1
|
##################################### METHODS COMPARISON ##########################################
|
2
|
#################################### Spatial Analysis ########################################
|
3
|
#This script is not aimed at producing new interpolation surfaces. It utilizes the R ojbects created
|
4
|
# during the interpolation phase. #
|
5
|
# At this stage the script produces figures of various accuracy metrics and compare methods: #
|
6
|
#- multisampling plots #
|
7
|
#- spatial accuracy in terms of distance to closest station #
|
8
|
#- spatial density of station network and accuracy metric
|
9
|
#- visualization of maps of prediction and difference for comparison
|
10
|
#AUTHOR: Benoit Parmentier #
|
11
|
#DATE: 10/30/2012 #
|
12
|
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#491 -- #
|
13
|
###################################################################################################
|
14
|
|
15
|
###Loading R library and packages
|
16
|
#library(gtools) # loading some useful tools
|
17
|
library(mgcv) # GAM package by Wood 2006 (version 2012)
|
18
|
library(sp) # Spatial pacakge with class definition by Bivand et al. 2008
|
19
|
library(spdep) # Spatial package with methods and spatial stat. by Bivand et al. 2012
|
20
|
library(rgdal) # GDAL wrapper for R, spatial utilities (Keitt et al. 2012)
|
21
|
library(gstat) # Kriging and co-kriging by Pebesma et al. 2004
|
22
|
library(automap) # Automated Kriging based on gstat module by Hiemstra et al. 2008
|
23
|
library(spgwr)
|
24
|
library(gpclib)
|
25
|
library(maptools)
|
26
|
library(graphics)
|
27
|
library(parallel) # Urbanek S. and Ripley B., package for multi cores & parralel processing
|
28
|
library(raster)
|
29
|
library(rasterVis)
|
30
|
library(plotrix) #Draw circle on graph
|
31
|
library(reshape)
|
32
|
## Functions
|
33
|
#loading R objects that might have similar names
|
34
|
load_obj <- function(f)
|
35
|
{
|
36
|
env <- new.env()
|
37
|
nm <- load(f, env)[1]
|
38
|
env[[nm]]
|
39
|
}
|
40
|
|
41
|
###Parameters and arguments
|
42
|
|
43
|
infile1<- "ghcn_or_tmax_covariates_06262012_OR83M.shp" #GHCN shapefile containing variables for modeling 2010
|
44
|
#infile2<-"list_10_dates_04212012.txt" #List of 10 dates for the regression
|
45
|
infile2<-"list_365_dates_04212012.txt" #list of dates
|
46
|
infile3<-"LST_dates_var_names.txt" #LST dates name
|
47
|
infile4<-"models_interpolation_05142012.txt" #Interpolation model names
|
48
|
infile5<-"mean_day244_rescaled.rst" #mean LST for day 244
|
49
|
inlistf<-"list_files_05032012.txt" #list of raster images containing the Covariates
|
50
|
infile6<-"OR83M_state_outline.shp"
|
51
|
#stat_loc<-read.table(paste(path,"/","location_study_area_OR_0602012.txt",sep=""),sep=",", header=TRUE)
|
52
|
|
53
|
|
54
|
obj_list<-"list_obj_08262012.txt" #Results of fusion from the run on ATLAS
|
55
|
path<-"/home/parmentier/Data/IPLANT_project/methods_interpolation_comparison" #Jupiter LOCATION on Atlas for kriging #Jupiter Location on XANDERS
|
56
|
#path<-"/Users/benoitparmentier/Dropbox/Data/NCEAS/Oregon_covariates/" #Local dropbox folder on Benoit's laptop
|
57
|
setwd(path)
|
58
|
proj_str="+proj=lcc +lat_1=43 +lat_2=45.5 +lat_0=41.75 +lon_0=-120.5 +x_0=400000 +y_0=0 +ellps=GRS80 +units=m +no_defs";
|
59
|
out_prefix<-"methods_09262012_" #User defined output prefix
|
60
|
|
61
|
sampling_CAI<-load_obj("results2_CAI_sampling_obj_09132012_365d_GAM_CAI2_multisampling2.RData")
|
62
|
sampling_fus<-load_obj("results2_fusion_sampling_obj_10d_GAM_fusion_multisamp4_09192012.RData")
|
63
|
fus_CAI_mod<-load_obj("results2_CAI_Assessment_measure_all_09132012_365d_GAM_CAI2_multisampling2.RData")
|
64
|
gam_fus_mod1<-load_obj("results2_fusion_Assessment_measure_all_10d_GAM_fusion_multisamp4_09192012.RData")
|
65
|
|
66
|
filename<-sub(".shp","",infile1) #Removing the extension from file.
|
67
|
ghcn<-readOGR(".", filename) #reading shapefile
|
68
|
|
69
|
#CRS<-proj4string(ghcn) #Storing projection information (ellipsoid, datum,etc.)
|
70
|
lines<-read.table(paste(path,"/",inlistf,sep=""), sep="") #Column 1 contains the names of raster files
|
71
|
inlistvar<-lines[,1]
|
72
|
inlistvar<-paste(path,"/",as.character(inlistvar),sep="")
|
73
|
covar_names<-as.character(lines[,2]) #Column two contains short names for covaraites
|
74
|
|
75
|
s_raster<- stack(inlistvar) #Creating a stack of raster images from the list of variables.
|
76
|
layerNames(s_raster)<-covar_names #Assigning names to the raster layers
|
77
|
projection(s_raster)<-proj_str
|
78
|
|
79
|
#Create mask using land cover data
|
80
|
pos<-match("LC10",layerNames(s_raster)) #Find the layer which contains water bodies
|
81
|
LC10<-subset(s_raster,pos)
|
82
|
LC10[is.na(LC10)]<-0 #Since NA values are 0, we assign all zero to NA
|
83
|
mask_land<-LC10<100 #All values below 100% water are assigned the value 1, value 0 is "water"
|
84
|
mask_land_NA<-mask_land
|
85
|
mask_land_NA[mask_land_NA==0]<-NA #Water bodies are assigned value 1
|
86
|
|
87
|
data_name<-"mask_land_OR"
|
88
|
raster_name<-paste(data_name,".rst", sep="")
|
89
|
writeRaster(mask_land, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...(IDRISI)
|
90
|
#writeRaster(r2, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...(IDRISI)
|
91
|
|
92
|
pos<-match("mm_01",layerNames(s_raster))
|
93
|
mm_01<-subset(s_raster,pos)
|
94
|
mm_01<-mm_01-273.15
|
95
|
mm_01<-mask(mm_01,mask_land_NA) #Raster image used as backround
|
96
|
#mention this is the last... files
|
97
|
|
98
|
### RESULTS COMPARISON
|
99
|
|
100
|
### CODE BEGIN #####
|
101
|
|
102
|
### PART I MULTISAMPLING COMPARISON ####
|
103
|
|
104
|
tb_diagnostic2<-sampling_CAI$tb #Extracting the accuracy metric information...
