1 |
230a3ae4
|
Benoit Parmentier
|
################## Functions for use in the raster prediction stage #######################################
|
2 |
|
|
############################ Interpolation in a given tile/region ##########################################
|
3 |
|
|
#This script contains 5 functions used in the interpolation of temperature in the specfied study/processing area:
|
4 |
|
|
# 1)predict_raster_model<-function(in_models,r_stack,out_filename)
|
5 |
|
|
# 2)fit_models<-function(list_formulas,data_training)
|
6 |
|
|
# 3)runClim_KGCAI<-function(j,list_param) : function that peforms GAM CAI method
|
7 |
|
|
# 4)runClim_KGFusion<-function(j,list_param) function for monthly step (climatology) in the fusion method
|
8 |
|
|
# 5)runGAMFusion <- function(i,list_param) : daily step for fusion method, perform daily prediction
|
9 |
|
|
#
|
10 |
|
|
#AUTHOR: Benoit Parmentier
|
11 |
40e4d58b
|
Benoit Parmentier
|
#DATE: 06/05/2013
|
12 |
230a3ae4
|
Benoit Parmentier
|
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#363--
|
13 |
|
|
|
14 |
|
|
##Comments and TODO:
|
15 |
|
|
#This script is meant to be for general processing tile by tile or region by region.
|
16 |
|
|
# Note that the functions are called from GAM_fusion_analysis_raster_prediction_mutlisampling.R.
|
17 |
|
|
# This will be expanded to other methods.
|
18 |
|
|
##################################################################################################
|
19 |
|
|
|
20 |
|
|
|
21 |
|
|
predict_raster_model<-function(in_models,r_stack,out_filename){
|
22 |
|
|
#This functions performs predictions on a raster grid given input models.
|
23 |
|
|
#Arguments: list of fitted models, raster stack of covariates
|
24 |
|
|
#Output: spatial grid data frame of the subset of tiles
|
25 |
|
|
list_rast_pred<-vector("list",length(in_models))
|
26 |
|
|
for (i in 1:length(in_models)){
|
27 |
|
|
mod <-in_models[[i]] #accessing GAM model ojbect "j"
|
28 |
|
|
raster_name<-out_filename[[i]]
|
29 |
|
|
if (inherits(mod,"gam")) { #change to c("gam","autoKrige")
|
30 |
|
|
raster_pred<- predict(object=r_stack,model=mod,na.rm=FALSE) #Using the coeff to predict new values.
|
31 |
|
|
names(raster_pred)<-"y_pred"
|
32 |
|
|
writeRaster(raster_pred, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...(IDRISI)
|
33 |
|
|
#print(paste("Interpolation:","mod", j ,sep=" "))
|
34 |
|
|
list_rast_pred[[i]]<-raster_name
|
35 |
|
|
}
|
36 |
|
|
}
|
37 |
|
|
if (inherits(mod,"try-error")) {
|
38 |
|
|
print(paste("no gam model fitted:",mod[1],sep=" ")) #change message for any model type...
|
39 |
|
|
}
|
40 |
|
|
return(list_rast_pred)
|
41 |
|
|
}
|
42 |
|
|
|
43 |
|
|
fit_models<-function(list_formulas,data_training){
|
44 |
|
|
#This functions several models and returns model objects.
|
45 |
|
|
#Arguments: - list of formulas for GAM models
|
46 |
|
|
# - fitting data in a data.frame or SpatialPointDataFrame
|
47 |
|
|
#Output: list of model objects
|
48 |
|
|
list_fitted_models<-vector("list",length(list_formulas))
|
49 |
|
|
for (k in 1:length(list_formulas)){
|
50 |
|
|
formula<-list_formulas[[k]]
|
51 |
|
|
mod<- try(gam(formula, data=data_training)) #change to any model!!
|
52 |
|
|
#mod<- try(autoKrige(formula, input_data=data_s,new_data=s_sgdf,data_variogram=data_s))
|
53 |
|
|
model_name<-paste("mod",k,sep="")
|
54 |
|
|
assign(model_name,mod)
|
55 |
|
|
list_fitted_models[[k]]<-mod
|
56 |
|
|
}
|
57 |
|
|
return(list_fitted_models)
|
58 |
|
|
}
|
59 |
|
|
|
60 |
40e4d58b
|
Benoit Parmentier
|
select_var_stack <-function(r_stack,formula_mod,spdf=TRUE){
|
61 |
|
|
##Write function to return only the relevant layers!!
|
62 |
|
|
#Note that default behaviour of the function is to remove na values in the subset
|
63 |
|
|
#of raster layers and return a spdf
|
64 |
|
|
|
65 |
|
|
### Start
|
66 |
|
|
|
67 |
|
|
covar_terms<-all.vars(formula_mod) #all covariates terms...+ y_var
|
68 |
|
|
if (length(covar_terms)==1){
|
69 |
|
|
r_stack_covar<-subset(r_stack,1)
|
70 |
|
|
} #use one layer
|
71 |
|
|
if (length(covar_terms)> 1){
|
72 |
|
|
r_stack_covar <-subset(r_stack,covar_terms[-1])
|
73 |
|
|
}
|
74 |
|
|
if (spdf==TRUE){
|
75 |
|
|
s_sgdf<-as(r_stack_covar,"SpatialGridDataFrame") #Conversion to spatial grid data frame, only convert the necessary layers!!
|
76 |
|
|
s_spdf<-as.data.frame(s_sgdf) #Note that this automatically removes all NA rows
|
77 |
|
|
s_spdf<-na.omit(s_spdf) #removes all rows that have na...
|
78 |
|
|
coords<- s_spdf[,c('s1','s2')]
|
79 |
|
|
coordinates(s_spdf)<-coords
|
80 |
|
|
proj4string(s_spdf)<-proj4string(s_sgdf) #Need to assign coordinates...
