1
|
################## CLIMATE INTERPOLATION FUSION METHOD #######################################
|
2
|
############################ Merging LST and station data ##########################################
|
3
|
#This script interpolates tmax values using MODIS LST and GHCND station data #
|
4
|
#interpolation area. It requires the text file of stations and a shape file of the study area. #
|
5
|
#Note that the projection for both GHCND and study area is lonlat WGS84. #
|
6
|
#AUTHOR: Brian McGill #
|
7
|
#DATE: 06/19/212 #
|
8
|
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#363-- #
|
9
|
###################################################################################################
|
10
|
|
11
|
###Loading R library and packages
|
12
|
library(gtools) # loading some useful tools
|
13
|
library(mgcv) # GAM package by Simon Wood
|
14
|
library(sp) # Spatial pacakge with class definition by Bivand et al.
|
15
|
library(spdep) # Spatial pacakge with methods and spatial stat. by Bivand et al.
|
16
|
library(rgdal) # GDAL wrapper for R, spatial utilities
|
17
|
library(gstat) # Kriging and co-kriging by Pebesma et al.
|
18
|
library(fields) # NCAR Spatial Interpolation methods such as kriging, splines
|
19
|
library(raster) # Hijmans et al. package for raster processing
|
20
|
### Parameters and argument
|
21
|
|
22
|
infile1<- "ghcn_or_tmax_covariates_06262012_OR83M.shp" #GHCN shapefile containing variables for modeling 2010
|
23
|
infile2<-"list_10_dates_04212012.txt" #List of 10 dates for the regression
|
24
|
#infile2<-"list_365_dates_04212012.txt"
|
25
|
infile3<-"LST_dates_var_names.txt" #LST dates name
|
26
|
infile4<-"models_interpolation_05142012.txt" #Interpolation model names
|
27
|
infile5<-"mean_day244_rescaled.rst" #Raster or grid for the locations of predictions
|
28
|
|
29
|
#path<-"/home/parmentier/Data/IPLANT_project/data_Oregon_stations"
|
30
|
path<-"M:/Data/IPLANT_project/data_Oregon_stations" #Locations on Atlas
|
31
|
|
32
|
#Station location of the study area
|
33
|
stat_loc<-read.table(paste(path,"/","location_study_area_OR_0602012.txt",sep=""),sep=",", header=TRUE)
|
34
|
#GHCN Database for 1980-2010 for study area (OR)
|
35
|
data3<-read.table(paste(path,"/","ghcn_data_TMAXy1980_2010_OR_0602012.txt",sep=""),sep=",", header=TRUE)
|
36
|
|
37
|
prop<-0.3 #Proportion of testing retained for validation
|
38
|
#prop<-0.25
|
39
|
seed_number<- 100 #Seed number for random sampling
|
40
|
out_prefix<-"_07022012_10d_fusion12" #User defined output prefix
|
41
|
setwd(path)
|
42
|
############ START OF THE SCRIPT ##################
|
43
|
|
44
|
#
|
45
|
### Step 0/Step 6 in Brian's code...preparing year 2010 data for modeling
|
46
|
#
|
47
|
|
48
|
|
49
|
###Reading the station data and setting up for models' comparison
|
50
|
filename<-sub(".shp","",infile1) #Removing the extension from file.
|
51
|
ghcn<-readOGR(".", filename) #reading shapefile
|
52
|
|
53
|
CRS<-proj4string(ghcn) #Storing projection information (ellipsoid, datum,etc.)
|
54
|
|
55
|
mean_LST<- readGDAL(infile5) #Reading the whole raster in memory. This provides a grid for kriging
|
56
|
proj4string(mean_LST)<-CRS #Assigning coordinate information to prediction grid.
|
57
|
|
58
|
ghcn = transform(ghcn,Northness = cos(ASPECT*pi/180)) #Adding a variable to the dataframe
|
59
|
#ghcn = transform(ghcn,Northness = cos(ASPECT)) #Adding a variable to the dataframe
|
60
|
#ghcn = transform(ghcn,Eastness = sin(ASPECT)) #adding variable to the dataframe.
|
61
|
ghcn = transform(ghcn,Eastness = sin(ASPECT*pi/180)) #adding variable to the dataframe.
|
62
|
#ghcn = transform(ghcn,Northness_w = sin(slope)*cos(ASPECT)) #Adding a variable to the dataframe
|
63
|
#ghcn = transform(ghcn,Eastness_w = sin(slope)*sin(ASPECT)) #adding variable to the dataframe.