|
105
|
tb_diagnostic<-sampling_fus$tb
|
106
|
|
107
|
tb_diagnostic[["prop"]]<-as.factor(tb_diagnostic[["prop"]])
|
108
|
tb_diagnostic2[["prop"]]<-as.factor(tb_diagnostic2[["prop"]])
|
109
|
|
110
|
#Preparing the data for the plot
|
111
|
#fus data
|
112
|
t<-melt(tb_diagnostic,
|
113
|
measure=c("mod1","mod2","mod3","mod4", "mod5", "mod6", "mod7", "mod8","mod9"),
|
114
|
id=c("dates","metric","prop"),
|
115
|
na.rm=F)
|
116
|
avg_tb<-cast(t,metric+prop~variable,mean)
|
117
|
sd_tb<-cast(t,metric+prop~variable,sd)
|
118
|
n_tb<-cast(t,metric+prop~variable,length)
|
119
|
avg_tb[["prop"]]<-as.numeric(as.character(avg_tb[["prop"]]))
|
120
|
avg_RMSE<-subset(avg_tb,metric=="RMSE")
|
121
|
|
122
|
#CAI data
|
123
|
t2<-melt(tb_diagnostic2,
|
124
|
measure=c("mod1","mod2","mod3","mod4", "mod5", "mod6", "mod7", "mod8","mod9"),
|
125
|
id=c("dates","metric","prop"),
|
126
|
na.rm=F)
|
127
|
avg_tb2<-cast(t2,metric+prop~variable,mean)
|
128
|
sd_tb2<-cast(t2,metric+prop~variable,sd)
|
129
|
n_tb2<-cast(t2,metric+prop~variable,length)
|
130
|
avg_tb2[["prop"]]<-as.numeric(as.character(avg_tb2[["prop"]]))
|
131
|
avg_RMSE2<-subset(avg_tb2,metric=="RMSE")
|
132
|
|
133
|
#Select only information related to FUSION
|
134
|
|
135
|
x<-avg_RMSE[["prop"]]
|
136
|
i=9
|
137
|
mod_name<-paste("mod",i,sep="")
|
138
|
y<-avg_RMSE[[mod_name]]
|
139
|
|
140
|
sd_tb_RMSE <- subset(sd_tb, metric=="RMSE")
|
141
|
x_sd<-sd_tb_RMSE[["prop"]]
|
142
|
i=9
|
143
|
mod_name<-paste("mod",i,sep="")
|
144
|
y_sd<-sd_tb_RMSE[[mod_name]]
|
145
|
|
146
|
#Select only information related to CAI
|
147
|
|
148
|
x2<-avg_RMSE2[["prop"]]
|
149
|
i=9
|
150
|
mod_name<-paste("mod",i,sep="")
|
151
|
y2<-avg_RMSE2[[mod_name]]
|
152
|
|
153
|
sd_tb_RMSE2 <- subset(sd_tb2, metric=="RMSE")
|
154
|
x_sd2<-sd_tb_RMSE2[["prop"]]
|
155
|
i=9
|
156
|
mod_name<-paste("mod",i,sep="")
|
157
|
y_sd2<-sd_tb_RMSE2[[mod_name]]
|
158
|
|
159
|
n=150
|
160
|
ciw <- qt(0.975, n) * y_sd / sqrt(n)
|
161
|
ciw2 <- qt(0.975, n) * y_sd2 / sqrt(n)
|
162
|
|
163
|
#Comparison of MAE for different proportions for FUSION and CAI using CI
|
164
|
X11()
|
165
|
plotCI(y=y, x=x, uiw=ciw, col="red", main=" FUS: RMSE proportion of validation hold out", barcol="blue", lwd=1,
|
166
|
ylab="RMSE (C)", xlab="Proportions of validation hold out (in %)")
|
167
|
lines(x,y,col="red")
|
168
|
legend("bottomright",legend=c("fus"), cex=1.2, col=c("red"),
|
169
|
lty=1, title="RMSE")
|
170
|
savePlot(paste("Comparison_multisampling_fus_RMSE_CI",out_prefix,".png", sep=""), type="png")
|
171
|
|
172
|
plotCI(y=y2, x=x2, uiw=ciw2, col="black", main=" CAI: RMSE proportion of validation hold out", barcol="blue", lwd=1,
|
173
|
ylab="RMSE (C)", xlab="Proportions of validation hold out (in %)")
|
174
|
lines(x2,y2,col="grey")
|
175
|
legend("bottomright",legend=c("CAI"), cex=1.2, col=c("grey"),
|
176
|
lty=1, title="RMSE")
|
177
|
savePlot(paste("Comparison_multisampling_CAI_RMSE_CI",out_prefix,".png", sep=""), type="png")
|
178
|
dev.off()
|
179
|
|
180
|
#Comparison of MAE for different proportions for FUSION and CAI
|
181
|
X11()
|
182
|
plot(x,y,col="red",type="b", ylab="RMSE (C)", xlab="Proportions of validation hold out (in %)")
|
183
|
lines(x2,y2,col="grey")
|
184
|
points(x2,y2,col="grey")
|
185
|
title("MAE in terms of proportions and random sampling")
|
186
|
legend("bottomright",legend=c("fus","CAI"), cex=1.2, col=c("red","grey"),
|
187
|
lty=1, title="RMSE")
|
188
|
savePlot(paste("Comparison_multisampling_fus_CAI_RMSE_averages",out_prefix,".png", sep=""), type="png")
|
189
|
dev.off()
|
190
|
|
191
|
############################################################################
|
192
|
#### PART II EXAMINIG PREDICTIONS AND RESIDUALS TEMPORAL PROFILES ##########
|
193
|
|
194
|
l_f<-list.files(pattern="*tmax_predicted.*fusion5.rst$") #Search for files in relation to fusion
|
195
|
l_f2<-list.files(pattern="CAI_tmax_predicted.*_GAM_CAI2.rst$")
|
196
|
inlistpred<-paste(path,"/",as.character(l_f),sep="")
|
197
|
inlistpred2<-paste(path,"/",as.character(l_f2),sep="")
|
198
|
|
199
|
fus_rast<- stack(inlistpred) #Creating a stack of raster images from the list of variables.
|
200
|
cai_rast<- stack(inlistpred2) #Creating a stack of raster images from the list of variables.
|
201
|
|
202
|
id<-unique(ghcn$station)
|
203
|
ghcn_id<-as.data.frame(subset(ghcn,select=c("station","x_OR83M","y_OR83M")))
|
204
|
|
205
|
ghcn_melt<-melt(ghcn_id,
|
206
|
measure=c("x_OR83M","y_OR83M"),
|
207
|
id=c("station"),
|
208
|
na.rm=F)
|
209
|
|
210
|
ghcn_cast<-cast(ghcn_melt,station~variable,mean)
|
211
|
ghcn_locs<-as.data.frame(ghcn_cast)
|
212
|
|
213
|
coords<- ghcn_locs[,c('x_OR83M','y_OR83M')]
|
214
|
coordinates(ghcn_locs)<-coords
|
215
|
proj4string(ghcn_locs)<-proj_str #Need to assign coordinates...
