1
|
######################################## METHOD COMPARISON #######################################
|
2
|
############################ Constant sampling for GAM CAI method #####################################
|
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
|
#Method is assedsed using constant sampling with variation of validation sample with different #
|
7
|
#hold out proportions. #
|
8
|
#AUTHOR: Benoit Parmentier #
|
9
|
#DATE: 12/27/2012 #
|
10
|
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#491-- #
|
11
|
###################################################################################################
|
12
|
|
13
|
###Loading R library and packages
|
14
|
library(gtools) # loading some useful tools
|
15
|
library(mgcv) # GAM package by Simon Wood
|
16
|
library(sp) # Spatial pacakge with class definition by Bivand et al.
|
17
|
library(spdep) # Spatial pacakge with methods and spatial stat. by Bivand et al.
|
18
|
library(rgdal) # GDAL wrapper for R, spatial utilities
|
19
|
library(gstat) # Kriging and co-kriging by Pebesma et al.
|
20
|
library(fields) # NCAR Spatial Interpolation methods such as kriging, splines
|
21
|
library(raster) # Hijmans et al. package for raster processing
|
22
|
library(rasterVis)
|
23
|
library(parallel) # Urbanek S. and Ripley B., package for multi cores & parralel processing
|
24
|
library(reshape)
|
25
|
|
26
|
### Parameters and argument
|
27
|
|
28
|
infile1<- "ghcn_or_tmax_covariates_06262012_OR83M.shp" #GHCN shapefile containing variables for modeling 2010
|
29
|
#infile2<-"list_10_dates_04212012.txt" #List of 10 dates for the regression
|
30
|
infile2<-"list_365_dates_04212012.txt"
|
31
|
infile3<-"LST_dates_var_names.txt" #LST dates name
|
32
|
infile4<-"models_interpolation_05142012.txt" #Interpolation model names
|
33
|
infile5<-"mean_day244_rescaled.rst" #Raster or grid for the locations of predictions
|
34
|
#infile6<-"lst_climatology.txt"
|
35
|
infile6<-"LST_files_monthly_climatology.txt"
|
36
|
inlistf<-"list_files_05032012.txt" #Stack of images containing the Covariates
|
37
|
|
38
|
path<-"/home/parmentier/Data/IPLANT_project/data_Oregon_stations_10242012_CAI" #Atlas location
|
39
|
setwd(path)
|
40
|
|
41
|
#Station location for the study area
|
42
|
stat_loc<-read.table(paste(path,"/","location_study_area_OR_0602012.txt",sep=""),sep=",", header=TRUE)
|
43
|
#GHCN Database for 1980-2010 for study area (OR)
|
44
|
data3<-read.table(paste(path,"/","ghcn_data_TMAXy1980_2010_OR_0602012.txt",sep=""),sep=",", header=TRUE)
|
45
|
|
46
|
nmodels<-9 #number of models running
|
47
|
y_var_name<-"dailyTmax" #climate variable interpolated
|
48
|
climgam<-1 #if 1, then GAM is run on the climatology rather than the daily deviation surface...
|
49
|
predval<-1 #if 1, produce raster prediction
|
50
|
prop<-0.3 #Proportion of testing retained for validation
|
51
|
|
52
|
seed_number<- 100 #Seed number for random sampling, if seed_number<0, no seed number is used..
|
53
|
#out_prefix<-"_365d_GAM_CAI2_const_10222012_" #User defined output prefix
|
54
|
#out_prefix<-"_365d_GAM_CAI2_const_all_lstd_10272012" #User defined output prefix
|
55
|
out_prefix<-"_365d_GAM_CAI4_all_12272012" #User defined output prefix
|
56
|
|
57
|
bias_val<-0 #if value 1 then daily training data is used in the bias surface rather than the all monthly stations (added on 07/11/2012)
|
58
|
bias_prediction<-1 #if value 1 then use GAM for the BIAS prediction otherwise GAM direct reprediction for y_var (daily tmax)
|
59
|
nb_sample<-1 #number of time random sampling must be repeated for every hold out proportion
|
60
|
prop_min<-0.3 #if prop_min=prop_max and step=0 then predicitons are done for the number of dates...