|
64
|
ghcn = transform(ghcn,Northness_w = sin(slope*pi/180)*cos(ASPECT*pi/180)) #Adding a variable to the dataframe
|
65
|
ghcn = transform(ghcn,Eastness_w = sin(slope*pi/180)*sin(ASPECT*pi/180)) #adding variable to the dataframe.
|
66
|
|
67
|
#Remove NA for LC and CANHEIGHT
|
68
|
ghcn$LC1[is.na(ghcn$LC1)]<-0
|
69
|
ghcn$LC3[is.na(ghcn$LC3)]<-0
|
70
|
ghcn$CANHEIGHT[is.na(ghcn$CANHEIGHT)]<-0
|
71
|
|
72
|
|
73
|
set.seed(seed_number) #Using a seed number allow results based on random number to be compared...
|
74
|
|
75
|
dates <-readLines(paste(path,"/",infile2, sep=""))
|
76
|
LST_dates <-readLines(paste(path,"/",infile3, sep=""))
|
77
|
models <-readLines(paste(path,"/",infile4, sep=""))
|
78
|
|
79
|
#Model assessment: specific diagnostic/metrics for GAM
|
80
|
results_AIC<- matrix(1,length(dates),length(models)+3)
|
81
|
results_GCV<- matrix(1,length(dates),length(models)+3)
|
82
|
results_DEV<- matrix(1,length(dates),length(models)+3)
|
83
|
results_RMSE_f<- matrix(1,length(dates),length(models)+3)
|
84
|
|
85
|
#Model assessment: general diagnostic/metrics
|
86
|
results_RMSE <- matrix(1,length(dates),length(models)+4)
|
87
|
results_MAE <- matrix(1,length(dates),length(models)+4)
|
88
|
results_ME <- matrix(1,length(dates),length(models)+4) #There are 8+1 models
|
89
|
results_R2 <- matrix(1,length(dates),length(models)+4) #Coef. of determination for the validation dataset
|
90
|
results_RMSE_f<- matrix(1,length(dates),length(models)+4) #RMSE fit, RMSE for the training dataset
|
91
|
results_RMSE_f_kr<- matrix(1,length(dates),length(models)+4)
|
92
|
|
93
|
# #Tracking relationship between LST AND LC
|
94
|
# cor_LST_LC1<-matrix(1,10,1) #correlation LST-LC1
|
95
|
# cor_LST_LC3<-matrix(1,10,1) #correlation LST-LC3
|
96
|
# cor_LST_tmax<-matrix(1,10,1) #correlation LST-tmax
|
97
|
|
98
|
#Screening for bad values: value is tmax in this case
|
99
|
#ghcn$value<-as.numeric(ghcn$value)
|
100
|
ghcn_all<-ghcn
|
101
|
ghcn_test<-subset(ghcn,ghcn$value>-150 & ghcn$value<400)
|
102
|
ghcn_test2<-subset(ghcn_test,ghcn_test$ELEV_SRTM>0)
|
103
|
ghcn<-ghcn_test2
|
104
|
#coords<- ghcn[,c('x_OR83M','y_OR83M')]
|
105
|
|
106
|
month_var<-c("mm_01","mm_02","mm_03","mm_04","mm_05","mm_06","mm_07","mm_08","mm_09", "mm_10", "mm_11", "mm_12")
|
107
|
ghcn.subsets <-lapply(dates, function(d) subset(ghcn, date==d)) #this creates a list of 10 or 365 subsets dataset based on dates
|
108
|
|
109
|
|
110
|
#Start loop here...