|
81 |
|
|
#raster_pred <- rasterize(s_spdf,r1,"pred",fun=mean)
|
82 |
|
|
covar_obj<-s_spdf
|
83 |
|
|
} else{
|
84 |
|
|
covar_obj<-r_stack_covar
|
85 |
|
|
}
|
86 |
|
|
|
87 |
|
|
return(covar_obj)
|
88 |
|
|
}
|
89 |
|
|
|
90 |
|
|
remove_na_spdf<-function(col_names,d_spdf){
|
91 |
|
|
#Purpose: remote na items from a subset of a SpatialPointsDataFrame
|
92 |
|
|
x<-d_spdf
|
93 |
|
|
coords <-coordinates(x)
|
94 |
|
|
x$s1<-coords[,1]
|
95 |
|
|
x$s2<-coords[,2]
|
96 |
|
|
|
97 |
|
|
x1<-x[c(col_names,"s1","s2")]
|
98 |
|
|
#x1$y_var <-data_training$y_var
|
99 |
|
|
#names(x1)
|
100 |
|
|
x1<-na.omit(as.data.frame(x1))
|
101 |
|
|
coordinates(x1)<-x1[c("s1","s2")]
|
102 |
|
|
proj4string(x1)<-proj4string(d_spdf)
|
103 |
|
|
return(x1)
|
104 |
|
|
}
|
105 |
|
|
|
106 |
|
|
predict_auto_krige_raster_model<-function(list_formulas,r_stack,data_training,out_filename){
|
107 |
230a3ae4
|
Benoit Parmentier
|
#This functions performs predictions on a raster grid given input models.
|
108 |
|
|
#Arguments: list of fitted models, raster stack of covariates
|
109 |
|
|
#Output: spatial grid data frame of the subset of tiles
|
110 |
|
|
|
111 |
|
|
list_fitted_models<-vector("list",length(list_formulas))
|
112 |
40e4d58b
|
Benoit Parmentier
|
list_rast_pred<-vector("list",length(list_formulas))
|
113 |
|
|
#s_sgdf<-as(r_stack,"SpatialGridDataFrame") #Conversion to spatial grid data frame, only convert the necessary layers!!
|
114 |
|
|
proj4string(data_training) <- projection(r_stack)
|
115 |
230a3ae4
|
Benoit Parmentier
|
for (k in 1:length(list_formulas)){
|
116 |
40e4d58b
|
Benoit Parmentier
|
formula_mod<-list_formulas[[k]]
|
117 |
|
|
raster_name<-out_filename[[k]]
|
118 |
|
|
#mod<- try(gam(formula, data=data_training)) #change to any model!!
|
119 |
|
|
s_spdf<-select_var_stack(r_stack,formula_mod,spdf=TRUE)
|
120 |
|
|
col_names<-all.vars(formula_mod)
|
121 |
|
|
if (length(col_names)==1){
|
122 |
|
|
data_fit <-data_training
|
123 |
|
|
}else{
|
124 |
|
|
data_fit <- remove_na_spdf(col_names,data_training)
|
125 |
|
|
}
|
126 |
|
|
|
127 |
|
|
mod <- try(autoKrige(formula_mod, input_data=data_fit,new_data=s_spdf,data_variogram=data_fit))
|
128 |
|
|
#mod <- try(autoKrige(formula_mod, input_data=data_training,new_data=s_spdf,data_variogram=data_training))
|
129 |
230a3ae4
|
Benoit Parmentier
|
model_name<-paste("mod",k,sep="")
|
130 |
|
|
assign(model_name,mod)
|
131 |
40e4d58b
|
Benoit Parmentier
|
|
132 |
|
|
if (inherits(mod,"autoKrige")) { #change to c("gam","autoKrige")
|
133 |
|
|
rpred<-mod$krige_output #Extracting the SptialGriDataFrame from the autokrige object
|
134 |
|
|
y_pred<-rpred$var1.pred #is the order the same?
|
135 |
|
|
raster_pred <- rasterize(rpred,r_stack,"var1.pred",fun=mean)
|
136 |
|
|
names(raster_pred)<-"y_pred"
|
137 |
|
|
writeRaster(raster_pred, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...
|
138 |
230a3ae4
|
Benoit Parmentier
|
#print(paste("Interpolation:","mod", j ,sep=" "))
|
139 |
40e4d58b
|
Benoit Parmentier
|
list_rast_pred[[k]]<-raster_name
|
140 |
|
|
mod$krige_output<-NULL
|
141 |
|
|
list_fitted_models[[k]]<-mod
|
142 |
|
|
|
143 |
|
|
}
|
144 |
|
|
if (inherits(mod,"try-error")) {
|
145 |
|
|
print(paste("no autokrige model fitted:",mod,sep=" ")) #change message for any model type...
|
146 |
|
|
list_fitted_models[[k]]<-mod
|
147 |
230a3ae4
|
Benoit Parmentier
|
}
|
148 |
|
|
}
|
149 |
40e4d58b
|
Benoit Parmentier
|
day_prediction_obj <-list(list_fitted_models,list_rast_pred)
|
150 |
|
|
names(day_prediction_obj) <-c("list_fitted_models","list_rast_pred")
|
151 |
|
|
return(day_prediction_obj)
|
152 |
230a3ae4
|
Benoit Parmentier
|
}
|
153 |
|
|
|
154 |
ab884b16
|
Benoit Parmentier
|
#Could merge both auto?
|
155 |
|
|
predict_autokrige_gwr_raster_model<-function(method_interp,list_formulas,r_stack,data_training,out_filename){
|
156 |
|
|
#This functions performs predictions on a raster grid given input models.
|
157 |
|
|
#It can be used at the daily or/and monthly time scale...