|
61
|
prop_max<-0.3
|
62
|
step<-0
|
63
|
constant<-0 #if value 1 then use the same sample used in the first date for interpolation over the set of dates
|
64
|
#projection used in the interpolation of the study area
|
65
|
CRS_interp<-"+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";
|
66
|
CRS_locs_WGS84<-CRS("+proj=longlat +ellps=WGS84 +datum=WGS84 +towgs84=0,0,0") #Station coords WGS84
|
67
|
|
68
|
#This can be entered as textfile or option later...ok for running now on 12/07/2012
|
69
|
list_formulas<-vector("list",nmodels)
|
70
|
|
71
|
list_formulas[[1]] <- as.formula("y_var~ s(ELEV_SRTM)", env=.GlobalEnv)
|
72
|
list_formulas[[2]] <- as.formula("y_var~ s(LST)", env=.GlobalEnv)
|
73
|
list_formulas[[3]] <- as.formula("y_var~ s(ELEV_SRTM,LST)", env=.GlobalEnv)
|
74
|
list_formulas[[4]] <- as.formula("y_var~ s(lat)+s(lon)+s(ELEV_SRTM)", env=.GlobalEnv)
|
75
|
list_formulas[[5]] <- as.formula("y_var~ s(lat,lon,ELEV_SRTM)", env=.GlobalEnv)
|
76
|
list_formulas[[6]] <- as.formula("y_var~ s(lat,lon)+s(ELEV_SRTM)+s(Northness_w,Eastness_w)+s(LST)", env=.GlobalEnv)
|
77
|
list_formulas[[7]] <- as.formula("y_var~ s(lat,lon)+s(ELEV_SRTM)+s(Northness_w,Eastness_w)+s(LST)+s(LC1)", env=.GlobalEnv)
|
78
|
list_formulas[[8]] <- as.formula("y_var~ s(lat,lon)+s(ELEV_SRTM)+s(Northness_w,Eastness_w)+s(LST)+s(LC3)", env=.GlobalEnv)
|
79
|
list_formulas[[9]] <- as.formula("y_var~ s(x_OR83M,y_OR83M)", env=.GlobalEnv)
|
80
|
|
81
|
#source("GAM_CAI_function_multisampling_10252012.R")
|
82
|
source("GAM_CAI_function_multisampling_12072012.R")
|
83
|
|
84
|
############ START OF THE SCRIPT ##################
|
85
|
|
86
|
###Reading the station data and setting up for models' comparison
|
87
|
filename<-sub(".shp","",infile1) #Removing the extension from file.
|
88
|
ghcn<-readOGR(".", filename) #reading shapefile
|
89
|
|
90
|
CRS<-proj4string(ghcn) #Storing projection information (ellipsoid, datum,etc.)
|
91
|
|
92
|
mean_LST<- readGDAL(infile5) #Reading the whole raster in memory. This provides a grid for kriging
|
93
|
proj4string(mean_LST)<-CRS #Assigning coordinate information to prediction grid.
|
94
|
|
95
|
ghcn <- transform(ghcn,Northness = cos(ASPECT*pi/180)) #Adding a variable to the dataframe
|
96
|
ghcn <- transform(ghcn,Eastness = sin(ASPECT*pi/180)) #adding variable to the dataframe.
|
97
|
ghcn <- transform(ghcn,Northness_w = sin(slope*pi/180)*cos(ASPECT*pi/180)) #Adding a variable to the dataframe
|
98
|
ghcn <- transform(ghcn,Eastness_w = sin(slope*pi/180)*sin(ASPECT*pi/180)) #adding variable to the dataframe.
|
99
|
|
100
|
#Remove NA for LC and CANHEIGHT
|
101
|
ghcn$LC1[is.na(ghcn$LC1)]<-0
|
102
|
ghcn$LC3[is.na(ghcn$LC3)]<-0
|
103
|
ghcn$CANHEIGHT[is.na(ghcn$CANHEIGHT)]<-0
|
104
|
ghcn$LC4[is.na(ghcn$LC4)]<-0
|
105
|
ghcn$LC6[is.na(ghcn$LC6)]<-0
|
106
|
|
107
|
#Use file.path for to construct pathfor independent os platform? !!!
