temp4=temp3*temp1
temp5=temp3*temp2
ANSE=1-temp4.sum()/temp5.sum()
return ANSE
def R(self):
return self.pd_ser['O'].corr(self.pd_ser['S'])
def NS_lnQ(self):
temp1=(np.log(self.pd_ser['S']+0.00001)-np.log(self.pd_ser['O']+0.00001))**2
temp2=((np.log(self.pd_ser['O']+0.00001)-(np.log(self.pd_ser['O']+0.00001)).mean())**2)
return 1-temp1.sum()/temp2.sum()
self.pd_ser['O']['a'].corr(self.pd_ser['O'])
#%% Read data from file
# -*- coding: utf-8 -*-
"""
Created on Thu Mar 10 13:26:54 2016
@author: christian_massari
"""
def MILC(name,data_input,PAR,Ab,fig,Wobs=[],K=0):
PIO=data_input['P']
TEMPER=data_input['T']
Qobs=data_input['Q']
N=len(Qobs)
data_input.index=pd.to_datetime(data_input.index)
MESE = data_input.index.month.values
#% READ MODEL PARAMETERS
W_p = PAR[0] # initial conditions, fraction of W_max (0-1)
W_max = PAR[1] # Field capacity
m2 = PAR[2] # exponent of drainage
Ks = PAR[3] # Ks parameter of infiltration and drainage
Nu = PAR[4] # fraction of drainage verusu interflow
gamma1 = PAR[5] # coefficient lag-time relationship
Kc = PAR[6] # parameter of potential evapotranspiration
alpha = PAR[7] # runoff exponent
delta_T = 24 # input data time step in hour
dt = 0.2 # computation time step in hour
Ks = Ks*24
#% Potential Evapotranspiration parameter
L=np.array([0.2100,0.2200,0.2300,0.2800,0.3000,0.3100,0.3000,0.2900,0.2700,0.2500,0.2200,0.2000])
Ka=1.26
T=TEMPER.values
EPOT=(T>0)*(Kc*(Ka*L[MESE-1]*(0.46*T+8)-2))/(24/delta_T)
#% INITIALIZATION
BF=np.zeros(N)
QS=np.zeros(N)
WW=np.zeros(N)
PERC=np.zeros(N)
#% MAIN ROUTINE
P=PIO.values
#Q=Qobs.values
W=W_p*W_max
PIOprec=0
S=np.nan
Pcum=0
IE=0
for t in range(N):
IE=P[t]*(W/W_max)**alpha
E=EPOT[t]*W/W_max
PERC=Nu*Ks*(W/W_max)**(m2)
BF[t]=(1-Nu)*Ks*(W/W_max)**(m2)
W=W+(P[t]-BF[t]-IE-PERC-E)
# data assimilation with nudging
if K>0:
if ~np.isnan(Wobs[t]):
W=K*(Wobs[t]*W_max)+(1-K)*W
if W>=W_max:
SE=W-W_max
W=W_max
else:
SE=0
QS[t]=IE+SE
WW[t]=W/W_max
if t>2:
PIOprec=np.sum(P[t-3:t])
WWW=pd.DataFrame(WW, index=data_input.index)
WWW.columns=['W']
df2=data_input.join(WWW)
#% Convolution (GIUH)
IUH1=IUH_comp(gamma1,Ab,dt,delta_T)*dt
IUH1=IUH1/np.sum(IUH1)
IUH2=IUH_NASH(1,0.5*gamma1,Ab,dt,delta_T)*dt
IUH2=IUH2/np.sum(IUH2);
QSint=np.interp(np.arange(0,N,dt),np.arange(0,N,1),QS)
BFint=np.interp(np.arange(0,N,dt),np.arange(0,N,1),BF)
temp1=np.convolve(IUH1,QSint)
temp2=np.convolve(IUH2,BFint)
yy=np.arange(0,N*np.round(1/dt),np.round(1/dt))
ii=yy.astype(int)
#Qsim1=temp2[ii]*(Ab*1000./delta_T/3600)
Qsim=(temp1[ii]+temp2[ii])*(Ab*1000./delta_T/3600)
te1 = pd.DataFrame(Qsim, index=data_input.index, columns=list('S'))
Qout=Qobs.copy()
te2=pd.Series.to_frame(Qout)
QQ=te1.join(te2)
QQ.columns=['S','O']
df3=df2.join(te1)
out=Perf(QQ)
#% PRINT FIGURE
if fig>0:
stringa_per= name[0:-4]+" NS="+ "%0.3f" % out.NS()+" ANSE="+ "%0.3f" % out.ANSE()+" RMSE="+ "%0.3f" % out.RMSE()+' $ m^3/s$'
f, ax = plt.subplots(2, sharex=True, figsize=(12, 12))
ax[0].plot(df3.index, df3['P'].values,label='Rainfall',color='b')
ax[0].set_ylim(0,np.max(df3['P'].values)+5)
ax[0].set_ylabel('Rainfall [mm]', fontsize=16)
ax2 = ax[0].twinx()
ax2.plot(df3.index, df3['W'].values,label='Soil Moisture',color='g')
ax2.set_title(stringa_per,fontsize=20)
ax2.set_ylim(0,np.max(df3['W'].values)+0.05)
ax2.set_ylabel('Relative saturation [-]', fontsize=16)
ax[0].grid(True)
ax[0].tick_params(axis='y', labelsize=16)
ax2.tick_params(axis='y', labelsize=16)
ax2.legend(loc='upper right', shadow=True)
ax[0].legend(loc='upper right', shadow=True)
ax[1].set_ylabel('Rainfall [mm]')
ax[1].plot(df3.index, df3['Q'].values,label='Qobs',color='g')
ax[1].plot(df3.index, df3['S'].values,label='Qsim',color='r')
ax[1].set_ylim(0,np.max(df3.max())+10)
#ax[1].set_ylim(0,df3.+10)
ax[1].set_ylabel('Discharge [$m^3/s$]', fontsize=16)
ax[1].grid(True)
ax[1].tick_params(axis='y', labelsize=16)
ax[1].tick_params(axis='x', labelsize=14)
ax[1].legend(loc='upper right', shadow=True)
#plt.legend(loc='lower right')
f.savefig(name[0:-4]+'.png',dpi=200)
return out,df3
#%% MODDEL RUN
if __name__ == '__main__':
name='migi_0406.txt'
data_input=pd.read_csv(name,index_col=0,header = None, names = ['P','T','Q'])
PAR=np.loadtxt('X_opt_'+name)
fig=1
QobsQsim,data=MILC(name,data_input,PAR,fig)
Wmodel=data['W']
ASC=pd.read_csv('ASCAT.csv',index_col=0)
# here must match the ASCAT time series with those of the model i.e. ASC and Wmodel
QobsQsim,data=MILC(name,data_input,PAR,Ab,fig,ASC,K=0.1)
print(QobsQsim.NS())
print(QobsQsim.R())
print(data)