|
216
|
|
217
|
tmp<-extract(fus_rast,ghcn_locs)
|
218
|
tmp2<-extract(cai_rast,ghcn_locs)
|
219
|
|
220
|
tmp_names<-paste("fusd",seq(1,365),sep="")
|
221
|
colnames(tmp)<-tmp_names
|
222
|
tmp_names<-paste("caid",seq(1,365),sep="")
|
223
|
colnames(tmp2)<-tmp_names
|
224
|
ghcn_fus_pred<-cbind(as.data.frame(ghcn_locs),as.data.frame(tmp))
|
225
|
ghcn_cai_pred<-cbind(as.data.frame(ghcn_locs),as.data.frame(tmp2))
|
226
|
|
227
|
write.table(ghcn_fus_pred,file="extract3_fus_y2010.txt",sep=",")
|
228
|
write.table(ghcn_cai_pred,file="extract3_cai_y2010.txt",sep=",")
|
229
|
|
230
|
ghcn$value[ghcn$value< -150 | ghcn$value>400]<-NA #screenout values out of range
|
231
|
ghcn$value<-ghcn$value/10
|
232
|
ghcn_m<-melt(as.data.frame(ghcn),
|
233
|
measure=c("value"),
|
234
|
id=c("station","date"),
|
235
|
na.rm=F)
|
236
|
|
237
|
ghcn_mc<-cast(ghcn_m,station~date~variable,mean) #This creates an array of dimension 186,366,1
|
238
|
|
239
|
ghcn_value<-as.data.frame(ghcn_mc[,,1])
|
240
|
ghcn_value<-cbind(ghcn_locs,ghcn_value[,1:365]) #This data frame contains values for 365 days
|
241
|
#for 186 stations of year 2010...
|
242
|
write.table(ghcn_value,na="",file="extract3_dailyTmax_y2010.txt",sep=",")
|
243
|
|
244
|
id<-c("USW00094261","USW00004141","USC00356252","USC00357208")
|
245
|
#id<-c("USW00024284","USC00354126","USC00358536","USC00354835",
|
246
|
"USC00356252","USC00359316","USC00358246","USC00350694",
|
247
|
"USC00350699","USW00024230","USC00353542")
|
248
|
|
249
|
m<-match(id,ghcn_locs$station)
|
250
|
dat_id<-ghcn_locs[m,] #creating new subset
|
251
|
#dat_id<-subset(ghcn_locs[gj])
|
252
|
|
253
|
filename<-sub(".shp","",infile6) #Removing the extension from file.
|
254
|
reg_outline<-readOGR(".", filename) #reading shapefile
|
255
|
X11()
|
256
|
s.range <- c(min(minValue(mm_01)), max(maxValue(mm_01)))
|
257
|
col.breaks <- pretty(s.range, n=50)
|
258
|
lab.breaks <- pretty(s.range, n=5)
|
259
|
temp.colors <- colorRampPalette(c('blue', 'white', 'red'))
|
260
|
plot(mm_01, breaks=col.breaks, col=temp.colors(length(col.breaks)-1),
|
261
|
axis=list(at=lab.breaks, labels=lab.breaks))
|
262
|
plot(reg_outline, add=TRUE)
|
263
|
plot(dat_id,cex=1.5,add=TRUE)
|
264
|
title("Selected stations for comparison",line=3)
|
265
|
title("(Background: mean January LST)", cex=0.5, line=2)
|
266
|
coords<-coordinates(dat_id)
|
267
|
text(x=coords[,1],y=coords[,2],labels=id,cex=0.8, adj=c(0,1),offset=2) #c(0,1) for lower right corner!
|
268
|
savePlot(paste("temporal_profile_station_locations_map",out_prefix,".png", sep=""), type="png")
|
269
|
dev.off()
|
270
|
|
271
|
stat_list<-vector("list",3 ) #list containing the selected stations...
|
272
|
stat_list[[1]]<-ghcn_fus_pred
|
273
|
stat_list[[2]]<-ghcn_cai_pred
|
274
|
stat_list[[3]]<-ghcn_value
|
275
|
ac_temp<-matrix(NA,length(id),2)
|
276
|
|
277
|
#id<-ghcn_value$station #if runinng on all the station...
|
278
|
for (i in 1:length(id)){
|
279
|
m1<-match(id[i],ghcn_fus_pred$station)
|
280
|
m2<-match(id[i],ghcn_cai_pred$station)
|
281
|
m3<-match(id[i],ghcn_value$station)
|
282
|
y1<-as.numeric(ghcn_fus_pred[m1,6:ncol(ghcn_fus_pred)]) #vector containing fusion time series of predictecd tmax
|
283
|
y2<-as.numeric(ghcn_cai_pred[m2,6:ncol(ghcn_cai_pred)]) #vector containing CAI time series of predictecd
|
284
|
y3<-as.numeric(ghcn_value[m3,6:ncol(ghcn_value)]) #vector containing observed time series of predictecd
|
285
|
res2<-y2-y3 #CAI time series residuals
|
286
|
res1<-y1-y3 #fusion time series residuals
|
287
|
x<-1:365
|
288
|
X11(6,15)
|
289
|
plot(x,y1,type="l",col="red",ylab="tmax (C)",xlab="Day of year")
|
290
|
lines(x,y2,col="blue")
|
291
|
lines(x,y3,col="green")
|
292
|
title(paste("temporal profile for station ", id[i],sep=""))
|
293
|
# add a legend
|
294
|
legend("topright",legend=c("fus","CAI","OBS"), cex=1.2, col=c("red","blue","green"),
|
295
|
lty=1, title="tmax")
|
296
|
savePlot(paste("Temporal_profile_",id[i],out_prefix,".png", sep=""), type="png")
|
297
|
|
298
|
### RESIDUALS PLOT
|
299
|
zero<-rep(0,365)
|
300
|
plot(x,res2,type="l",col="blue", ylab="tmax (C)",xlab="Day of year") #res2 contains residuals from cai
|
301
|
lines(x,res1,col="red") #res1 contains fus
|
302
|
lines(x,zero,col="green")
|
303
|
legend("topright",legend=c("fus","CAI"), cex=1.2, col=c("red","blue"),
|
304
|
lty=1)
|
305
|
title(paste("temporal profile for station ", id[i],sep=""))
|
306
|
|
307
|
savePlot(paste("Temporal_profile_res",id[i],out_prefix,".png", sep=""), type="png")
|
308
|
|
309
|
ac_temp[i,1]<-mean(abs(res1),na.rm=T)
|
310
|
ac_temp[i,2]<-mean(abs(res2),na.rm=T)
|
311
|
dev.off()
|
312
|
}
|
313
|
ac_temp<-as.data.frame(ac_temp)
|
314
|
ac_temp$station<-id
|
315
|
names(ac_temp)<-c("fus","CAI","station") #ac_temp contains the MAE per station
|
316
|
|
317
|
### RESIDUALS FOR EVERY STATION...############
|
318
|
|
319
|
id<-ghcn_value$station #if runinng on all the station...
|
320
|
ac_temp2<-matrix(NA,length(id),2)
|
321
|
|
322
|
for (i in 1:length(id)){
|
323
|
m1<-match(id[i],ghcn_fus_pred$station)
|
324
|
m2<-match(id[i],ghcn_cai_pred$station)
|
325
|
m3<-match(id[i],ghcn_value$station)
|
326
|
y1<-as.numeric(ghcn_fus_pred[m1,6:ncol(ghcn_fus_pred)])
|
327
|
y2<-as.numeric(ghcn_cai_pred[m2,6:ncol(ghcn_cai_pred)])
|
328
|
y3<-as.numeric(ghcn_value[m3,6:ncol(ghcn_value)])
|
329
|
res2<-y2-y3
|
330
|
res1<-y1-y3
|
331
|
ac_temp2[i,1]<-mean(abs(res1),na.rm=T)
|
332
|
ac_temp2[i,2]<-mean(abs(res2),na.rm=T)
|
333
|
}
|
334
|
|
335
|
|
336
|
ac_temp2<-as.data.frame(ac_temp2)
|
337
|
ac_temp2$station<-id
|
338
|
names(ac_temp2)<-c("fus","CAI","station")
|
339
|
|
340
|
ac_temp2<-ac_temp2[order(ac_temp2$fus,ac_temp2$CAI), ]
|
341
|
ghcn_MAE<-merge(ghcn_locs,ac_temp2,by.x=station,by.y=station)
|
342
|
|
343
|
########### TRANSECT-- DAY OF YEAR PLOT...#########
|
344
|
|
345
|
id<-c("USW00024284","USC00354126","USC00358536","USC00354835",
|
346
|
"USC00356252","USC00359316","USC00358246","USC00350694",
|
347
|
"USC00350699","USW00024230","USC00353542")
|
348
|
id_order<-1:11
|
349
|
m<-match(id,ghcn_locs$station)
|
350
|
dat_id<-ghcn_locs[m,] #creating new subset
|
351
|
#dat_id<-subset(ghcn_locs[gj])
|
352
|
|
353
|
filename<-sub(".shp","",infile6) #Removing the extension from file.