|
158 |
|
|
#Arguments: list of fitted models, raster stack of covariates
|
159 |
|
|
# method_interp must be equal to "gwr" or "kriging"
|
160 |
|
|
#Output: spatial grid data frame of the subset of tiles
|
161 |
|
|
|
162 |
|
|
list_fitted_models<-vector("list",length(list_formulas))
|
163 |
|
|
list_rast_pred<-vector("list",length(list_formulas))
|
164 |
|
|
#s_sgdf<-as(r_stack,"SpatialGridDataFrame") #Conversion to spatial grid data frame, only convert the necessary layers!!
|
165 |
|
|
proj4string(data_training) <- projection(r_stack)
|
166 |
|
|
for (k in 1:length(list_formulas)){
|
167 |
|
|
formula_mod<-list_formulas[[k]]
|
168 |
|
|
raster_name<-out_filename[[k]]
|
169 |
|
|
#mod<- try(gam(formula, data=data_training)) #change to any model!!
|
170 |
|
|
s_spdf<-select_var_stack(r_stack,formula_mod,spdf=TRUE)
|
171 |
|
|
col_names<-all.vars(formula_mod) #extract terms names from formula object
|
172 |
|
|
if (length(col_names)==1){
|
173 |
|
|
data_fit <-data_training
|
174 |
|
|
}else{
|
175 |
|
|
data_fit <- remove_na_spdf(col_names,data_training)
|
176 |
|
|
}
|
177 |
|
|
|
178 |
|
|
if(method_interp=="kriging"){
|
179 |
|
|
mod <- try(autoKrige(formula_mod, input_data=data_fit,new_data=s_spdf,data_variogram=data_fit))
|
180 |
|
|
}
|
181 |
|
|
|
182 |
|
|
if(method_interp=="gwr"){
|
183 |
|
|
bwGm <-try(gwr.sel(formula_mod,data=data_fit,gweight=gwr.Gauss, verbose = FALSE))
|
184 |
|
|
mod <- try(gwr(formula_mod, data=data_fit, bandwidth=bwGm, gweight=gwr.Gauss, hatmatrix=TRUE))
|
185 |
|
|
}
|
186 |
|
|
#mod <- try(autoKrige(formula_mod, input_data=data_training,new_data=s_spdf,data_variogram=data_training))
|
187 |
|
|
|
188 |
|
|
model_name<-paste("mod",k,sep="")
|
189 |
|
|
assign(model_name,mod)
|
190 |
|
|
|
191 |
|
|
if (inherits(mod,"autoKrige") | inherits(mod,"gwr")){ #change to c("gam","autoKrige")
|
192 |
|
|
if(method_interp=="kriging"){
|
193 |
|
|
rpred<-mod$krige_output #Extracting the SptialGriDataFrame from the autokrige object
|
194 |
|
|
y_pred<-rpred$var1.pred #is the order the same?
|
195 |
|
|
raster_pred <- rasterize(rpred,r_stack,"var1.pred",fun=mean)
|
196 |
|
|
mod$krige_output<-NULL
|
197 |
|
|
}
|
198 |
|
|
if(method_interp=="gwr"){
|
199 |
|
|
rpred <- gwr(formula_mod, data_fit, bandwidth=bwGm, fit.points =s_spdf,predict=TRUE, se.fit=TRUE,fittedGWRobject=mod)
|
200 |
|
|
#y_pred<-rpred$var1.pred #is the order the same?
|
201 |
|
|
raster_pred<-rasterize(rpred$SDF,r_stack,"pred",fun=mean)
|
202 |
|
|
}
|
203 |
|
|
|
204 |
|
|
names(raster_pred)<-"y_pred"
|
205 |
|
|
writeRaster(raster_pred, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...
|
206 |
|
|
#print(paste("Interpolation:","mod", j ,sep=" "))
|
207 |
|
|
list_rast_pred[[k]]<-raster_name
|
208 |
|
|
list_fitted_models[[k]]<-mod
|
209 |
|
|
|
210 |
|
|
}
|
211 |
|
|
if (inherits(mod,"try-error")) {
|
212 |
|
|
print(paste("no autokrige/gwr model fitted:",mod,sep=" ")) #change message for any model type...
|
213 |
|
|
list_fitted_models[[k]]<-mod
|
214 |
|
|
}
|
215 |
|
|
}
|
216 |
|
|
day_prediction_obj <-list(list_fitted_models,list_rast_pred)
|
217 |
|
|
names(day_prediction_obj) <-c("list_fitted_models","list_rast_pred")
|
218 |
|
|
return(day_prediction_obj)
|
219 |
|
|
}
|
220 |
|
|
|
221 |
230a3ae4
|
Benoit Parmentier
|
fit_models<-function(list_formulas,data_training){
|
222 |
|
|
#This functions several models and returns model objects.
|
223 |
|
|
#Arguments: - list of formulas for GAM models
|
224 |
|
|
# - fitting data in a data.frame or SpatialPointDataFrame
|
225 |
|
|
#Output: list of model objects
|
226 |
|
|
list_fitted_models<-vector("list",length(list_formulas))
|
227 |
|
|
for (k in 1:length(list_formulas)){
|
228 |
|
|
formula<-list_formulas[[k]]
|
229 |
|
|
mod<- try(gam(formula, data=data_training)) #change to any model!!
|
230 |
|
|
#mod<- try(autoKrige(formula, input_data=data_s,new_data=s_sgdf,data_variogram=data_s))
|
231 |
|
|
model_name<-paste("mod",k,sep="")
|
232 |
|
|
assign(model_name,mod)
|
233 |
|
|
list_fitted_models[[k]]<-mod
|
234 |
|
|
}
|
235 |
|
|
return(list_fitted_models)
|
236 |
|
|
}
|
237 |
|
|
|
238 |
|
|
####
|
239 |
ca16094e
|
Benoit Parmentier
|
#TODO:Should use interp_day_fun!!
|
240 |
230a3ae4
|
Benoit Parmentier
|
#Add log file and calculate time and sizes for processes-outputs
|
241 |
|
|
runGAM_day_fun <-function(i,list_param){
|
242 |
|
|
|
243 |
|
|
#Make this a function with multiple argument that can be used by mcmapply??