|
111
|
|
112
|
## looping through the dates...this is the main part of the code
|
113
|
#i=1 #for debugging
|
114
|
#j=1 #for debugging
|
115
|
for(i in 1:length(dates)){ # start of the for loop #1
|
116
|
|
117
|
date<-strptime(dates[i], "%Y%m%d") # interpolation date being processed
|
118
|
month<-strftime(date, "%m") # current month of the date being processed
|
119
|
LST_month<-paste("mm_",month,sep="") # name of LST month to be matched
|
120
|
|
121
|
###Regression part 1: Creating a validation dataset by creating training and testing datasets
|
122
|
|
123
|
mod_LST <-ghcn.subsets[[i]][,match(LST_month, names(ghcn.subsets[[i]]))] #Match interpolation date and monthly LST average
|
124
|
ghcn.subsets[[i]] = transform(ghcn.subsets[[i]],LST = mod_LST) #Add the variable LST to the subset dataset
|
125
|
n<-nrow(ghcn.subsets[[i]])
|
126
|
ns<-n-round(n*prop) #Create a sample from the data frame with 70% of the rows
|
127
|
nv<-n-ns #create a sample for validation with prop of the rows
|
128
|
ind.training <- sample(nrow(ghcn.subsets[[i]]), size=ns, replace=FALSE) #This selects the index position for 70% of the rows taken randomly
|
129
|
ind.testing <- setdiff(1:nrow(ghcn.subsets[[i]]), ind.training)
|
130
|
data_s <- ghcn.subsets[[i]][ind.training, ] #Training dataset currently used in the modeling
|
131
|
data_v <- ghcn.subsets[[i]][ind.testing, ] #Testing/validation dataset
|
132
|
|
133
|
#i=1
|
134
|
date_proc<-dates[i]
|
135
|
date_proc<-strptime(dates[i], "%Y%m%d") # interpolation date being processed
|
136
|
mo<-as.integer(strftime(date_proc, "%m")) # current month of the date being processed
|
137
|
day<-as.integer(strftime(date_proc, "%d"))
|
138
|
year<-as.integer(strftime(date_proc, "%Y"))
|
139
|
|
140
|
#setup
|
141
|
|
142
|
#mo=9 #Commented out by Benoit on June 14
|
143
|
#day=2
|
144
|
#year=2010
|
145
|
datelabel=format(ISOdate(year,mo,day),"%b %d, %Y")
|
146
|
|
147
|
###########
|
148
|
# STEP 1 - 10 year monthly averages
|
149
|
###########
|
150
|
|
151
|
#library(raster)
|
152
|
#old<-getwd()
|
153
|
#setwd("c:/data/benoit/data_Oregon_stations_Brian_04242012")
|
154
|
#l=list.files(pattern="mean_month.*rescaled.tif")
|
155
|
l=list.files(pattern="mean_month.*rescaled.rst")
|
156
|
molst<-stack(l) #Creating a raster stack...
|
157
|
#setwd(old)
|
158
|
molst=molst-273.16 #K->C
|
159
|
idx <- seq(as.Date('2010-01-15'), as.Date('2010-12-15'), 'month')
|
160
|
molst <- setZ(molst, idx)
|
161
|
layerNames(molst) <- month.abb
|
162
|
themolst<-raster(molst,mo) #current month being processed saved in a raster image
|
163
|
plot(themolst)
|
164
|
|
165
|
###########
|
166
|
# STEP 2 - Weather station means across same days: Monthly mean calculation
|
167
|
###########
|
168
|
|
169
|
# ??? which years & what quality flags???
|
170
|
#select ghcn.id, lat,lon, elev, month, avg(value/10) as "TMax", count(*) as "NumDays" from ghcn, stations where ghcn.id in (select id from stations where state=='OR') and ghcn.id==stations.id and value<>-9999 and year>=2000 and element=='TMAX' group by stations.id, month;select ghcn.id, lat,lon, elev, month, avg(value/10) as "TMax", count(*) as "NumDays" from ghcn, stations where ghcn.id in (select id from stations where state=='OR') and ghcn.id==stations.id and value<>-9999 and year>=2000 and element=='TMAX' group by stations.id, month;
|
171
|
#below table from above SQL query
|
172
|
#dst<-read.csv('/data/benoit/data_oregon_stations_brian_04242012/station_means.csv',h=T)
|
173
|
|
174
|
##Added by Benoit ######
|
175
|
date1<-ISOdate(data3$year,data3$month,data3$day) #Creating a date object from 3 separate column
|
176
|
date2<-as.POSIXlt(as.Date(date1))
|
177
|
data3$date<-date2
|
178
|
d<-subset(data3,year>=2000 & mflag=="0" ) #Selecting dataset 2000-2010 with good quality
|
179
|
#May need some screeing??? i.e. range of temp and elevation...
|
180
|
d1<-aggregate(value~station+month, data=d, mean) #Calculate monthly mean for every station in OR
|
181
|
id<-as.data.frame(unique(d1$station)) #Unique station in OR for year 2000-2010
|
182
|
|
183
|
dst<-merge(d1, stat_loc, by.x="station", by.y="STAT_ID") #Inner join all columns are retained
|
184
|
|
185
|
#This allows to change only one name of the data.frame
|
186
|
pos<-match("value",names(dst)) #Find column with name "value"
|
187
|
names(dst)[pos]<-c("TMax")
|
188
|
dst$TMax<-dst$TMax/10 #TMax is the average max temp for months
|
189
|
#dstjan=dst[dst$month==9,] #dst contains the monthly averages for tmax for every station over 2000-2010
|
190
|
##############
|
191
|
|
192
|
modst=dst[dst$month==mo,] #Subsetting dataset for the relevnat month of the date being processed
|
193
|
|
194
|
##########
|
195
|
# STEP 3 - get LST at stations
|
196
|
##########
|
197
|
|
198
|
sta_lola=modst[,c("lon","lat")] #Extracting locations of stations for the current month...
