forked from stineb/nn_fluxnet2015
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathfit_fvar_vs_soilm_nn_fluxnet2015.R
272 lines (220 loc) · 10.9 KB
/
fit_fvar_vs_soilm_nn_fluxnet2015.R
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
library( dplyr )
library( tidyr )
library( minpack.lm )
##------------------------------------------------
## Select all sites for which method worked (codes 1 and 2 determined by 'nn_getfail_fluxnet2015.R')
##------------------------------------------------
successcodes <- read.csv( "successcodes.csv", as.is = TRUE )
do.sites <- dplyr::filter( successcodes, successcode==1 | successcode==2 )$mysitename
## Manual settings ----------------
# do.sites = "FR-Pue"
nam_target = "lue_obs_evi"
use_weights= FALSE
use_fapar = FALSE
package = "nnet"
nrep = 5
dotrain = FALSE
overwrite_modis = FALSE
overwrite_mte = FALSE
##---------------------------------
siteinfo <- read.csv( "siteinfo_fluxnet2015_sofun.csv" )
##------------------------------------------------
## Load overview L1 and initialise additional columns created here
##------------------------------------------------
load( "data/overview_data_fluxnet2015_L1.Rdata" )
## fLUE in the lower quartile [bysite]
overview$fvar_min <- rep( NA, nrow(overview) )
## fLUE in the upper 10% quantile [bysite]
overview$fvar_max <- rep( NA, nrow(overview) )
## initialise matrix of WUE medians by soil moisture-quantile
nintervals <- 5
fvar_vs_soilm <- matrix( NA, length(do.sites), nintervals )
fitparams <- data.frame()
##------------------------------------------------
## Define functions for the fitting and visualisation
##------------------------------------------------
stress_exp <- function( x, ymin, xmin, par_nonlin ){
outstress <- (1.0 - ymin) * (1.0 - exp( -par_nonlin * (x - xmin) ) ) + ymin
return( outstress )
}
# stress_quot <- function( x, ymin, xmin, par_nonlin ){
# outstress <- (1.0 - ymin) * (x - xmin) / (par_nonlin + (x - xmin)) + ymin
# return( outstress )
# }
stress_quot <- function( x, xmin, xmax, par_nonlin ){
if ( x > xmax ){
outstress <- 1.0
} else {
outstress <- ( ( x - xmin ) / ( xmax - xmin ) ) ^ par_nonlin
}
return( outstress )
}
stress_quad <- function( x, x0, off, apar ){
outstress <- 1.0 + off - apar * ( x - x0 ) ^ 2
return( outstress )
}
mycurve <- function( func, from, to, col='black', add=FALSE, lwd=1, lty=1 ){
range_x <- seq( from, to, by=(to-from)/100 )
range_y <- sapply( range_x, func )
if (add){
lines( range_x, range_y, type="l", col=col, lwd=lwd, lty=lty )
} else {
plot( range_x, range_y, type="l", col=col, lwd=lwd, lty=lty )
}
}
## check and override if necessary
if ( nam_target=="lue_obs_evi" || nam_target=="lue_obs_fpar" ){
plotlue <- TRUE
if (nam_target=="lue_obs_evi"){
fapar_data <- "evi"
} else if (nam_target=="lue_obs_fpar"){
fapar_data <- "fpar"
}
if (use_fapar){
print("WARNING: setting use_fapar to FALSE")
use_fapar <- FALSE
}
}
## identifier for output files
if (use_fapar){
if (nam_target=="lue_obs_evi"){
char_fapar <- "_withEVI"
} else if (nam_target=="lue_obs_fpar"){
char_fapar <- "_withFPAR"
} else {
print("ERROR: PROVIDE VALID FAPAR DATA!")