|
108
|
dates<-readLines(file.path(path,infile2))
|
109
|
#dates <-readLines(paste(path,"/",infile2, sep=""))
|
110
|
LST_dates <-readLines(paste(path,"/",infile3, sep=""))
|
111
|
models <-readLines(paste(path,"/",infile4, sep=""))
|
112
|
|
113
|
##Extracting the variables values from the raster files
|
114
|
|
115
|
lines<-read.table(paste(path,"/",inlistf,sep=""), sep=" ") #Column 1 contains the names of raster files
|
116
|
inlistvar<-lines[,1]
|
117
|
inlistvar<-paste(path,"/",as.character(inlistvar),sep="")
|
118
|
covar_names<-as.character(lines[,2]) #Column two contains short names for covaraites
|
119
|
|
120
|
s_raster<- stack(inlistvar) #Creating a stack of raster images from the list of variables.
|
121
|
layerNames(s_raster)<-covar_names #Assigning names to the raster layers
|
122
|
projection(s_raster)<-CRS
|
123
|
|
124
|
#stat_val<- extract(s_raster, ghcn3) #Extracting values from the raster stack for every point location in coords data frame.
|
125
|
pos<-match("ASPECT",layerNames(s_raster)) #Find column with name "value"
|
126
|
r1<-raster(s_raster,layer=pos) #Select layer from stack
|
127
|
pos<-match("slope",layerNames(s_raster)) #Find column with name "value"
|
128
|
r2<-raster(s_raster,layer=pos) #Select layer from stack
|
129
|
N<-cos(r1*pi/180)
|
130
|
E<-sin(r1*pi/180)
|
131
|
Nw<-sin(r2*pi/180)*cos(r1*pi/180) #Adding a variable to the dataframe
|
132
|
Ew<-sin(r2*pi/180)*sin(r1*pi/180) #Adding variable to the dataframe.
|
133
|
|
134
|
pos<-match("LC1",layerNames(s_raster)) #Find column with name "value"
|
135
|
LC1<-raster(s_raster,layer=pos) #Select layer from stack
|
136
|
s_raster<-dropLayer(s_raster,pos)
|
137
|
LC1[is.na(LC1)]<-0 #NA must be set to zero.
|
138
|
pos<-match("LC3",layerNames(s_raster)) #Find column with name "value"
|
139
|
LC3<-raster(s_raster,layer=pos) #Select layer from stack
|
140
|
s_raster<-dropLayer(s_raster,pos)
|
141
|
LC3[is.na(LC3)]<-0
|
142
|
|
143
|
#Modification added to account for other land cover
|
144
|
|
145
|
pos<-match("LC4",layerNames(s_raster)) #Find column with name "value"
|
146
|
LC4<-raster(s_raster,layer=pos) #Select layer from stack
|
147
|
s_raster<-dropLayer(s_raster,pos)
|
148
|
LC4[is.na(LC4)]<-0
|
149
|
|
150
|
pos<-match("LC6",layerNames(s_raster)) #Find column with name "value"
|
151
|
LC6<-raster(s_raster,layer=pos) #Select layer from stack
|
152
|
s_raster<-dropLayer(s_raster,pos)
|
153
|
LC6[is.na(LC6)]<-0
|
154
|
|
155
|
LC_s<-stack(LC1,LC3,LC4,LC6)
|
156
|
layerNames(LC_s)<-c("LC1_forest","LC3_grass","LC4_crop","LC6_urban")
|
157
|
#plot(LC_s)
|
158
|
|
159
|
pos<-match("CANHEIGHT",layerNames(s_raster)) #Find column with name "value"
|
160
|
CANHEIGHT<-raster(s_raster,layer=pos) #Select layer from stack
|
161
|
s_raster<-dropLayer(s_raster,pos)
|
162
|
CANHEIGHT[is.na(CANHEIGHT)]<-0
|
163
|
pos<-match("ELEV_SRTM",layerNames(s_raster)) #Find column with name "ELEV_SRTM"
|
164
|
ELEV_SRTM<-raster(s_raster,layer=pos) #Select layer from stack on 10/30
|
165
|
s_raster<-dropLayer(s_raster,pos)
|
166
|
ELEV_SRTM[ELEV_SRTM <0]<-NA
|
167
|
|
168
|
xy<-coordinates(r1) #get x and y projected coordinates...