|
354
|
reg_outline<-readOGR(".", filename) #reading shapefile
|
355
|
X11()
|
356
|
s.range <- c(min(minValue(mm_01)), max(maxValue(mm_01)))
|
357
|
col.breaks <- pretty(s.range, n=50)
|
358
|
lab.breaks <- pretty(s.range, n=5)
|
359
|
temp.colors <- colorRampPalette(c('blue', 'white', 'red'))
|
360
|
plot(mm_01, breaks=col.breaks, col=temp.colors(length(col.breaks)-1),
|
361
|
axis=list(at=lab.breaks, labels=lab.breaks))
|
362
|
plot(reg_outline, add=TRUE)
|
363
|
plot(dat_id,cex=1.5,add=TRUE)
|
364
|
title("Selected stations for comparison",line=3)
|
365
|
title("(Background: mean January LST)", cex=0.5, line=2)
|
366
|
coords<-coordinates(dat_id)
|
367
|
text(x=coords[,1],y=coords[,2],labels=as.character(id_order),cex=1.5, adj=c(0,1),offset=2) #c(0,1) for lower right corner!
|
368
|
savePlot(paste("temporal_profile_station_locations_map",out_prefix,".png", sep=""), type="png")
|
369
|
dev.off()
|
370
|
|
371
|
m1<-match(id,ghcn_fus_pred$station)
|
372
|
m2<-match(id,ghcn_cai_pred$station)
|
373
|
m3<-match(id,ghcn_value$station)
|
374
|
ghcn_dsub<-subset(ghcn,ghcn$station==id)
|
375
|
all.equal(m1,m2,m3) #OK order is the same
|
376
|
date_selection<-c("01-01-2010","01-09-2010")
|
377
|
#date_str<-gsub(date_selection,"-","")
|
378
|
date_str<-c("20100101","20100901")
|
379
|
covar_dsub<-subset(ghcn_dsub,ghcn_dsub$date==date_str[i],select=c("station","ELEV_SRTM","LC1"))
|
380
|
|
381
|
date_pred<-as.Date(date_selection)
|
382
|
#mo<-as.integer(strftime(date_pred, "%m")) # current month of the date being processed
|
383
|
doy_pred<-(strptime(date_pred, "%d-%m-%Y")$yday+1)
|
384
|
|
385
|
for (i in 1:length(date_pred)){
|
386
|
doy<-doy_pred[i]+5 #column label
|
387
|
doy<-243+5
|
388
|
stat_subset<-cbind(id,ghcn_fus_pred[m1,doy],ghcn_cai_pred[m1,doy],ghcn_value[m1,doy])
|
389
|
colnames(stat_subset)<-c("station","fus","cai","value")
|
390
|
stat_subset<-as.data.frame(stat_subset)
|
391
|
for(j in 2:4){ # start of the for loop #1
|
392
|
stat_subset[,j]<-as.numeric(as.character(stat_subset[,j]))
|
393
|
}
|
394
|
X11()
|
395
|
plot(1:11,stat_subset$value,type="b",col="green",ylab="tmax",xlab="station transtect number")
|
396
|
# xlabels())
|
397
|
lines(1:11,stat_subset$fus,type="b",col="red")
|
398
|
lines(1:11,stat_subset$cai,type="b",col="blue")
|
399
|
|
400
|
legend("bottomright",legend=c("obs","fus","cai"), cex=1.2, col=c("green","red","blue"),
|
401
|
lty=1, title="tmax")
|
402
|
title(paste("Daily tmax prediction ",date_selection[i],sep=" "))
|
403
|
savePlot(paste("transect_profile_tmax_",date_str[i],out_prefix,".png", sep=""), type="png")
|
404
|
dev.off()
|
405
|
}
|
406
|
|
407
|
##############################################
|
408
|
########## USING TEMPORAL IMAGES...############
|
409
|
|
410
|
date_list<-vector("list", length(l_f))
|
411
|
for (k in 1:length(l_f)){
|
412
|
tmp<-(unlist(strsplit(l_f[k],"_"))) #spliting file name to obtain the prediction date
|
413
|
date_list[k]<-tmp[4]
|
414
|
}
|
415
|
|
416
|
date_list2<-vector("list", length(l_f2))
|
417
|
for (k in 1:length(l_f2)){
|
418
|
tmp<-(unlist(strsplit(l_f2[k],"_"))) #spliting file name to obtain the prediction date
|
419
|
date_list2[k]<-tmp[4]
|
420
|
}
|
421
|
|
422
|
setdiff(date_list,date_list2)
|
423
|
all.equal(date_list,date_list2) #This checks that both lists are equals
|
424
|
|
425
|
nel<-length(gam_fus_mod1)
|
426
|
list_fus_data_s<-vector("list", nel)
|
427
|
list_cai_data_s<-vector("list", nel)
|
428
|
list_fus_data_v<-vector("list", nel)
|
429
|
list_cai_data_v<-vector("list", nel)
|
430
|
|
431
|
list_fus_data<-vector("list", nel)
|
432
|
list_cai_data<-vector("list", nel)
|
433
|
|
434
|
list_dstspat_er<-vector("list", nel)
|
435
|
list_dstspat_er2<-vector("list", nel)
|
436
|
k=1
|
437
|
|
438
|
for (k in 1:nel){
|
439
|
#for (k in 1:365){
|
440
|
|
441
|
#Start loop over the full year!!!
|
442
|
names(gam_fus_mod1[[k]])
|
443
|
data_s<-gam_fus_mod1[[k]]$data_s
|
444
|
data_v<-gam_fus_mod1[[k]]$data_v
|
445
|
|
446
|
date_proc<-unique(data_s$date)
|
447
|
index<-match(as.character(date_proc),unlist(date_list)) #find the correct date..
|
448
|
#raster_pred<-raster(rp_raster,index)
|
449
|
|
450
|
#####second series added
|
451
|
data_v2<-fus_CAI_mod[[k]]$data_v
|
452
|
data_s2<-fus_CAI_mod[[k]]$data_s
|
453
|
|
454
|
date_proc<-unique(data_s$date)
|
455
|
index<-match(as.character(date_proc),unlist(date_list)) #find the correct date..