|
244 |
|
|
#Arguments:
|
245 |
|
|
#1)list_index: j
|
246 |
|
|
#2)covar_rast: covariates raster images used in the modeling
|
247 |
|
|
#3)covar_names: names of input variables
|
248 |
|
|
#4)lst_avg: list of LST climatogy names, may be removed later on
|
249 |
|
|
#5)list_models: list input models for bias calculation
|
250 |
|
|
#6)sampling_obj: data at the daily time scale
|
251 |
|
|
#7)var: TMAX or TMIN, variable being interpolated
|
252 |
|
|
#8)y_var_name: output name, not used at this stage
|
253 |
|
|
#9)out_prefix
|
254 |
|
|
#10) out_path
|
255 |
|
|
|
256 |
|
|
#The output is a list of four shapefile names produced by the function:
|
257 |
|
|
#1) clim: list of output names for raster climatogies
|
258 |
|
|
#2) data_month: monthly training data for bias surface modeling
|
259 |
|
|
#3) mod: list of model objects fitted
|
260 |
|
|
#4) formulas: list of formulas used in bias modeling
|
261 |
|
|
|
262 |
|
|
### PARSING INPUT ARGUMENTS
|
263 |
|
|
#list_param_runGAMFusion<-list(i,clim_yearlist,sampling_obj,var,y_var_name, out_prefix)
|
264 |
|
|
|
265 |
|
|
index<-list_param$list_index
|
266 |
|
|
s_raster<-list_param$covar_rast
|
267 |
|
|
covar_names<-list_param$covar_names
|
268 |
|
|
lst_avg<-list_param$lst_avg
|
269 |
|
|
list_models<-list_param$list_models
|
270 |
|
|
dst<-list_param$dst #monthly station dataset
|
271 |
|
|
sampling_obj<-list_param$sampling_obj
|
272 |
|
|
var<-list_param$var
|
273 |
|
|
y_var_name<-list_param$y_var_name
|
274 |
|
|
interpolation_method <-list_param$interpolation_method
|
275 |
|
|
out_prefix<-list_param$out_prefix
|
276 |
|
|
out_path<-list_param$out_path
|
277 |
ca16094e
|
Benoit Parmentier
|
screen_data_training<-list_param$screen_data_training
|
278 |
230a3ae4
|
Benoit Parmentier
|
|
279 |
|
|
ghcn.subsets<-sampling_obj$ghcn_data_day
|
280 |
|
|
sampling_dat <- sampling_obj$sampling_dat
|
281 |
|
|
sampling <- sampling_obj$sampling_index
|
282 |
|
|
|
283 |
|
|
##########
|
284 |
|
|
# STEP 1 - Read in information and get traing and testing stations
|
285 |
|
|
#############
|
286 |
|
|
|
287 |
|
|
date<-strptime(sampling_dat$date[i], "%Y%m%d") # interpolation date being processed
|
288 |
|
|
month<-strftime(date, "%m") # current month of the date being processed
|
289 |
|
|
LST_month<-paste("mm_",month,sep="") # name of LST month to be matched
|
290 |
|
|
proj_str<-proj4string(dst) #get the local projection information from monthly data
|
291 |
|
|
|
292 |
|
|
#Adding layer LST to the raster stack
|
293 |
|
|
#names(s_raster)<-covar_names
|
294 |
|
|
pos<-match("LST",names(s_raster)) #Find the position of the layer with name "LST", if not present pos=NA
|
295 |
|
|
s_raster<-dropLayer(s_raster,pos) # If it exists drop layer
|
296 |
|
|
LST<-subset(s_raster,LST_month)
|
297 |
|
|
names(LST)<-"LST"
|
298 |
|
|
s_raster<-addLayer(s_raster,LST) #Adding current month
|
299 |
|
|
|
300 |
|
|
###Regression part 1: Creating a validation dataset by creating training and testing datasets
|
301 |
|
|
data_day<-ghcn.subsets[[i]]
|
302 |
|
|
mod_LST <- ghcn.subsets[[i]][,match(LST_month, names(ghcn.subsets[[i]]))] #Match interpolation date and monthly LST average
|
303 |
|
|
data_day$LST <- as.data.frame(mod_LST)[,1] #Add the variable LST to the daily dataset
|
304 |
|
|
dst$LST<-dst[[LST_month]] #Add the variable LST to the monthly dataset
|
305 |
|
|
|
306 |
|
|
ind.training<-sampling[[i]]
|
307 |
|
|
ind.testing <- setdiff(1:nrow(data_day), ind.training)
|
308 |
|
|
data_s <- data_day[ind.training, ] #Training dataset currently used in the modeling
|
309 |
|
|
data_v <- data_day[ind.testing, ] #Testing/validation dataset using input sampling
|
310 |
|
|
|
311 |
|
|
ns<-nrow(data_s)
|
312 |
|
|
nv<-nrow(data_v)
|
313 |
|
|
#i=1
|
314 |
|
|
date_proc<-sampling_dat$date[i]
|
315 |
|
|
date_proc<-strptime(sampling_dat$date[i], "%Y%m%d") # interpolation date being processed
|
316 |
|
|
mo<-as.integer(strftime(date_proc, "%m")) # current month of the date being processed
|
317 |
|
|
day<-as.integer(strftime(date_proc, "%d"))
|
318 |
|
|
year<-as.integer(strftime(date_proc, "%Y"))
|
319 |
|
|
|