|
169
|
xy_latlon<-project(xy, CRS, inv=TRUE) # find lat long for projected coordinats (or pixels...)
|
170
|
lon<-raster(xy_latlon) #Transform a matrix into a raster object ncol=ncol(r1), nrow=nrow(r1))
|
171
|
ncol(lon)<-ncol(r1)
|
172
|
nrow(lon)<-nrow(r1)
|
173
|
extent(lon)<-extent(r1)
|
174
|
projection(lon)<-CRS #At this stage this is still an empty raster with 536 nrow and 745 ncell
|
175
|
lat<-lon
|
176
|
values(lon)<-xy_latlon[,1]
|
177
|
values(lat)<-xy_latlon[,2]
|
178
|
|
179
|
r<-stack(N,E,Nw,Ew,lon,lat,LC1,LC3,LC4,LC6, CANHEIGHT,ELEV_SRTM)
|
180
|
rnames<-c("Northness","Eastness","Northness_w","Eastness_w", "lon","lat","LC1","LC3","LC4","LC6","CANHEIGHT","ELEV_SRTM")
|
181
|
layerNames(r)<-rnames
|
182
|
s_raster<-addLayer(s_raster, r)
|
183
|
|
184
|
#s_sgdf<-as(s_raster,"SpatialGridDataFrame") #Conversion to spatial grid data frame
|
185
|
|
186
|
####### Preparing LST stack of climatology...
|
187
|
|
188
|
#l=list.files(pattern="mean_month.*rescaled.rst")
|
189
|
l <-readLines(paste(path,"/",infile6, sep=""))
|
190
|
molst<-stack(l) #Creating a raster stack...
|
191
|
#setwd(old)
|
192
|
molst<-molst-273.16 #K->C #LST stack of monthly average...
|
193
|
idx <- seq(as.Date('2010-01-15'), as.Date('2010-12-15'), 'month')
|
194
|
molst <- setZ(molst, idx)
|
195
|
layerNames(molst) <- month.abb
|
196
|
|
197
|
###### Preparing tables for model assessment: specific diagnostic/metrics
|
198
|
|
199
|
#Model assessment: specific diagnostics/metrics
|
200
|
results_m1<- matrix(1,1,nmodels+3)
|
201
|
results_m2<- matrix(1,1,nmodels+3)
|
202
|
results_m3<- matrix(1,1,nmodels+3)
|
203
|
#results_RMSE_f<- matrix(1,length(models)+3)
|
204
|
|
205
|
#Model assessment: general diagnostic/metrics
|
206
|
results_RMSE <- matrix(1,1,nmodels+4)
|
207
|
results_MAE <- matrix(1,1,nmodels+4)
|
208
|
results_ME <- matrix(1,1,nmodels+4) #There are 8+1 models
|
209
|
results_R2 <- matrix(1,1,nmodels+4) #Coef. of determination for the validation dataset
|
210
|
|
211
|
results_RMSE_f<- matrix(1,1,nmodels+4) #RMSE fit, RMSE for the training dataset
|
212
|
results_MAE_f <- matrix(1,1,nmodels+4)
|
213
|
results_R2_f<-matrix(1,1,nmodels+4)
|
214
|
######## Preparing monthly averages from the ProstGres database
|
215
|
|
216
|
# do this work outside of (before) this function
|
217
|
# to avoid making a copy of the data frame inside the function call
|
218
|
date1<-ISOdate(data3$year,data3$month,data3$day) #Creating a date object from 3 separate column
|
219
|
date2<-as.POSIXlt(as.Date(date1))
|
220
|
data3$date<-date2
|
221
|
d<-subset(data3,year>=2000 & mflag=="0" ) #Selecting dataset 2000-2010 with good quality: 193 stations
|
222
|
#May need some screeing??? i.e. range of temp and elevation...