|
199
|
library(rgdal)
|
200
|
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";
|
201
|
lookup<-function(r,lat,lon) {
|
202
|
xy<-project(cbind(lon,lat),proj_str);
|
203
|
cidx<-cellFromXY(r,xy);
|
204
|
return(r[cidx])
|
205
|
}
|
206
|
sta_tmax_from_lst=lookup(themolst,sta_lola$lat,sta_lola$lon) #Extracted values of LST for the stations
|
207
|
|
208
|
#########
|
209
|
# STEP 4 - bias at stations
|
210
|
#########
|
211
|
|
212
|
sta_bias=sta_tmax_from_lst-modst$TMax; #That is the difference between the monthly LST mean and monthly station mean
|
213
|
#Added by Benoit
|
214
|
modst$LSTD_bias<-sta_bias #Adding bias to data frame modst containning the monthly average for 10 years
|
215
|
|
216
|
bias_xy=project(as.matrix(sta_lola),proj_str)
|
217
|
# windows()
|
218
|
X11()
|
219
|
plot(modst$TMax,sta_tmax_from_lst,xlab="Station mo Tmax",ylab="LST mo Tmax",main=paste("LST vs TMax for",datelabel,sep=" "))
|
220
|
abline(0,1)
|
221
|
savePlot(paste("LST_TMax_scatterplot_",dates[i],out_prefix,".png", sep=""), type="png")
|
222
|
dev.off()
|
223
|
|
224
|
########
|
225
|
# STEP 5 - interpolate bias
|
226
|
########
|
227
|
|
228
|
# ?? include covariates like elev, distance to coast, cloud frequency, tree height
|
229
|
#library(fields)
|
230
|
#windows()
|
231
|
quilt.plot(sta_lola,sta_bias,main="Bias at stations",asp=1)
|
232
|
US(add=T,col="magenta",lwd=2)
|
233
|
#fitbias<-Tps(bias_xy,sta_bias) #use TPS or krige
|
234
|
fitbias<-Krig(bias_xy,sta_bias,theta=1e5) #use TPS or krige
|
235
|
#The output is a krig object using fields
|
236
|
# Creating plot of bias surface and saving it
|
237
|
X11()
|
238
|
datelabel2=format(ISOdate(year,mo,day),"%B ") #added by Benoit, label
|
239
|
surface(fitbias,col=rev(terrain.colors(100)),asp=1,main=paste("Interpolated bias for",datelabel2,sep=" "))
|
240
|
savePlot(paste("Bias_surface_LST_TMax_",dates[i],out_prefix,".png", sep=""), type="png")
|
241
|
dev.off()
|
242
|
|
243
|
#US(add=T,col="magenta",lwd=2)
|
244
|
|
245
|
##########
|
246
|
# STEP 6 - return to daily station data & calcualate delta=daily T-monthly T from stations
|
247
|
##########
|
248
|
|
249
|
#Commmented out by Benoit 06/14
|
250
|
# library(RSQLite)
|
251
|
# m<-dbDriver("SQLite")
|
252
|
# con<-dbConnect(m,dbname='c:/data/ghcn_tmintmax.db')
|
253
|
# querystr=paste("select ghcn.id, value as 'dailyTmax' from ghcn where ghcn.id in (select id from stations where state=='OR') and value<>-9999",
|
254
|
# "and year==",year,"and month==",mo,"and day==",day,
|
255
|
# "and element=='TMAX' ")
|
256
|
# rs<-dbSendQuery(con,querystr)
|
257
|
# d<-fetch(rs,n=-1)
|
258
|
# dbClearResult(rs)
|
259
|
# dbDisconnect(con)
|
260
|
#
|
261
|
# d$dailyTmax=d$dailyTmax/10 #stored as 1/10 degree C to allow integer storage
|
262
|
# dmoday=merge(modst,d,by="id")
|
263
|
##########################Commented out by Benoit
|
264
|
|
265
|
#added by Benoit
|
266
|
#x<-ghcn.subsets[[i]] #Holds both training and testing for instance 161 rows for Jan 1
|
267
|
x<-data_v
|
268
|
d<-data_s
|
269
|
|
270
|
pos<-match("value",names(d)) #Find column with name "value"
|
271
|
names(d)[pos]<-c("dailyTmax")
|
272
|
names(x)[pos]<-c("dailyTmax")
|
273
|
d$dailyTmax=(as.numeric(d$dailyTmax))/10 #stored as 1/10 degree C to allow integer storage
|
274
|
x$dailyTmax=(as.numeric(x$dailyTmax))/10 #stored as 1/10 degree C to allow integer storage
|
275
|
pos<-match("station",names(d)) #Find column with name "value"
|
276
|
names(d)[pos]<-c("id")
|
277
|
names(x)[pos]<-c("id")
|
278
|
names(modst)[1]<-c("id") #modst contains the average tmax per month for every stations...