|
456
|
#raster_pred<-raster(rp_raster,index)
|
457
|
|
458
|
###Checking if training and validation have the same columns
|
459
|
nd<-setdiff(names(data_s),names(data_v))
|
460
|
nd2<-setdiff(names(data_s2),names(data_v2))
|
461
|
|
462
|
data_v[[nd]]<-NA #daily_delta is not the same
|
463
|
|
464
|
data_v$training<-rep(0,nrow(data_v))
|
465
|
data_v2$training<-rep(0,nrow(data_v2))
|
466
|
data_s$training<-rep(1,nrow(data_s))
|
467
|
data_s2$training<-rep(1,nrow(data_s2))
|
468
|
|
469
|
list_fus_data_s[[k]]<-data_s
|
470
|
list_cai_data_s[[k]]<-data_s2
|
471
|
list_fus_data_v[[k]]<-data_v
|
472
|
list_cai_data_v[[k]]<-data_v2
|
473
|
list_fus_data[[k]]<-rbind(data_v,data_s)
|
474
|
list_cai_data[[k]]<-rbind(data_v2,data_s2)
|
475
|
|
476
|
d_s_v<-matrix(0,nrow(data_v),nrow(data_s))
|
477
|
for(i in 1:nrow(data_s)){
|
478
|
pt<-data_s[i,]
|
479
|
d_pt<-(spDistsN1(data_v,pt,longlat=FALSE))/1000 #Distance to stataion i in km
|
480
|
d_s_v[,i]<-d_pt
|
481
|
}
|
482
|
|
483
|
d_s_v2<-matrix(0,nrow(data_v2),nrow(data_s2))
|
484
|
for(i in 1:nrow(data_s2)){
|
485
|
pt2<-data_s2[i,]
|
486
|
d_pt2<-(spDistsN1(data_v2,pt2,longlat=FALSE))/1000 #Distance to stataion i in km
|
487
|
d_s_v2[,i]<-d_pt2
|
488
|
}
|
489
|
|
490
|
#Create data.frame with position, ID, dst and residuals...YOU HAVE TO DO IT SEPARATELY FOR EACH MODEL!!!
|
491
|
#Do first fusion and then CAI
|
492
|
pos<-vector("numeric",nrow(data_v))
|
493
|
y<-vector("numeric",nrow(data_v))
|
494
|
dst<-vector("numeric",nrow(data_v))
|
495
|
|
496
|
pos2<-vector("numeric",nrow(data_v))
|
497
|
y2<-vector("numeric",nrow(data_v))
|
498
|
dst2<-vector("numeric",nrow(data_v))
|
499
|
|
500
|
for (i in 1:nrow(data_v)){
|
501
|
pos[i]<-match(min(d_s_v[i,]),d_s_v[i,])
|
502
|
dst[i]<-min(d_s_v[i,])
|
503
|
}
|
504
|
|
505
|
for (i in 1:nrow(data_v2)){
|
506
|
pos2[i]<-match(min(d_s_v2[i,]),d_s_v2[i,])
|
507
|
dst2[i]<-min(d_s_v2[i,])
|
508
|
}
|
509
|
|
510
|
res_fus<-data_v$res_mod9
|
511
|
res_CAI<-data_v2$res_mod9
|
512
|
|
513
|
dstspat_er<-as.data.frame(cbind(as.vector(data_v$id),as.vector(data_s$id[pos]),pos, data_v$lat, data_v$lon, data_v$x_OR83M,data_v$y_OR83M,
|
514
|
dst,
|
515
|
res_fus))
|
516
|
dstspat_er2<-as.data.frame(cbind(as.vector(data_v2$id),as.vector(data_s2$id[pos2]),pos2, data_v2$lat, data_v2$lon, data_v2$x_OR83M,data_v2$y_OR83M,
|
517
|
dst2,
|
518
|
res_CAI))
|
519
|
names(dstspat_er2)[1:7]<-c("v_id","s_id","pos","lat","lon","x_OR83M","y_OR83M")
|
520
|
|
521
|
names(dstspat_er)[1:7]<-c("v_id","s_id","pos","lat","lon","x_OR83M","y_OR83M")
|
522
|
# names(dstspat_er)[10:15]<-c("res_mod1","res_mod2","res_mod3","res_mod4","res_mod5","res_CAI")
|
523
|
#names(dstspat_er)[10:15]<-c("res_mod1","res_mod2","res_mod3","res_mod4","res_mod5","res_CAI")
|
524
|
list_dstspat_er[[k]]<-dstspat_er
|
525
|
list_dstspat_er2[[k]]<-dstspat_er2
|
526
|
|
527
|
}
|
528
|
save(list_dstspat_er,file="spat1_ac5.RData")
|
529
|
save(list_dstspat_er2,file="spat2_ac5.RData")
|
530
|
|
531
|
#obj_tmp2<-load_obj("spat_ac4.RData")
|
532
|
save(list_fus_data,file="list_fus_data_combined.RData")
|
533
|
save(list_cai_data,file="list_cai_data_combined.RData")
|
534
|
|
535
|
save(list_fus_data_s,file="list_fus_data_s_combined.RData")
|
536
|
save(list_cai_data_s,file="list_cai_data_s_combined.RData")
|
537
|
save(list_fus_data_v,file="list_fus_data_v_combined.RData")
|
538
|
save(list_cai_data_v,file="list_cai_data_v_combined.RData")
|
539
|
|
540
|
for (k in 1:nel){
|
541
|
data_s<-as.data.frame(list_fus_data_s[[k]])
|
542
|
data_v<-as.data.frame(list_fus_data_v[[k]])
|
543
|
list_fus_data[[k]]<-rbind(data_s,data_v)
|
544
|
data_s2<-as.data.frame(list_cai_data_s[[k]])
|
545
|
data_v2<-as.data.frame(list_cai_data_v[[k]])
|
546
|
list_cai_data[[k]]<-rbind(data_s2,data_v2)
|
547
|
}
|
548
|
data_fus<-do.call(rbind.fill,list_fus_data)
|
549
|
data_cai<-do.call(rbind.fill,list_cai_data)
|
550
|
|
551
|
data_fus_melt<-melt(data_fus,
|
552
|
measure=c("x_OR83M","y_OR83M","res_fus","res_mod1","res_mod2","res_mod3","res_mod4","res_mod5","pred_fus","dailyTmax","TMax","LST","training"),
|
553
|
id=c("id","date"),
|
554
|
na.rm=F)
|
555
|
data_fus_cast<-cast(data_fus_melt,id+date~variable,mean)
|
556
|
|
557
|
test_dst<-list_dstspat_er
|
558
|
test<-do.call(rbind,list_dstspat_er)
|
559
|
|
560
|
test_dst2<-list_dstspat_er2
|
561
|
test2<-do.call(rbind,list_dstspat_er2)
|
562
|
|
563
|
for(i in 4:ncol(test)){ # start of the for loop #1
|
564
|
test[,i]<-as.numeric(as.character(test[,i]))
|
565
|
}
|
566
|
|
567
|
for(i in 4:ncol(test2)){ # start of the for loop #1
|
568
|
test2[,i]<-as.numeric(as.character(test2[,i]))
|
569
|
}
|
570
|
|
571
|
# Plot results
|
572
|
plot(test$dst,abs(test$res_fus))
|
573
|
limit<-seq(0,150, by=10)
|
574
|
tmp<-cut(test$dst,breaks=limit)
|
575
|
tmp2<-cut(test2$dst,breaks=limit)
|
576
|
|
577
|
erd1<-tapply(test$res_fus,tmp, mean)
|
578
|
erd2<-as.numeric(tapply(abs(test$res_fus),tmp, mean))
|
579
|
plot(erd2)
|
580
|
|
581
|
erd2_CAI<-tapply(abs(test2$res_CAI),tmp2, mean)
|
582
|
n<-tapply(abs(test$res_fus),tmp, length)
|
583
|
n2<-tapply(abs(test2$res_CAI),tmp2, length)
|
584
|
|