|
223
|
d1<-aggregate(value~station+month, data=d, mean) #Calculate monthly mean for every station in OR
|
224
|
id<-as.data.frame(unique(d1$station)) #Unique station in OR for year 2000-2010: 193 but 7 loss of monthly avg
|
225
|
|
226
|
dst<-merge(d1, stat_loc, by.x="station", by.y="STAT_ID") #Inner join all columns are retained
|
227
|
|
228
|
#This allows to change only one name of the data.frame
|
229
|
pos<-match("value",names(dst)) #Find column with name "value"
|
230
|
names(dst)[pos]<-c("TMax")
|
231
|
dst$TMax<-dst$TMax/10 #TMax is the average max temp for monthy data
|
232
|
#dstjan=dst[dst$month==9,] #dst contains the monthly averages for tmax for every station over 2000-2010
|
233
|
|
234
|
#Extracting covariates from stack for the monthly dataset...
|
235
|
coords<- dst[c('lon','lat')] #Define coordinates in a data frame
|
236
|
coordinates(dst)<-coords #Assign coordinates to the data frame
|
237
|
proj4string(dst)<-CRS_locs_WGS84 #Assign coordinates reference system in PROJ4 format
|
238
|
dst_month<-spTransform(dst,CRS(CRS_interp)) #Project from WGS84 to new coord. system
|
239
|
|
240
|
stations_val<-extract(s_raster,dst_month) #extraction of the infomration at station location
|
241
|
stations_val<-as.data.frame(stations_val)
|
242
|
dst_extract<-cbind(dst_month,stations_val)
|
243
|
dst<-dst_extract
|
244
|
#Now clean and screen monthly values
|
245
|
dst_all<-dst
|
246
|
dst<-subset(dst,dst$TMax>-15 & dst$TMax<40)
|
247
|
dst<-subset(dst,dst$ELEV_SRTM>0) #This will drop two stations...or 24 rows
|
248
|
|
249
|
######### Preparing daily values for training and testing
|
250
|
|
251
|
#Screening for bad values: value is tmax in this case
|
252
|
#ghcn$value<-as.numeric(ghcn$value)
|
253
|
ghcn_all<-ghcn
|
254
|
ghcn_test<-subset(ghcn,ghcn$value>-150 & ghcn$value<400)
|
255
|
ghcn_test2<-subset(ghcn_test,ghcn_test$ELEV_SRTM>0)
|
256
|
ghcn<-ghcn_test2
|
257
|
#coords<- ghcn[,c('x_OR83M','y_OR83M')]
|
258
|
|
259
|
##Sampling: training and testing sites...
|
260
|
|
261
|
if (seed_number>0) {
|
262
|
set.seed(seed_number) #Using a seed number allow results based on random number to be compared...
|
263
|
}
|
264
|
nel<-length(dates)
|
265
|
dates_list<-vector("list",nel) #list of one row data.frame
|
266
|
|
267
|
prop_range<-(seq(from=prop_min,to=prop_max,by=step))*100
|
268
|
sn<-length(dates)*nb_sample*length(prop_range)
|
269
|
|
270
|
for(i in 1:length(dates)){
|
271
|
d_tmp<-rep(dates[i],nb_sample*length(prop_range)) #repeating same date
|
272
|
s_nb<-rep(1:nb_sample,length(prop_range)) #number of random sample per proportion
|
273
|
prop_tmp<-sort(rep(prop_range, nb_sample))
|
274
|
tab_run_tmp<-cbind(d_tmp,s_nb,prop_tmp)
|
275
|
dates_list[[i]]<-tab_run_tmp
|
276
|
}
|
277
|
|
278
|
sampling_dat<-as.data.frame(do.call(rbind,dates_list))
|
279
|
names(sampling_dat)<-c("date","run_samp","prop")
|
280
|
|
281
|
for(i in 2:3){ # start of the for loop #1
|
282
|
sampling_dat[,i]<-as.numeric(as.character(sampling_dat[,i]))
|
283
|
}
|
284
|
|
285
|
sampling_dat$date<- as.character(sampling_dat[,1])
|
286
|
#ghcn.subsets <-lapply(dates, function(d) subset(ghcn, date==d)) #this creates a list of 10 or 365 subsets dataset based on dates
|
287
|
ghcn.subsets <-lapply(as.character(sampling_dat$date), function(d) subset(ghcn, date==d)) #this creates a list of 10 or 365 subsets dataset based on dates
|
288
|
|
289
|
if (seed_number>0) {
|
290
|
set.seed(seed_number) #Using a seed number allow results based on random number to be compared...