|
279
|
dmoday=merge(modst,d,by="id") #LOOSING DATA HERE!!! from 113 t0 103
|
280
|
xmoday=merge(modst,x,by="id") #LOOSING DATA HERE!!! from 48 t0 43
|
281
|
names(dmoday)[4]<-c("lat")
|
282
|
names(dmoday)[5]<-c("lon") #dmoday contains all the the information: BIAS, monn
|
283
|
names(xmoday)[4]<-c("lat")
|
284
|
names(xmoday)[5]<-c("lon") #dmoday contains all the the information: BIAS, monn
|
285
|
|
286
|
data_v<-xmoday
|
287
|
###
|
288
|
|
289
|
#dmoday contains the daily tmax values with TMax being the monthly station tmax mean
|
290
|
#dmoday contains the daily tmax values with TMax being the monthly station tmax mean
|
291
|
|
292
|
# windows()
|
293
|
X11()
|
294
|
plot(dailyTmax~TMax,data=dmoday,xlab="Mo Tmax",ylab=paste("Daily for",datelabel),main="across stations in OR")
|
295
|
savePlot(paste("Daily_tmax_monthly_TMax_scatterplot_",dates[i],out_prefix,".png", sep=""), type="png")
|
296
|
dev.off()
|
297
|
##########
|
298
|
# STEP 7 - interpolate delta across space
|
299
|
##########
|
300
|
|
301
|
daily_sta_lola=dmoday[,c("lon","lat")] #could be same as before but why assume merge does this - assume not
|
302
|
daily_sta_xy=project(as.matrix(daily_sta_lola),proj_str)
|
303
|
daily_delta=dmoday$dailyTmax-dmoday$TMax
|
304
|
#windows()
|
305
|
quilt.plot(daily_sta_lola,daily_delta,asp=1,main="Station delta for Jan 15")
|
306
|
US(add=T,col="magenta",lwd=2)
|
307
|
#fitdelta<-Tps(daily_sta_xy,daily_delta) #use TPS or krige
|
308
|
fitdelta<-Krig(daily_sta_xy,daily_delta,theta=1e5) #use TPS or krige
|
309
|
#Kriging using fields package
|
310
|
|
311
|
# Creating plot of bias surface and saving it
|
312
|
X11()
|
313
|
surface(fitdelta,col=rev(terrain.colors(100)),asp=1,main=paste("Interpolated delta for",datelabel,sep=" "))
|
314
|
savePlot(paste("Delta_surface_LST_TMax_",dates[i],out_prefix,".png", sep=""), type="png")
|
315
|
dev.off()
|
316
|
#US(add=T,col="magenta",lwd=2)
|
317
|
#
|
318
|
|
319
|
#### Added by Benoit on 06/19
|
320
|
data_s<-dmoday #put the
|
321
|
data_s$daily_delta<-daily_delta
|
322
|
|
323
|
|
324
|
#data_s$y_var<-daily_delta #y_var is the variable currently being modeled, may be better with BIAS!!