585
|
distance<-seq(5,145,by=10)
|
586
|
|
587
|
X11()
|
588
|
plot(distance,erd2,ylim=c(1,3), type="b", col="red",ylab=" Average MAE",
|
589
|
xlab="distance to closest training station (km)")
|
590
|
lines(distance,erd2_CAI,col="grey")
|
591
|
title("MAE in terms of distance to closest station GAM and FUSION")
|
592
|
legend("bottomright",legend=c("fus","CAI"), cex=1.2, col=c("red","grey"),
|
593
|
lty=1, title="MAE")
|
594
|
savePlot(paste("Comparison_models_er_spat",out_prefix,".png", sep=""), type="png")
|
595
|
dev.off()
|
596
|
|
597
|
means <- erd2_CAI
|
598
|
means2<- erd2
|
599
|
stdev <-tapply(abs(test2$res_CAI),tmp2, sd)
|
600
|
stdev2 <-tapply(abs(test$res_fus),tmp, sd)
|
601
|
|
602
|
ciw <- qt(0.975, n) * stdev / sqrt(n)
|
603
|
ciw2 <- qt(0.975, n) * stdev2 / sqrt(n)
|
604
|
|
605
|
X11()
|
606
|
plotCI(y=means, x=distance, uiw=ciw, col="black", main=" CAI: MAE and distance to clostest training station", barcol="blue", lwd=1)
|
607
|
lines(distance,erd2_CAI,col="grey")
|
608
|
points(distance,erd2_CAI,col="grey")
|
609
|
savePlot(paste("CI_CAI_er_spat_",out_prefix,".png", sep=""), type="png")
|
610
|
dev.off()
|
611
|
|
612
|
X11()
|
613
|
plotCI(y=means2, x=distance, uiw=ciw2, col="black", main=" FUSION: MAE and distance to clostest training station", barcol="blue", lwd=1)
|
614
|
lines(distance,erd2,col="black")
|
615
|
savePlot(paste("CI_fusion_er_spat_",out_prefix,".png", sep=""), type="png")
|
616
|
dev.off()
|
617
|
|
618
|
X11()
|
619
|
barplot(n,names.arg=as.character(distance))
|
620
|
savePlot(paste("Barplot_freq_er_spat_",out_prefix,".png", sep=""), type="png")
|
621
|
dev.off()
|
622
|
|
623
|
############################################################
|
624
|
############## PART III #############
|
625
|
### Average MAE per station and coarse grid box (0.5 deg)
|
626
|
|
627
|
#For all stations
|
628
|
|
629
|
ghcn$station
|
630
|
|
631
|
# For validation and training stations...
|
632
|
|
633
|
test$abs_res_fus<-abs(test$res_fus)
|
634
|
test2$abs_res_CAI<-abs(test2$res_CAI)
|
635
|
|
636
|
station_melt<-melt(test,
|
637
|
measure=c("x_OR83M","y_OR83M","res_mod_v","res_mod1","res_mod2","res_mod3","res_mod4","res_mod5","abs_res_fus","abs_res_CAI"),
|
638
|
id=c("v_id"),
|
639
|
na.rm=F)
|
640
|
station_v_er<-cast(station_melt,v_id~variable,mean)
|
641
|
#station_v_er2<-as.data.frame(station_v_er)
|
642
|
station_v_er<-as.data.frame(station_v_er)
|
643
|
oc<-vector("numeric",nrow(station_v_er))
|
644
|
oc<-oc+1
|
645
|
station_v_er$oc<-oc
|
646
|
|
647
|
unique(ghcn$station)
|
648
|
|
649
|
coords<- station_v_er[,c('x_OR83M','y_OR83M')]
|
650
|
coordinates(station_v_er)<-coords
|
651
|
proj4string(station_v_er)<-CRS #Need to assign coordinates...
|
652
|
|
653
|
bubble(station_v_er,"abs_res_fus")
|
654
|
list_agg_MAE<-vector("list",nel)
|
655
|
list_agg_RMSE<-vector("list",nel)
|
656
|
list_density_training<-vector("list",nel)
|
657
|
list_density_station<-vector("list",nel)
|
658
|
|
659
|
for (k in 1:nel){
|
660
|
data_s<-as.data.frame(list_fus_data_s[[k]])
|
661
|
data_v<-as.data.frame(list_fus_data_v[[k]])
|
662
|
list_fus_data[[k]]<-rbind(data_s,data_v)
|
663
|
data_s2<-as.data.frame(list_cai_data_s[[k]])
|
664
|
data_v2<-as.data.frame(list_cai_data_v[[k]])
|
665
|
list_cai_data[[k]]<-rbind(data_s2,data_v2)
|
666
|
}
|
667
|
|
668
|
############ GRID BOX AVERAGING ####################
|
669
|
####### Create the averaged grid box...##############
|
670
|
|
671
|
rast_agg<-aggregate(raster_pred,fact=50,fun=mean,na.rm=TRUE) #Changing the raster resolution by aggregation factor
|
672
|
|
673
|
ghcn_sub<-as.data.frame(subset(ghcn, select=c("station","x_OR83M","y_OR83M")))
|
674
|
ghcn_sub_melt<-melt(ghcn_sub,
|
675
|
measure=c("x_OR83M","y_OR83M"),
|
676
|
id=c("station"),
|
677
|
na.rm=F)
|
678
|
ghcn_stations<-as.data.frame(cast(ghcn_sub_melt,station~variable,mean))
|
679
|
coords<- ghcn_stations[,c('x_OR83M','y_OR83M')]
|
680
|
coordinates(ghcn_stations)<-coords
|
681
|
proj4string(ghcn_stations)<-proj_str #Need to assign coordinates...
|
682
|
oc_all<-vector("numeric",nrow(ghcn_stations))
|
683
|
oc_all<-oc_all+1
|
684
|
|
685
|
ghcn_stations$oc_all<-oc_all
|
686
|
rast_oc_all<-rasterize(ghcn_stations,rast_agg,"oc_all",na.rm=TRUE,fun=sum)
|
687
|
ac_agg50$oc_all<-values(rast_oc_all)
|
688
|
|
689
|
plot(rast_oc_all, main="Number of stations in coarsened 50km grid")
|
690
|
plot(reg_outline, add=TRUE)
|
691
|
fdens_all50<-as.data.frame(freq(rast_oc_all))
|
692
|
tot50<-sum(fdens_all50$count[1:(nrow(fdens_all50)-1)])
|
693
|
percent<-(fdens_all50$count/tot50)*100
|
694
|
percent[length(percent)]<-NA
|
695
|
fdens_all50$percent<-percent
|
696
|
#list_agg_MAE<-vector("list",nel)
|
697
|
#list_agg_RMSE<-vector("list",nel)
|
698
|
#list_density_training<-vector("list",nel)
|
699
|
#list_density_station<-vector("list",nel)
|
700
|
#start loop here for grid box aggregation
|
701
|
list_density_fus<-vector("list",nel)
|
702
|
list_density_cai<-vector("list",nel)
|
703
|
|
704
|
nel<-365
|
705
|
for (k in 1:nel){
|
706
|
|
707
|
data_s<-list_fus_data_s[[k]] #Extracting the relevant spdf from the list: this is 1050?