|
291
|
}
|
292
|
|
293
|
sampling<-vector("list",length(ghcn.subsets))
|
294
|
sampling_station_id<-vector("list",length(ghcn.subsets))
|
295
|
for(i in 1:length(ghcn.subsets)){
|
296
|
n<-nrow(ghcn.subsets[[i]])
|
297
|
prop<-(sampling_dat$prop[i])/100
|
298
|
ns<-n-round(n*prop) #Create a sample from the data frame with 70% of the rows
|
299
|
nv<-n-ns #create a sample for validation with prop of the rows
|
300
|
ind.training <- sample(nrow(ghcn.subsets[[i]]), size=ns, replace=FALSE) #This selects the index position for 70% of the rows taken randomly
|
301
|
ind.testing <- setdiff(1:nrow(ghcn.subsets[[i]]), ind.training)
|
302
|
#Find the corresponding
|
303
|
data_sampled<-ghcn.subsets[[i]][ind.training,] #selected the randomly sampled stations
|
304
|
station_id.training<-data_sampled$station #selected id for the randomly sampled stations (115)
|
305
|
#Save the information
|
306
|
sampling[[i]]<-ind.training
|
307
|
sampling_station_id[[i]]<- station_id.training
|
308
|
}
|
309
|
## Use same samples across the year...
|
310
|
if (constant==1){
|
311
|
sampled<-sampling[[1]]
|
312
|
data_sampled<-ghcn.subsets[[1]][sampled,] #selected the randomly sampled stations
|
313
|
station_sampled<-data_sampled$station #selected id for the randomly sampled stations (115)
|
314
|
list_const_sampling<-vector("list",sn)
|
315
|
list_const_sampling_station_id<-vector("list",sn)
|
316
|
for(i in 1:sn){
|
317
|
station_id.training<-intersect(station_sampled,ghcn.subsets[[i]]$station)
|
318
|
ind.training<-match(station_id.training,ghcn.subsets[[i]]$station)
|
319
|
list_const_sampling[[i]]<-ind.training
|
320
|
list_const_sampling_station_id[[i]]<-station_id.training
|
321
|
}
|
322
|
sampling<-list_const_sampling
|
323
|
sampling_station_id<-list_const_sampling_station_id
|
324
|
}
|
325
|
|
326
|
######## Prediction for the range of dates
|
327
|
|
328
|
#Start loop here...
|
329
|
|
330
|
#gam_CAI_mod<-mclapply(1:length(dates), runGAMCAI,mc.preschedule=FALSE,mc.cores = 8) #This is the end bracket from mclapply(...) statement
|
331
|
gam_CAI_mod<-mclapply(1:length(ghcn.subsets), runGAMCAI,mc.preschedule=FALSE,mc.cores = 9) #This is the end bracket from mclapply(...) statement
|
332
|
#gam_CAI_mod<-mclapply(1:2, runGAMCAI,mc.preschedule=FALSE,mc.cores = 2) #This is the end bracket from mclapply(...) statement#
|
333
|
#gam_CAI_mod<-mclapply(1:2, runGAMCAI,mc.preschedule=FALSE,mc.cores = 2) #This is the end bracket from mclapply(...) statement
|
334
|
|
335
|
tb<-gam_CAI_mod[[1]][[3]][0,] #empty data frame with metric table structure that can be used in rbinding...