320 |
|
|
#### STEP 2: PREPARE DATA
|
321 |
|
|
|
322 |
|
|
#Clean out this part: make this a function call
|
323 |
|
|
x<-as.data.frame(data_v)
|
324 |
|
|
d<-as.data.frame(data_s)
|
325 |
|
|
for (j in 1:nrow(x)){
|
326 |
|
|
if (x$value[j]== -999.9){
|
327 |
|
|
x$value[j]<-NA
|
328 |
|
|
}
|
329 |
|
|
}
|
330 |
|
|
for (j in 1:nrow(d)){
|
331 |
|
|
if (d$value[j]== -999.9){
|
332 |
|
|
d$value[j]<-NA
|
333 |
|
|
}
|
334 |
|
|
}
|
335 |
|
|
pos<-match("value",names(d)) #Find column with name "value"
|
336 |
|
|
names(d)[pos]<-y_var_name
|
337 |
|
|
pos<-match("value",names(x)) #Find column with name "value"
|
338 |
|
|
names(x)[pos]<-y_var_name
|
339 |
|
|
pos<-match("station",names(d)) #Find column with station ID
|
340 |
|
|
names(d)[pos]<-c("id")
|
341 |
|
|
pos<-match("station",names(x)) #Find column with name station ID
|
342 |
|
|
names(x)[pos]<-c("id")
|
343 |
|
|
|
344 |
|
|
data_s<-d
|
345 |
|
|
data_v<-x
|
346 |
|
|
|
347 |
|
|
data_s$y_var <- data_s[[y_var_name]] #Adding the variable modeled
|
348 |
|
|
data_v$y_var <- data_v[[y_var_name]]
|
349 |
|
|
|
350 |
|
|
#Adding back spatal definition
|
351 |
|
|
|
352 |
|
|
coordinates(data_s)<-cbind(data_s$x,data_s$y)
|
353 |
|
|
proj4string(data_s)<-proj_str
|
354 |
|
|
coordinates(data_v)<-cbind(data_v$x,data_v$y)
|
355 |
|
|
proj4string(data_v)<-proj_str
|
356 |
|
|
#### STEP3: NOW FIT AND PREDICT MODEL
|
357 |
|
|
|
358 |
|
|
list_formulas<-lapply(list_models,as.formula,env=.GlobalEnv) #mulitple arguments passed to lapply!!
|
359 |
|
|
|
360 |
ca16094e
|
Benoit Parmentier
|
if(screen_data_training==TRUE){
|
361 |
|
|
col_names <-unlist(lapply(list_formulas,all.vars)) #extract all covariates names used in the models
|
362 |
|
|
col_names<-unique(col_names)
|
363 |
|
|
data_fit <- remove_na_spdf(col_names,data_s)
|
364 |
|
|
}else{
|
365 |
|
|
data_fit <- data_s
|
366 |
|
|
}
|
367 |
|
|
mod_list<-fit_models(list_formulas,data_fit) #only gam at this stage
|
368 |
|
|
#mod_list<-fit_models(list_formulas,data_s) #only gam at this stage
|
369 |
230a3ae4
|
Benoit Parmentier
|
cname<-paste("mod",1:length(mod_list),sep="") #change to more meaningful name?
|
370 |
|
|
names(mod_list)<-cname
|
371 |
|
|
|
372 |
|
|
#Now generate file names for the predictions...
|
373 |
|
|
list_out_filename<-vector("list",length(mod_list))
|
374 |
|
|
names(list_out_filename)<-cname
|
375 |
|
|
|
376 |
|
|
for (k in 1:length(list_out_filename)){
|
377 |
|
|
#i indicate which day is predicted, y_var_name indicates TMIN or TMAX
|
378 |
|
|
data_name<-paste(y_var_name,"_predicted_",names(mod_list)[k],"_",
|
379 |
|
|
sampling_dat$date[i],"_",sampling_dat$prop[i],
|
380 |
|
|
"_",sampling_dat$run_samp[i],sep="")
|
381 |
|
|
raster_name<-file.path(out_path,paste(interpolation_method,"_",data_name,out_prefix,".tif", sep=""))
|
382 |
|
|
list_out_filename[[k]]<-raster_name
|
383 |
|
|
}
|
384 |
|
|
|
385 |
|
|
#now predict values for raster image...
|
386 |
|
|
rast_day_list<-predict_raster_model(mod_list,s_raster,list_out_filename)
|
387 |
|
|
names(rast_day_list)<-cname
|
388 |
|
|
#Some models will not be predicted...remove them
|
389 |
|
|
rast_day_list<-rast_day_list[!sapply(rast_day_list,is.null)] #remove NULL elements in list
|
390 |
|
|
|
391 |
|
|
#Prepare object to return
|
392 |
|
|
|
393 |
|
|
day_obj<- list(rast_day_list,data_s,data_v,sampling_dat[i,],mod_list,list_models)
|
394 |
|
|
obj_names<-c(y_var_name,"data_s","data_v","sampling_dat","mod","formulas")
|
395 |
|
|
names(day_obj)<-obj_names
|
396 |
|
|
save(day_obj,file= file.path(out_path,paste("day_obj_",interpolation_method,"_",var,"_",sampling_dat$date[i],"_",sampling_dat$prop[i],
|
397 |
|
|
"_",sampling_dat$run_samp[i],out_prefix,".RData",sep="")))
|
398 |
|
|
return(day_obj)
|
399 |
|
|
|
400 |
|
|
}
|
401 |
|
|
|
402 |
40e4d58b
|
Benoit Parmentier
|
#Maybe should just use the same code...
|
403 |
230a3ae4
|
Benoit Parmentier
|
|
404 |
40e4d58b
|
Benoit Parmentier
|
runKriging_day_fun <-function(i,list_param){
|
405 |
230a3ae4
|
Benoit Parmentier
|
|
406 |
|
|
#Make this a function with multiple argument that can be used by mcmapply??