|
325
|
data_s$y_var<-data_s$LSTD_bias
|
326
|
#data_s$y_var<-(data_s$dailyTmax)*10
|
327
|
#Model and response variable can be changed without affecting the script
|
328
|
|
329
|
mod1<- gam(y_var~ s(lat) + s (lon) + s (ELEV_SRTM), data=data_s)
|
330
|
mod2<- gam(y_var~ s(lat,lon)+ s(ELEV_SRTM), data=data_s) #modified nesting....from 3 to 2
|
331
|
mod3<- gam(y_var~ s(lat) + s (lon) + s (ELEV_SRTM) + s (Northness)+ s (Eastness) + s(DISTOC), data=data_s)
|
332
|
mod4<- gam(y_var~ s(lat) + s (lon) + s(ELEV_SRTM) + s(Northness) + s (Eastness) + s(DISTOC) + s(LST), data=data_s)
|
333
|
mod5<- gam(y_var~ s(lat,lon) +s(ELEV_SRTM) + s(Northness,Eastness) + s(DISTOC) + s(LST), data=data_s)
|
334
|
mod6<- gam(y_var~ s(lat,lon) +s(ELEV_SRTM) + s(Northness,Eastness) + s(DISTOC) + s(LST)+s(LC1), data=data_s)
|
335
|
mod7<- gam(y_var~ s(lat,lon) +s(ELEV_SRTM) + s(Northness,Eastness) + s(DISTOC) + s(LST)+s(LC3), data=data_s)
|
336
|
mod8<- gam(y_var~ s(lat,lon) +s(ELEV_SRTM) + s(Northness,Eastness) + s(DISTOC) + s(LST) + s(LC1), data=data_s)
|
337
|
|
338
|
#Added
|
339
|
#tmax_predicted=themolst+daily_delta_rast-bias_rast #Final surface?? but daily_rst
|
340
|
|
341
|
#### Added by Benoit ends
|
342
|
|
343
|
#########
|
344
|
# STEP 8 - assemble final answer - T=LST+Bias(interpolated)+delta(interpolated)
|
345
|
#########
|
346
|
|
347
|
bias_rast=interpolate(themolst,fitbias) #interpolation using function from raster package
|
348
|
#themolst is raster layer, fitbias is "Krig" object from bias surface
|
349
|
plot(bias_rast,main="Raster bias") #This not displaying...
|
350
|
|
351
|
daily_delta_rast=interpolate(themolst,fitdelta) #Interpolation of the bias surface...
|
352
|
|
353
|
plot(daily_delta_rast,main="Raster Daily Delta")
|
354
|
|
355
|
tmax_predicted=themolst+daily_delta_rast-bias_rast #Final surface as a raster layer...
|
356
|
#tmax_predicted=themolst+daily_delta_rast+bias_rast #Added by Benoit, why is it -bias_rast
|
357
|
plot(tmax_predicted,main="Predicted daily")
|
358
|
|
359
|
########
|
360
|
# check: assessment of results: validation
|
361
|
########
|
362
|
RMSE<-function(x,y) {return(mean((x-y)^2)^0.5)}
|
363
|
|
364
|
#FIT ASSESSMENT
|
365
|
sta_pred_data_s=lookup(tmax_predicted,data_s$lat,data_s$lon)
|
366
|
rmse_fit=RMSE(sta_pred_data_s,data_s$dailyTmax)
|
367
|
|
368
|
sta_pred=lookup(tmax_predicted,data_v$lat,data_v$lon)
|
369
|
#sta_pred=lookup(tmax_predicted,daily_sta_lola$lat,daily_sta_lola$lon)
|
370
|
#rmse=RMSE(sta_pred,dmoday$dailyTmax)
|
371
|
#pos<-match("value",names(data_v)) #Find column with name "value"
|
372
|
#names(data_v)[pos]<-c("dailyTmax")
|
373
|
tmax<-data_v$dailyTmax
|
374
|
#data_v$dailyTmax<-tmax
|
375
|
rmse=RMSE(sta_pred,tmax)
|
376
|
#plot(sta_pred~dmoday$dailyTmax,xlab=paste("Actual daily for",datelabel),ylab="Pred daily",main=paste("RMSE=",rmse))
|
377
|
X11()
|
378
|
plot(sta_pred~tmax,xlab=paste("Actual daily for",datelabel),ylab="Pred daily",main=paste("RMSE=",rmse))
|
379
|
abline(0,1)
|
380
|
savePlot(paste("Predicted_tmax_versus_observed_scatterplot_",dates[i],out_prefix,".png", sep=""), type="png")
|
381
|
dev.off()
|
382
|
#resid=sta_pred-dmoday$dailyTmax
|
383
|
resid=sta_pred-tmax
|
384
|
quilt.plot(daily_sta_lola,resid)
|
385
|
|
386
|
### END OF BRIAN's code
|
387
|
|
388
|
### Added by benoit
|
389
|
#Store results using TPS
|
390
|
j=9
|
391
|
results_RMSE[i,1]<- dates[i] #storing the interpolation dates in the first column
|
392
|
results_RMSE[i,2]<- ns #number of stations used in the training stage
|
393
|
results_RMSE[i,3]<- "RMSE"
|
394
|
results_RMSE[i,j+3]<- rmse #Storing RMSE for the model j
|
395
|
|
396
|
results_RMSE_f[i,1]<- dates[i] #storing the interpolation dates in the first column
|
397
|
results_RMSE_f[i,2]<- ns #number of stations used in the training stage
|
398
|
results_RMSE_f[i,3]<- "RMSE"
|
399
|
results_RMSE_f[i,j+3]<- rmse_fit #Storing RMSE for the model j
|
400
|
|
401
|
ns<-nrow(data_s) #This is added to because some loss of data might have happened because of the averaging...