|
708
|
data_s2<-list_cai_data_s[[k]]
|
709
|
data_v<-list_fus_data_v[[k]]
|
710
|
data_v2<-list_cai_data_v[[k]]
|
711
|
|
712
|
data_v$abs_res_fus<-abs(data_v$res_mod9)
|
713
|
data_v2$abs_res_CAI<-abs(data_v2$res_mod9)
|
714
|
data_s$abs_res_fus<-abs(data_s$res_mod9)
|
715
|
data_s2$abs_res_CAI<-abs(data_s2$res_mod9)
|
716
|
data_s$oc<-rep(1,nrow(data_s))
|
717
|
data_s2$oc<-rep(1,nrow(data_s2))
|
718
|
|
719
|
#Computing MAE per grid box
|
720
|
rast_MAE_fus<-rasterize(data_v,rast_agg,"abs_res_fus",na.rm=TRUE,fun=mean)
|
721
|
rast_MAE_cai<-rasterize(data_v2,rast_agg,"abs_res_CAI",na.rm=TRUE,fun=mean)
|
722
|
#Computing density of station
|
723
|
rast_oc<-rasterize(data_s,rast_agg,"oc",na.rm=TRUE,fun=sum)
|
724
|
rast_oc2<-rasterize(data_s2,rast_agg,"oc",na.rm=TRUE,fun=sum)
|
725
|
|
726
|
#Creating plots adding to the list...to get a data frame...
|
727
|
ac_agg50<-as.data.frame(values(rast_oc))
|
728
|
ac_agg50$MAE_fus<-as.numeric(values(rast_MAE_fus))
|
729
|
ac_agg50$MAE_cai<-as.numeric(values(rast_MAE_cai))
|
730
|
names(ac_agg50)<-c("oc","MAE_fus","MAE_cai")
|
731
|
|
732
|
#ghcn_sub<-as.data.frame(subset(ghcn, select=c("station","x_OR83M","y_OR83M")))
|
733
|
#ghcn_sub_melt<-melt(ghcn_sub,
|
734
|
# measure=c("x_OR83M","y_OR83M"),
|
735
|
# id=c("station"),
|
736
|
# na.rm=F)
|
737
|
#ghcn_stations<-as.data.frame(cast(ghcn_sub_melt,station~variable,mean))
|
738
|
#coords<- ghcn_stations[,c('x_OR83M','y_OR83M')]
|
739
|
#coordinates(ghcn_stations)<-coords
|
740
|
#proj4string(ghcn_stations)<-proj_str #Need to assign coordinates...
|
741
|
#oc_all<-vector("numeric",nrow(ghcn_stations))
|
742
|
#oc_all<-oc_all+1
|
743
|
|
744
|
#ghcn_stations$oc_all<-oc_all
|
745
|
#rast_oc_all<-rasterize(ghcn_stations,rast_agg,"oc_all",na.rm=TRUE,fun=sum)
|
746
|
#ac_agg50$oc_all<-values(rast_oc_all)
|
747
|
|
748
|
td1<-aggregate(MAE_fus~oc,data=ac_agg50,mean)
|
749
|
td2<-aggregate(MAE_cai~oc,data=ac_agg50,mean)
|
750
|
td<-merge(td1,td2,by="oc")
|
751
|
|
752
|
td1_all<-aggregate(MAE_fus~oc_all,data=ac_agg50,mean)
|
753
|
td2_all<-aggregate(MAE_cai~oc_all,data=ac_agg50,mean)
|
754
|
td_all<-merge(td1_all,td2_all,by="oc_all")
|
755
|
|
756
|
plot(MAE_fus~oc,data=td,type="b")
|
757
|
lines(td$oc,td$MAE_cai, type="b", lwd=1.5,co="red")
|
758
|
plot(MAE_fus~oc_all,data=td_all,type="b")
|
759
|
lines(td_all$oc_all,td_all$MAE_cai, type="b", lwd=1.5,co="red")
|
760
|
|
761
|
filename<-sub(".shp","",infile6) #Removing the extension from file.
|
762
|
reg_outline<-readOGR(".", filename) #reading shapefile
|
763
|
plot(rast_MAE_fus, main="Fusion MAE in coarsened 50km grid")
|
764
|
plot(reg_outline, add=TRUE)
|
765
|
|
766
|
plot(rast_MAE_cai, main="CAI MAE in coarsened 50km grid")
|
767
|
plot(reg_outline, add=TRUE)
|
768
|
|
769
|
plot(rast_oc, main="Number of val stations in coarsened 50km grid")
|
770
|
plot(reg_outline, add=TRUE)
|
771
|
plot(rast_oc_t, main="Number of training stations in coarsened 50km grid")
|
772
|
plot(reg_outline, add=TRUE)
|
773
|
|
774
|
#MAKE IT AN OBJECT for future function return...
|
775
|
|
776
|
#list(rast_MAE,rast_RMSE,rast_oc,rast_all)
|
777
|
list_density_fus[[k]]<-list(rast_MAE_fus,rast_oc,rast_oc_all,ac_agg50)
|
778
|
list_density_cai[[k]]<-list(rast_MAE_cai,rast_oc,rast_oc_all,ac_agg50)
|
779
|
}
|
780
|
|
781
|
#meean over stack oc
|
782
|
#mean over stack MAE
|
783
|
do.call(rbind,list_density_)
|
784
|
list_var_stat<-vector("list", 365)
|
785
|
#list_var_stat<-vector("list", 2)
|
786
|
#k=2
|
787
|
|
788
|
for (k in 1:length(l_f)){
|
789
|
|
790
|
raster_pred<-raster(l_f[[k]])
|
791
|
layerNames(raster_pred)<-"fus"
|
792
|
projection(raster_pred)<-proj_str
|
793
|
|
794
|
raster_pred2<-raster(l_f2[[k]])
|
795
|
layerNames(raster_pred2)<-"fus"
|
796
|
projection(raster_pred2)<-proj_str
|
797
|
|
798
|
tmp_rast<-mask(raster_pred2,raster_pred)
|
799
|
raster_pred2<-tmp_rast
|
800
|
|
801
|
t1<-cellStats(raster_pred,na.rm=TRUE,stat=sd) #Calculating the standard deviation for the
|
802
|
t2<-cellStats(raster_pred2,na.rm=TRUE,stat=sd)
|
803
|
|
804
|
m1<-Moran(raster_pred,w=3) #Calculating Moran's I with window of 3 an default Queen's case
|
805
|
m2<-Moran(tmp_rast,w=3) #Calculating Moran's I with window of 3 an default Queen's case for prediction 2 (CAI)
|
806
|
stat<-as.data.frame(t(c(m1,m2,t1,t2)))
|
807
|
names(stat)<-c("moran_fus","moran_CAI","sd_fus","sd_CAI")
|
808
|
list_var_stat[[k]]<-stat
|
809
|
}
|
810
|
|
811
|
var_stat<-do.call(rbind,list_var_stat)
|
812
|
|
813
|
|
814
|
pos<-match("ELEV_SRTM",layerNames(s_raster)) #Find column with name "value"
|
815
|
elev<-raster(s_raster,layer=pos) #Select layer from stack
|
816
|
elev<-mask(elev,raster_pred)
|
817
|
te<-cellStats(elev,na.rm=TRUE,stat=sd)
|
818
|
|
819
|
pos<-match("mm_12",layerNames(s_raster)) #Find column with name "value"
|
820
|
m_12<-raster(s_raster,layer=pos) #Select layer from stack
|
821
|
m_LST<-Moran(m_12,w=3)
|
822
|
m_e<-Moran(elev,w=3)
|
823
|
m_12<-m_12-273.15
|
824
|
plot(MAE_fus~oc,data=td,type="b")
|
825
|
lines(td$oc,td$MAE_CAI, type="b", lwd=1.5,co="red")
|
826
|
|
827
|
data_dist<-as.data.frame(cbind(distance,erd2,erd2_mod1,erd2_mod2,erd2_mod3,erd2_mod4,erd2_mod5,erd2_CAI,n))
|
828
|
rownames(data_dist)<-NULL