|
407 |
|
|
#Arguments:
|
408 |
|
|
#1)list_index: j
|
409 |
|
|
#2)covar_rast: covariates raster images used in the modeling
|
410 |
|
|
#3)covar_names: names of input variables
|
411 |
|
|
#4)lst_avg: list of LST climatogy names, may be removed later on
|
412 |
|
|
#5)list_models: list input models for bias calculation
|
413 |
40e4d58b
|
Benoit Parmentier
|
#6)sampling_obj: data at the daily time scale
|
414 |
230a3ae4
|
Benoit Parmentier
|
#7)var: TMAX or TMIN, variable being interpolated
|
415 |
|
|
#8)y_var_name: output name, not used at this stage
|
416 |
|
|
#9)out_prefix
|
417 |
40e4d58b
|
Benoit Parmentier
|
#10) out_path
|
418 |
|
|
|
419 |
230a3ae4
|
Benoit Parmentier
|
#The output is a list of four shapefile names produced by the function:
|
420 |
|
|
#1) clim: list of output names for raster climatogies
|
421 |
|
|
#2) data_month: monthly training data for bias surface modeling
|
422 |
|
|
#3) mod: list of model objects fitted
|
423 |
|
|
#4) formulas: list of formulas used in bias modeling
|
424 |
|
|
|
425 |
|
|
### PARSING INPUT ARGUMENTS
|
426 |
|
|
#list_param_runGAMFusion<-list(i,clim_yearlist,sampling_obj,var,y_var_name, out_prefix)
|
427 |
|
|
|
428 |
40e4d58b
|
Benoit Parmentier
|
index<-list_param$list_index
|
429 |
230a3ae4
|
Benoit Parmentier
|
s_raster<-list_param$covar_rast
|
430 |
|
|
covar_names<-list_param$covar_names
|
431 |
|
|
lst_avg<-list_param$lst_avg
|
432 |
|
|
list_models<-list_param$list_models
|
433 |
|
|
dst<-list_param$dst #monthly station dataset
|
434 |
40e4d58b
|
Benoit Parmentier
|
sampling_obj<-list_param$sampling_obj
|
435 |
230a3ae4
|
Benoit Parmentier
|
var<-list_param$var
|
436 |
|
|
y_var_name<-list_param$y_var_name
|
437 |
40e4d58b
|
Benoit Parmentier
|
interpolation_method <-list_param$interpolation_method
|
438 |
230a3ae4
|
Benoit Parmentier
|
out_prefix<-list_param$out_prefix
|
439 |
|
|
out_path<-list_param$out_path
|
440 |
|
|
|
441 |
|
|
|
442 |
|
|
ghcn.subsets<-sampling_obj$ghcn_data_day
|
443 |
|
|
sampling_dat <- sampling_obj$sampling_dat
|
444 |
|
|
sampling <- sampling_obj$sampling_index
|
445 |
|
|
|
446 |
|
|
##########
|
447 |
|
|
# STEP 1 - Read in information and get traing and testing stations
|
448 |
|
|
#############
|
449 |
|
|
|
450 |
|
|
date<-strptime(sampling_dat$date[i], "%Y%m%d") # interpolation date being processed
|
451 |
|
|
month<-strftime(date, "%m") # current month of the date being processed
|
452 |
|
|
LST_month<-paste("mm_",month,sep="") # name of LST month to be matched
|
453 |
|
|
proj_str<-proj4string(dst) #get the local projection information from monthly data
|
454 |
40e4d58b
|
Benoit Parmentier
|
|
455 |
|
|
#Adding layer LST to the raster stack
|
456 |
|
|
#names(s_raster)<-covar_names
|
457 |
|
|
pos<-match("LST",names(s_raster)) #Find the position of the layer with name "LST", if not present pos=NA
|
458 |
|
|
s_raster<-dropLayer(s_raster,pos) # If it exists drop layer
|
459 |
|
|
LST<-subset(s_raster,LST_month)
|
460 |
|
|
names(LST)<-"LST"
|
461 |
|
|
s_raster<-addLayer(s_raster,LST) #Adding current month
|
462 |
|
|
|
463 |
230a3ae4
|
Benoit Parmentier
|
###Regression part 1: Creating a validation dataset by creating training and testing datasets
|
464 |
|
|
data_day<-ghcn.subsets[[i]]
|
465 |
|
|
mod_LST <- ghcn.subsets[[i]][,match(LST_month, names(ghcn.subsets[[i]]))] #Match interpolation date and monthly LST average
|
466 |
40e4d58b
|
Benoit Parmentier
|
data_day$LST <- as.data.frame(mod_LST)[,1] #Add the variable LST to the daily dataset
|
467 |
230a3ae4
|
Benoit Parmentier
|
dst$LST<-dst[[LST_month]] #Add the variable LST to the monthly dataset
|
468 |
|
|
|
469 |
|
|
ind.training<-sampling[[i]]
|
470 |
|
|
ind.testing <- setdiff(1:nrow(data_day), ind.training)
|
471 |
|
|
data_s <- data_day[ind.training, ] #Training dataset currently used in the modeling
|
472 |
|
|
data_v <- data_day[ind.testing, ] #Testing/validation dataset using input sampling
|
473 |
|
|
|
474 |
|
|
ns<-nrow(data_s)
|
475 |
|
|
nv<-nrow(data_v)
|
476 |
|
|
#i=1
|
477 |
|
|
date_proc<-sampling_dat$date[i]
|
478 |
|
|
date_proc<-strptime(sampling_dat$date[i], "%Y%m%d") # interpolation date being processed
|
479 |
|
|
mo<-as.integer(strftime(date_proc, "%m")) # current month of the date being processed
|
480 |
|
|
day<-as.integer(strftime(date_proc, "%d"))
|
481 |
|
|
year<-as.integer(strftime(date_proc, "%Y"))
|
482 |
|
|
|
483 |
40e4d58b
|
Benoit Parmentier
|
#### STEP 2: PREPARE DATA
|
484 |
230a3ae4
|
Benoit Parmentier
|
|
485 |
40e4d58b
|
Benoit Parmentier
|
#Clean out this part: make this a function call