|
402
|
nv<-nrow(data_v)
|
403
|
|
404
|
for (j in 1:length(models)){
|
405
|
|
406
|
##Model assessment: specific diagnostic/metrics for GAM
|
407
|
|
408
|
name<-paste("mod",j,sep="") #modj is the name of The "j" model (mod1 if j=1)
|
409
|
mod<-get(name) #accessing GAM model ojbect "j"
|
410
|
results_AIC[i,1]<- dates[i] #storing the interpolation dates in the first column
|
411
|
results_AIC[i,2]<- ns #number of stations used in the training stage
|
412
|
results_AIC[i,3]<- "AIC"
|
413
|
results_AIC[i,j+3]<- AIC (mod)
|
414
|
|
415
|
results_GCV[i,1]<- dates[i] #storing the interpolation dates in the first column
|
416
|
results_GCV[i,2]<- ns #number of stations used in the training
|
417
|
results_GCV[i,3]<- "GCV"
|
418
|
results_GCV[i,j+3]<- mod$gcv.ubre
|
419
|
|
420
|
results_DEV[i,1]<- dates[i] #storing the interpolation dates in the first column
|
421
|
results_DEV[i,2]<- ns #number of stations used in the training stage
|
422
|
results_DEV[i,3]<- "DEV"
|
423
|
results_DEV[i,j+3]<- mod$deviance
|
424
|
|
425
|
sta_LST_s=lookup(themolst,data_s$lat,data_s$lon)
|
426
|
sta_delta_s=lookup(daily_delta_rast,data_s$lat,data_s$lon) #delta surface has been calculated before!!
|
427
|
sta_bias_s= mod$fit
|
428
|
#Need to extract values from the kriged delta surface...
|
429
|
#sta_delta= lookup(delta_surface,data_v$lat,data_v$lon)
|
430
|
#tmax_predicted=sta_LST+sta_bias-y_mod$fit
|
431
|
tmax_predicted_s= sta_LST_s-sta_bias_s+sta_delta_s
|
432
|
|
433
|
results_RMSE_f[i,1]<- dates[i] #storing the interpolation dates in the first column
|
434
|
results_RMSE_f[i,2]<- ns #number of stations used in the training stage
|
435
|
results_RMSE_f[i,3]<- "RSME"
|
436
|
results_RMSE_f[i,j+3]<- sqrt(sum((tmax_predicted_s-data_s$dailyTmax)^2)/ns)
|
437
|
|
438
|
##Model assessment: general diagnostic/metrics
|
439
|
##validation: using the testing data
|
440
|
|
441
|
#This was modified on 06192012
|
442
|
|
443
|
#data_v$y_var<-data_v$LSTD_bias
|
444
|
#data_v$y_var<-tmax
|
445
|
y_mod<- predict(mod, newdata=data_v, se.fit = TRUE) #Using the coeff to predict new values.
|
446
|
|
447
|
####ADDED ON JULY 5th
|
448
|
sta_LST_v=lookup(themolst,data_v$lat,data_v$lon)
|
449
|
sta_delta_v=lookup(daily_delta_rast,data_v$lat,data_v$lon) #delta surface has been calculated before!!
|
450
|
sta_bias_v= y_mod$fit
|
451
|
#Need to extract values from the kriged delta surface...