|
829
|
|
830
|
############# PART IV COMPARISON OF SPATIAL PATTERN BY EXAMING MAPS OF PREDICTION
|
831
|
#PLOTING CAI AND FUSION TO COMPARE
|
832
|
|
833
|
infile2<-"list_10_dates_04212012.txt" #List of 10 dates for the regression
|
834
|
dates2<-read.table(paste(path,"/",infile2,sep=""), sep="") #Column 1 contains the names of raster files
|
835
|
date_list2<-as.list(as.character(dates2[,1]))
|
836
|
names_statistics<-c("mean","sd","min","max")
|
837
|
stat_val_m<-matrix(NA,nrow=length(date_list2),ncol=length(names_statistics))
|
838
|
colnames(stat_val_m)<-names_statistics
|
839
|
rownames(stat_val_m)<-date_list2
|
840
|
stat_val_m<-as.data.frame(stat_val_m)
|
841
|
|
842
|
for (k in 1:length(date_list2)){
|
843
|
|
844
|
date_proc2<-date_list2[[k]]
|
845
|
#date_proc<-date_list[[k]]
|
846
|
index<-match(as.character(date_proc2),unlist(date_list)) #find the correct date... in the 365 stack
|
847
|
#raster_pred<-raster(rp_raster,index)
|
848
|
raster_pred1<-raster(l_f[[index]]) # Fusion image
|
849
|
projection(raster_pred1)<-proj_str
|
850
|
raster_pred1<-mask(raster_pred1,mask_land_NA)
|
851
|
|
852
|
raster_pred2<-raster(l_f2[[index]]) # CAI image
|
853
|
projection(raster_pred2)<-proj_str
|
854
|
raster_pred2<-mask(raster_pred2,mask_land_NA)
|
855
|
|
856
|
predictions <- stack(raster_pred1,raster_pred2)
|
857
|
layerNames(predictions)<-c(paste('fusion',date_list2[[k]],sep=" "),paste('CAI',date_list2[[k]],sep=" "))
|
858
|
# use overall min and max values to generate an nice, consistent set
|
859
|
# of breaks for both colors (50 values) and legend labels (5 values)
|
860
|
s.range <- c(min(minValue(predictions)), max(maxValue(predictions)))
|
861
|
col.breaks <- pretty(s.range, n=50)
|
862
|
lab.breaks <- pretty(s.range, n=5)
|
863
|
temp.colors <- colorRampPalette(c('blue', 'white', 'red'))
|
864
|
|
865
|
# plot using these (common) breaks; note use of _reverse_ heat.colors,
|
866
|
# making it so that larger numbers are redder
|
867
|
X11(6,12)
|
868
|
#plot(predictions, breaks=col.breaks, col=rev(heat.colors(length(col.breaks)-1)),
|
869
|
# axis=list(at=lab.breaks, labels=lab.breaks))
|
870
|
plot(predictions, breaks=col.breaks, col=temp.colors(length(col.breaks)-1),
|
871
|
axis=list(at=lab.breaks, labels=lab.breaks))
|
872
|
#plot(reg_outline, add=TRUE)
|
873
|
savePlot(paste("comparison_raster1_CAI_fusion_tmax_prediction_",date_list2[[k]],out_prefix,".png", sep=""), type="png")
|
874
|
|
875
|
|
876
|
stat_val_m$mean[i]<-cellStats(raster_pred1,na.rm=TRUE,stat=mean) #Calculating the standard deviation for the
|
877
|
t1<-cellStats(raster_pred1,na.rm=TRUE,stat=mean) #Calculating the standard deviation for the
|
878
|
t2<-cellStats(raster_pred2,na.rm=TRUE,stat=mean) #Calculating the standard deviation for the
|
879
|
t1<-cellStats(raster_pred1,na.rm=TRUE,stat=sd) #Calculating the standard deviation for the
|
880
|
t2<-cellStats(raster_pred2,na.rm=TRUE,stat=sd) #Calculating the standard deviation for the
|
881
|
t1<-cellStats(raster_pred1,na.rm=TRUE,stat=min) #Calculating the standard deviation for the
|
882
|
t2<-cellStats(raster_pred2,na.rm=TRUE,stat=min) #Calculating the standard deviation for the
|
883
|
t1<-cellStats(raster_pred1,na.rm=TRUE,stat=max) #Calculating the standard deviation for the
|
884
|
t2<-cellStats(raster_pred2,na.rm=TRUE,stat=max) #Calculating the standard deviation for the
|
885
|
|
886
|
hist(predictions,freq=FALSE,maxpixels=ncells(predictions),xlabel="tmax (C)")
|
887
|
savePlot(paste("comparison_histo_CAI_fusion_tmax_prediction_",date_list2[[k]],out_prefix,".png", sep=""), type="png")
|
888
|
#plot(predictions)
|
889
|
dev.off()
|
890
|
|
891
|
X11(6,12)
|
892
|
diff<-raster_pred2-raster_pred1
|
893
|
s.range <- c(min(minValue(diff)), max(maxValue(diff)))
|
894
|
col.breaks <- pretty(s.range, n=50)
|
895
|
lab.breaks <- pretty(s.range, n=5)
|
896
|
temp.colors <- colorRampPalette(c('blue', 'white', 'red'))
|
897
|
plot(diff, breaks=col.breaks, col=temp.colors(length(col.breaks)-1),
|
898
|
axis=list(at=lab.breaks, labels=lab.breaks))
|
899
|
title(paste("Difference between CAI and fusion for ",date_list2[[k]],sep=""))
|
900
|
savePlot(paste("comparison_diff_CAI_fusion_tmax_prediction_",date_list2[[k]],out_prefix,".png", sep=""), type="png")
|
901
|
dev.off()
|
902
|
|
903
|
}
|
904
|
|
905
|
write.table(data_dist,file=paste("data_dist_",out_prefix,".txt",sep=""),sep=",")
|
906
|
write.table(test,file=paste("ac_spat_dist",out_prefix,".txt",sep=""),sep=",")
|
907
|
write.table(var_stat,file=paste("moran_var_stat_",out_prefix,".txt",sep=""),sep=",")
|
908
|
write.table(td,file=paste("MAE_density_station_",out_prefix,".txt",sep=""),sep=",")
|
909
|
write.table(td_all,file=paste("MAE_density_station_all_",out_prefix,".txt",sep=""),sep=",")
|
910
|
|
911
|
symbols(c(2e5, 4e5), c(2e5, 3e5), circles=rep(2e4, 2), inches=FALSE, add=TRUE)
|
912
|
text(c(2e5, 4e5), c(2e5, 3e5), labels=1:2,cex=0.5)
|
913
|
points(coordinates(pts), type="c")
|
914
|
text(coordinates(pts), labels=9:11, cex=0.8)
|
915
|
points(coordinates(pts), type="b", pch=as.character(1:length(pts))
|
916
|
points(coordinates(pts), type="b", pch=as.character(9:11)
|
917
|
|
918
|
########### END OF THE SCRIPT #############
|