|
486 |
230a3ae4
|
Benoit Parmentier
|
x<-as.data.frame(data_v)
|
487 |
|
|
d<-as.data.frame(data_s)
|
488 |
|
|
for (j in 1:nrow(x)){
|
489 |
|
|
if (x$value[j]== -999.9){
|
490 |
|
|
x$value[j]<-NA
|
491 |
|
|
}
|
492 |
|
|
}
|
493 |
|
|
for (j in 1:nrow(d)){
|
494 |
|
|
if (d$value[j]== -999.9){
|
495 |
|
|
d$value[j]<-NA
|
496 |
|
|
}
|
497 |
|
|
}
|
498 |
|
|
pos<-match("value",names(d)) #Find column with name "value"
|
499 |
|
|
names(d)[pos]<-y_var_name
|
500 |
|
|
pos<-match("value",names(x)) #Find column with name "value"
|
501 |
|
|
names(x)[pos]<-y_var_name
|
502 |
|
|
pos<-match("station",names(d)) #Find column with station ID
|
503 |
|
|
names(d)[pos]<-c("id")
|
504 |
|
|
pos<-match("station",names(x)) #Find column with name station ID
|
505 |
|
|
names(x)[pos]<-c("id")
|
506 |
|
|
|
507 |
40e4d58b
|
Benoit Parmentier
|
data_s<-d
|
508 |
|
|
data_v<-x
|
509 |
|
|
|
510 |
|
|
data_s$y_var <- data_s[[y_var_name]] #Adding the variable modeled
|
511 |
|
|
data_v$y_var <- data_v[[y_var_name]]
|
512 |
230a3ae4
|
Benoit Parmentier
|
|
513 |
40e4d58b
|
Benoit Parmentier
|
#Adding back spatal definition
|
514 |
230a3ae4
|
Benoit Parmentier
|
|
515 |
40e4d58b
|
Benoit Parmentier
|
coordinates(data_s)<-cbind(data_s$x,data_s$y)
|
516 |
|
|
proj4string(data_s)<-proj_str
|
517 |
|
|
coordinates(data_v)<-cbind(data_v$x,data_v$y)
|
518 |
|
|
proj4string(data_v)<-proj_str
|
519 |
|
|
#### STEP3: NOW FIT AND PREDICT MODEL
|
520 |
230a3ae4
|
Benoit Parmentier
|
|
521 |
40e4d58b
|
Benoit Parmentier
|
list_formulas<-lapply(list_models,as.formula,env=.GlobalEnv) #mulitple arguments passed to lapply!!
|
522 |
|
|
#models names
|
523 |
|
|
cname<-paste("mod",1:length(list_formulas),sep="") #change to more meaningful name?
|
524 |
|
|
names(list_formulas) <- cname
|
525 |
|
|
#Now generate output file names for the predictions...
|
526 |
|
|
list_out_filename<-vector("list",length(list_formulas))
|
527 |
|
|
names(list_out_filename)<-cname
|
528 |
230a3ae4
|
Benoit Parmentier
|
|
529 |
40e4d58b
|
Benoit Parmentier
|
for (k in 1:length(list_out_filename)){
|
530 |
|
|
#i indicate which day is predicted, y_var_name indicates TMIN or TMAX
|
531 |
|
|
data_name<-paste(y_var_name,"_predicted_",names(list_formulas)[k],"_",
|
532 |
230a3ae4
|
Benoit Parmentier
|
sampling_dat$date[i],"_",sampling_dat$prop[i],
|
533 |
|
|
"_",sampling_dat$run_samp[i],sep="")
|
534 |
|
|
raster_name<-file.path(out_path,paste(interpolation_method,"_",data_name,out_prefix,".tif", sep=""))
|
535 |
40e4d58b
|
Benoit Parmentier
|
list_out_filename[[k]]<-raster_name
|
536 |
230a3ae4
|
Benoit Parmentier
|
}
|
537 |
|
|
|
538 |
40e4d58b
|
Benoit Parmentier
|
#now fit and predict values for raster image...
|
539 |
230a3ae4
|
Benoit Parmentier
|
|
540 |
40e4d58b
|
Benoit Parmentier
|
if (interpolation_method=="gam_daily"){
|
541 |
|
|
mod_list<-fit_models(list_formulas,data_s) #only gam at this stage
|
542 |
|
|
names(mod_list)<-cname
|
543 |
|
|
rast_day_list<-predict_raster_model(mod_list,s_raster,list_out_filename)
|
544 |
|
|
names(rast_day_list)<-cname
|
545 |
|
|
}
|
546 |
230a3ae4
|
Benoit Parmentier
|
|
547 |
40e4d58b
|
Benoit Parmentier
|
if (interpolation_method=="kriging_daily"){
|
548 |
|
|
day_prediction_obj<-predict_auto_krige_raster_model(list_formulas,s_raster,data_s,list_out_filename)
|
549 |
|
|
mod_list <-day_prediction_obj$list_fitted_models
|
550 |
|
|
rast_day_list <-day_prediction_obj$list_rast_pred
|
551 |
|
|
names(rast_day_list)<-cname
|
552 |
|
|
}
|
553 |
|
|
|
554 |
|
|
#Some models will not be predicted...remove them
|
555 |
|
|
rast_day_list<-rast_day_list[!sapply(rast_day_list,is.null)] #remove NULL elements in list
|
556 |
230a3ae4
|
Benoit Parmentier
|
|
557 |
40e4d58b
|
Benoit Parmentier
|
#Prepare object to return
|
558 |
|
|
|
559 |
|
|
day_obj<- list(rast_day_list,data_s,data_v,sampling_dat[i,],mod_list,list_models)
|
560 |
|
|
obj_names<-c(y_var_name,"data_s","data_v","sampling_dat","mod","formulas")
|
561 |
|
|
names(day_obj)<-obj_names
|
562 |
|
|
save(day_obj,file= file.path(out_path,paste("day_obj_",interpolation_method,"_",var,"_",sampling_dat$date[i],"_",sampling_dat$prop[i],
|
563 |
|
|
"_",sampling_dat$run_samp[i],out_prefix,".RData",sep="")))
|
564 |
|
|
return(day_obj)
|
565 |
230a3ae4
|
Benoit Parmentier
|
|
566 |
|
|
}
|
567 |
40e4d58b
|
Benoit Parmentier
|
|
568 |
ab884b16
|
Benoit Parmentier
|
run_interp_day_fun <-function(i,list_param){
|
569 |
|
|
|
570 |
|
|
#Make this a function with multiple argument that can be used by mcmapply??