|
452
|
#sta_delta= lookup(delta_surface,data_v$lat,data_v$lon)
|
453
|
#tmax_predicted=sta_LST+sta_bias-y_mod$fit
|
454
|
tmax_predicted_v= sta_LST_v-sta_bias_v+sta_delta_v
|
455
|
|
456
|
#data_v$tmax<-(data_v$tmax)/10
|
457
|
res_mod<- data_v$dailyTmax - tmax_predicted_v #Residuals for the model for fusion
|
458
|
#res_mod<- data_v$y_var - y_mod$fit #Residuals for the model
|
459
|
|
460
|
RMSE_mod <- sqrt(sum(res_mod^2)/nv) #RMSE FOR REGRESSION STEP 1: GAM
|
461
|
MAE_mod<- sum(abs(res_mod))/nv #MAE, Mean abs. Error FOR REGRESSION STEP 1: GAM
|
462
|
ME_mod<- sum(res_mod)/nv #ME, Mean Error or bias FOR REGRESSION STEP 1: GAM
|
463
|
R2_mod<- cor(data_v$dailyTmax,tmax_predicted_v)^2 #R2, coef. of var FOR REGRESSION STEP 1: GAM
|
464
|
|
465
|
results_RMSE[i,1]<- dates[i] #storing the interpolation dates in the first column
|
466
|
results_RMSE[i,2]<- ns #number of stations used in the training stage
|
467
|
results_RMSE[i,3]<- "RMSE"
|
468
|
results_RMSE[i,j+3]<- RMSE_mod #Storing RMSE for the model j
|
469
|
results_MAE[i,1]<- dates[i] #storing the interpolation dates in the first column
|
470
|
results_MAE[i,2]<- ns #number of stations used in the training stage
|
471
|
results_MAE[i,3]<- "MAE"
|
472
|
results_MAE[i,j+3]<- MAE_mod #Storing MAE for the model j
|
473
|
results_ME[i,1]<- dates[i] #storing the interpolation dates in the first column
|
474
|
results_ME[i,2]<- ns #number of stations used in the training stage
|
475
|
results_ME[i,3]<- "ME"
|
476
|
results_ME[i,j+3]<- ME_mod #Storing ME for the model j
|
477
|
results_R2[i,1]<- dates[i] #storing the interpolation dates in the first column
|
478
|
results_R2[i,2]<- ns #number of stations used in the training stage
|
479
|
results_R2[i,3]<- "R2"
|
480
|
results_R2[i,j+3]<- R2_mod #Storing R2 for the model j
|
481
|
|
482
|
#Saving residuals and prediction in the dataframes: tmax predicted from GAM
|
483
|
pred<-paste("pred_mod",j,sep="")
|
484
|
#data_v[[pred]]<-as.numeric(y_mod$fit)
|
485
|
data_v[[pred]]<-as.numeric(tmax_predicted_v)
|
486
|
data_s[[pred]]<-as.numeric(tmax_predicted_s) #Storing model fit values (predicted on training sample)
|
487
|
#data_s[[pred]]<-as.numeric(mod$fit) #Storing model fit values (predicted on training sample)
|
488
|
|
489
|
name2<-paste("res_mod",j,sep="")
|
490
|
data_v[[name2]]<-as.numeric(res_mod)
|
491
|
temp<-tmax_predicted_s-data_s$dailyTmax
|
492
|
data_s[[name2]]<-as.numeric(temp)
|
493
|
#end of loop calculating RMSE
|
494
|
|
495
|
}
|
496
|
|
497
|
# end of the for loop1
|
498
|
|
499
|
}
|
500
|
|
501
|
|
502
|
## Plotting and saving diagnostic measures
|
503
|
|
504
|
#Specific diagnostic measures related to the testing datasets
|
505
|
|
506
|
results_table_RMSE<-as.data.frame(results_RMSE)
|
507
|
results_table_MAE<-as.data.frame(results_MAE)
|
508
|
results_table_ME<-as.data.frame(results_ME)
|
509
|
results_table_R2<-as.data.frame(results_R2)
|
510
|
|
511
|
results_table_RMSE_f<-as.data.frame(results_RMSE_f)
|
512
|
|
513
|
cname<-c("dates","ns","metric","mod1", "mod2","mod3", "mod4", "mod5", "mod6", "mod7","mod8","mod9")
|
514
|
colnames(results_table_RMSE)<-cname
|
515
|
colnames(results_table_RMSE_f)<-cname
|
516
|
|
517
|
#tb_diagnostic1<-rbind(results_table_RMSE,results_table_MAE, results_table_ME, results_table_R2) #
|
518
|
tb_diagnostic1<-results_table_RMSE #measures of validation
|
519
|
tb_diagnostic2<-results_table_RMSE_f #measures of fit
|
520
|
|
521
|
write.table(tb_diagnostic1, file= paste(path,"/","results_fusion_Assessment_measure1",out_prefix,".txt",sep=""), sep=",")
|
522
|
write.table(tb_diagnostic2, file= paste(path,"/","results_fusion_Assessment_measure2",out_prefix,".txt",sep=""), sep=",")
|
523
|
|
524
|
#### END OF SCRIPT
|