|
336
|
tb_tmp<-gam_CAI_mod #copy
|
337
|
|
338
|
for (i in 1:length(tb_tmp)){
|
339
|
tmp<-tb_tmp[[i]][[3]]
|
340
|
tb<-rbind(tb,tmp)
|
341
|
}
|
342
|
rm(tb_tmp)
|
343
|
|
344
|
for(i in 4:(nmodels+4)){ # start of the for loop #1
|
345
|
tb[,i]<-as.numeric(as.character(tb[,i]))
|
346
|
}
|
347
|
|
348
|
metrics<-as.character(unique(tb$metric)) #Name of accuracy metrics (RMSE,MAE etc.)
|
349
|
tb_metric_list<-vector("list",length(metrics))
|
350
|
|
351
|
for(i in 1:length(metrics)){ # Reorganizing information in terms of metrics
|
352
|
metric_name<-paste("tb_",metrics[i],sep="")
|
353
|
tb_metric<-subset(tb, metric==metrics[i])
|
354
|
tb_metric<-cbind(tb_metric,sampling_dat[,2:3])
|
355
|
assign(metric_name,tb_metric)
|
356
|
tb_metric_list[[i]]<-tb_metric
|
357
|
}
|
358
|
mod_labels<-rep("mod",nmodels+1)
|
359
|
index<-as.character(1:(nmodels+1))
|
360
|
mod_labels<-paste(mod_labels,index,sep="")
|
361
|
|
362
|
tb_diagnostic<-do.call(rbind,tb_metric_list) #produce a data.frame from the list ...
|
363
|
tb_diagnostic[["prop"]]<-as.factor(tb_diagnostic[["prop"]])
|
364
|
|
365
|
t<-melt(tb_diagnostic,
|
366
|
measure=mod_labels,
|
367
|
id=c("dates","metric","prop"),
|
368
|
na.rm=F)
|
369
|
avg_tb<-cast(t,metric+prop~variable,mean)
|
370
|
median_tb<-cast(t,metric+prop~variable,mean)
|
371
|
avg_tb[["prop"]]<-as.numeric(as.character(avg_tb[["prop"]]))
|
372
|
avg_RMSE<-subset(avg_tb,metric=="RMSE")
|
373
|
|
374
|
# Save before plotting
|
375
|
#sampling_obj<-list(sampling_dat=sampling_dat,training=sampling)
|
376
|
#sampling_obj<-list(sampling_dat=sampling_dat,training=sampling, tb=tb_diagnostic)
|
377
|
sampling_obj<-list(sampling_dat=sampling_dat,training=sampling, training_id=sampling_station_id, tb=tb_diagnostic)
|
378
|
|
379
|
write.table(avg_tb, file= paste(path,"/","results2_fusion_Assessment_measure_avg_",out_prefix,".txt",sep=""), sep=",")
|
380
|
write.table(median_tb, file= paste(path,"/","results2_fusion_Assessment_measure_median_",out_prefix,".txt",sep=""), sep=",")
|
381
|
write.table(tb_diagnostic, file= paste(path,"/","results2_fusion_Assessment_measure",out_prefix,".txt",sep=""), sep=",")
|
382
|
write.table(tb, file= paste(path,"/","results2_fusion_Assessment_measure_all",out_prefix,".txt",sep=""), sep=",")
|
383
|
|
384
|
save(sampling_obj, file= paste(path,"/","results2_CAI_sampling_obj",out_prefix,".RData",sep=""))
|
385
|
save(gam_CAI_mod,file= paste(path,"/","results2_CAI_Assessment_measure_all",out_prefix,".RData",sep=""))
|
386
|
|
387
|
#new combined object used since november 2012
|
388
|
gam_CAI_mod_obj<-list(gam_CAI_mod=gam_CAI_mod,sampling_obj=sampling_obj)
|
389
|
save(gam_CAI_mod_obj,file= paste(path,"/","results_mod_obj_",out_prefix,".RData",sep=""))
|
390
|
|
391
|
#### END OF SCRIPT
|