}
} else {
char_fapar <- ""
}
if (use_weights){
char_wgt <- "_wgt"
} else {
char_wgt <- ""
}
print( "Fitting functional relationship for all sites ..." )
jdx <- 0
for (sitename in do.sites){
jdx <- jdx + 1
infil <- paste( "data/fvar/nn_fluxnet2015_", sitename, "_", nam_target, char_wgt, char_fapar, ".Rdata", sep="" )
##------------------------------------------------
## load nn_fVAR data and "detatch"
##------------------------------------------------
load( infil ) ## gets list 'nn_fluxnet'
nice <- as.data.frame( nn_fluxnet[[ sitename ]]$nice )
varnams_swc <- nn_fluxnet[[ sitename ]]$varnams_swc
varnams_swc_obs <- nn_fluxnet[[ sitename ]]$varnams_swc_obs
# if (sitename == "SD-Dem" || sitename == "SN-Dhr"){
# varnams_swc_mod <- varnams_swc[ -which( varnams_swc=="soilm_obs" ) ]
# # nice$soilm_mean <- apply( dplyr::select( nice, one_of(varnams_swc_mod)), 1, FUN=mean, na.rm=TRUE )
# nice$soilm_mean <- apply( dplyr::select( nice, one_of(varnams_swc_obs) ), 1, FUN=mean, na.rm=TRUE )
# nice$soilm_mean[ is.nan( nice$soilm_mean ) ] <- NA
# }
## add row to aggregated data
mysitename <- data.frame( mysitename=rep( sitename, nrow(nice) ) )
##------------------------------------------------
## Get functional form of fVAR vs. soil moisture, treating (soilm_xxx, fvar) pairs as individual observations
##------------------------------------------------
## Get median by interval and get fvar_vs_soilm for this site (used for clustering)
intervals <- seq( 0.0, 1.0, 1.0/nintervals )
mid <- intervals[1:(length(intervals)-1)] + (intervals[2]-intervals[1])/2
nice$ininterval <- NULL
nice <- nice %>% mutate( ininterval = cut( soilm_mean , breaks = intervals ) ) %>% group_by( ininterval )
tmp <- nice %>% dplyr::summarise( median=median( fvar, na.rm=TRUE ) ) %>% complete( ininterval, fill = list( median = NA ) ) %>% dplyr::select( median )
fvar_vs_soilm[ jdx, ] <- unlist(tmp)[1:nintervals]
## Get median by interval and get fvar_vs_soilm for this site (used for clustering)
nsintervals <- 5
sintervals <- seq( 0.0, 1.0, 1.0/nsintervals )
xvals <- sintervals[1:(length(sintervals)-1)] + (sintervals[2]-sintervals[1])/2
nice$insinterval <- NULL
nice <- nice %>% mutate( insinterval = cut( soilm_mean , breaks = sintervals ) ) %>% group_by( insinterval )
tmp <- nice %>% dplyr::summarise( median=median( fvar, na.rm=TRUE ) ) %>% complete( insinterval, fill = list( median = NA ) ) %>% dplyr::select( median )
yvals <- unlist(tmp)[1:nsintervals]
# ## Fit by all data
# gpp_stressfit <- try(
# nlsLM(
# fvar ~ stress_quad( soilm_mean, x0, off, apar ),
# data=nice,
# start=list( x0=1.0, off=0.0, apar=1.0 ),
# lower=c( 0.01, -1.0, 0.001 ),
# upper=c( 1.0, 1.0, 1000 ),
# algorithm="port"
# )
# )
## Fit by medians in bis
df_tmp <- data.frame( xvals=xvals, yvals=yvals )
gpp_stressfit <- try(
nlsLM(
yvals ~ stress_quad( xvals, x0, off, apar ),
data=df_tmp,
start=list( x0=1.0, off=0.0, apar=1.0 ),
lower=c( 0.01, -1.0, 0.01 ),
algorithm="port"
)
)
## add fit parameters to aggregated data frame
addrow <- data.frame( x0=coef(gpp_stressfit)[[ "x0" ]], off=coef(gpp_stressfit)[[ "off" ]], apar=coef(gpp_stressfit)[[ "apar" ]] )
fitparams <- rbind( fitparams, addrow )
if (class(gpp_stressfit)!="try-error"){ rmse_gpp_stressfit <- mean(summary(gpp_stressfit)$residuals^2) }
##------------------------------------------------
## Plot fvar vs. soil moisture
##------------------------------------------------
pdf( paste( "fig_nn_fluxnet2015/fvar_vs_soilm/fvar_vs_soilm_mean_", sitename, ".pdf", sep="" ), width=6, height=5 )
par(las=1)
plot( nice$soilm_mean, nice$fvar, xlim=c(0,1), ylim=c(0,1.2), pch=16, xlab="soil water content (fraction)", ylab="fLUE", col=add_alpha("royalblue3", 0.2) )
bp <- boxplot( fvar ~ insinterval, data=nice, main=sitename, col=NA, las=1, outline = FALSE, na.rm=TRUE, add=TRUE, at=(sintervals[1:nsintervals]+(1.0/(2*nsintervals))), boxwex=0.05, axes=FALSE )
abline( h=1.0, lwd=0.5 )
points( xvals, yvals, pch=16, col='red' )
if (class(gpp_stressfit)!="try-error"){
mycurve( function(x) stress_quad( x, coef(gpp_stressfit)[[ "x0" ]], coef(gpp_stressfit)[[ "off" ]], coef(gpp_stressfit)[[ "apar" ]] ), from=0.0, to=1.0, col='red', add=TRUE, lwd=2 )
## alternative tested:
# mycurve( function(x) stress_quot( x, coef(gpp_stressfit)[[ "xmin" ]], coef(gpp_stressfit)[[ "xmax" ]], coef(gpp_stressfit)[[ "par_nonlin" ]] ), from=0.0, to=1.0, col='red', add=TRUE, lwd=2 )
# abline( v=coef(gpp_stressfit)[[ "xmin" ]], lty=3 )
# abline( v=coef(gpp_stressfit)[[ "xmax" ]], lty=3 )
# mtext( bquote( "RMSE" == .(format( rmse_gpp_stressfit, digits=3 ) ) ), side=3, line=0, adj=1 )
}
dev.off()
##------------------------------------------------
## Plot the same again but now not into PDFs so that it will be included in knitted file
##------------------------------------------------
if ( sitename %in% c( "AU-DaP", "FR-Pue", "IT-PT1") ){
## fLUE vs. soil moisture
par(las=1)
plot( nice$soilm_mean, nice$fvar, xlim=c(0,1), ylim=c(0,1.2), pch=16, xlab="soil water content (fraction)", ylab="fLUE", col=add_alpha("royalblue3", 0.2) )
bp <- boxplot( fvar ~ insinterval, data=nice, main=sitename, col=NA, las=1, outline = FALSE, na.rm=TRUE, add=TRUE, at=(sintervals[1:nsintervals]+(1.0/(2*nsintervals))), boxwex=0.05, axes=FALSE )
abline( h=1.0, lwd=0.5 )
points( xvals, yvals, pch=16, col='red' )
if (class(gpp_stressfit)!="try-error"){
mycurve( function(x) stress_quad( x, coef(gpp_stressfit)[[ "x0" ]], coef(gpp_stressfit)[[ "off" ]], coef(gpp_stressfit)[[ "apar" ]] ), from=0.0, to=1.0, col='red', add=TRUE, lwd=2 )
}
}
##------------------------------------------------
## Add to overview data table: fvar in lowest soil moisture quartile
##------------------------------------------------
overview$fvar_min[ which(overview$mysitename==sitename) ] <- fvar_vs_soilm[ jdx, 1 ]
##------------------------------------------------
## Add to overview data table: fvar in upper 10% soil moisture quantile
##------------------------------------------------
## Get median by interval and get fvar_vs_soilm for this site (used for clustering)
nsintervals <- 10
sintervals <- seq( 0.0, 1.0, 1.0/nsintervals )
xvals <- sintervals[1:(length(sintervals)-1)] + (sintervals[2]-sintervals[1])/2
nice <- nice %>% mutate( inssinterval = cut( soilm_mean , breaks = sintervals ) ) %>% group_by( inssinterval )
tmp <- nice %>% dplyr::summarise( median=median( fvar, na.rm=TRUE ) ) %>% complete( inssinterval, fill = list( median = NA ) ) %>% dplyr::select( median )
yvals <- unlist(tmp)[1:nsintervals]
overview$fvar_max[ which(overview$mysitename==sitename) ] <- yvals[nsintervals]
}
print( "... done." )
if ( length( dplyr::filter( successcodes, successcode==1 | successcode==2 )$mysitename ) == length( do.sites ) ){
##------------------------------------------------
## save collected data
##------------------------------------------------
## aggregated response of fLUE to soil moisture
rownames(fvar_vs_soilm) <- do.sites
save( fvar_vs_soilm, file="data/fvar_vs_soilm.Rdata")
rownames(fitparams) <- do.sites
save( fitparams, file="data/fitparams.Rdata")
save( overview, file="data/overview_data_fluxnet2015_L2.Rdata" )
} else {
print("WARNING: NO SAVING AT THE END!")
}