|
571 |
|
|
#This function performs interpolation at daily time scale. Modifications made
|
572 |
|
|
#to run three possible methods: gwr, kriging and gam.
|
573 |
|
|
#Arguments:
|
574 |
|
|
#1)list_index: j
|
575 |
|
|
#2)covar_rast: covariates raster images used in the modeling
|
576 |
|
|
#3)covar_names: names of input variables
|
577 |
|
|
#4)lst_avg: list of LST climatogy names, may be removed later on
|
578 |
|
|
#5)list_models: list input models for bias calculation
|
579 |
|
|
#6)sampling_obj: data at the daily time scale
|
580 |
|
|
#7)var: TMAX or TMIN, variable being interpolated
|
581 |
|
|
#8)y_var_name: output name, not used at this stage
|
582 |
|
|
#9)out_prefix
|
583 |
|
|
#10) out_path
|
584 |
|
|
|
585 |
|
|
#The output is a list of four shapefile names produced by the function:
|
586 |
|
|
#1) clim: list of output names for raster climatologies
|
587 |
|
|
#2) data_month: monthly training data for bias surface modeling
|
588 |
|
|
#3) mod: list of model objects fitted
|
589 |
|
|
#4) formulas: list of formulas used in bias modeling
|
590 |
|
|
|
591 |
|
|
### PARSING INPUT ARGUMENTS
|
592 |
|
|
#list_param_runGAMFusion<-list(i,clim_yearlist,sampling_obj,var,y_var_name, out_prefix)
|
593 |
|
|
|
594 |
|
|
index<-list_param$list_index
|
595 |
|
|
s_raster<-list_param$covar_rast
|
596 |
|
|
covar_names<-list_param$covar_names
|
597 |
|
|
lst_avg<-list_param$lst_avg
|
598 |
|
|
list_models<-list_param$list_models
|
599 |
|
|
dst<-list_param$dst #monthly station dataset
|
600 |
|
|
sampling_obj<-list_param$sampling_obj
|
601 |
|
|
var<-list_param$var
|
602 |
|
|
y_var_name<-list_param$y_var_name
|
603 |
|
|
interpolation_method <-list_param$interpolation_method
|
604 |
|
|
out_prefix<-list_param$out_prefix
|
605 |
|
|
out_path<-list_param$out_path
|
606 |
|
|
|
607 |
|
|
|
608 |
|
|
ghcn.subsets<-sampling_obj$ghcn_data_day
|
609 |
|
|
sampling_dat <- sampling_obj$sampling_dat
|
610 |
|
|
sampling <- sampling_obj$sampling_index
|
611 |
|
|
|
612 |
|
|
##########
|
613 |
|
|
# STEP 1 - Read in information and get traing and testing stations
|
614 |
|
|
#############
|
615 |
|
|
|
616 |
|
|
date<-strptime(sampling_dat$date[i], "%Y%m%d") # interpolation date being processed
|
617 |
|
|
month<-strftime(date, "%m") # current month of the date being processed
|
618 |
|
|
LST_month<-paste("mm_",month,sep="") # name of LST month to be matched
|
619 |
|
|
proj_str<-proj4string(dst) #get the local projection information from monthly data
|
620 |
|
|
|
621 |
|
|
#Adding layer LST to the raster stack
|
622 |
|
|
#names(s_raster)<-covar_names
|
623 |
|
|
pos<-match("LST",names(s_raster)) #Find the position of the layer with name "LST", if not present pos=NA
|
624 |
|
|
s_raster<-dropLayer(s_raster,pos) # If it exists drop layer
|
625 |
|
|
LST<-subset(s_raster,LST_month)
|
626 |
|
|
names(LST)<-"LST"
|
627 |
|
|
s_raster<-addLayer(s_raster,LST) #Adding current month
|
628 |
|
|
|
629 |
|
|
###Regression part 1: Creating a validation dataset by creating training and testing datasets
|
630 |
|
|
data_day<-ghcn.subsets[[i]]
|
631 |
|
|
mod_LST <- ghcn.subsets[[i]][,match(LST_month, names(ghcn.subsets[[i]]))] #Match interpolation date and monthly LST average
|
632 |
|
|
data_day$LST <- as.data.frame(mod_LST)[,1] #Add the variable LST to the daily dataset
|
633 |
|
|
dst$LST<-dst[[LST_month]] #Add the variable LST to the monthly dataset
|
634 |
|
|
|
635 |
|
|
ind.training<-sampling[[i]]
|
636 |
|
|
ind.testing <- setdiff(1:nrow(data_day), ind.training)
|
637 |
|
|
data_s <- data_day[ind.training, ] #Training dataset currently used in the modeling
|
638 |
|
|
data_v <- data_day[ind.testing, ] #Testing/validation dataset using input sampling
|
639 |
|
|
|
640 |
|
|
ns<-nrow(data_s)
|
641 |
|
|
nv<-nrow(data_v)
|
642 |
|
|
#i=1
|
643 |
|
|
date_proc<-sampling_dat$date[i]
|
644 |
|
|
date_proc<-strptime(sampling_dat$date[i], "%Y%m%d") # interpolation date being processed
|
645 |
|
|
mo<-as.integer(strftime(date_proc, "%m")) # current month of the date being processed
|
646 |
|
|
day<-as.integer(strftime(date_proc, "%d"))
|
647 |
|
|
year<-as.integer(strftime(date_proc, "%Y"))
|
648 |
|
|
|
649 |
|
|
#### STEP 2: PREPARE DATA
|
650 |
|
|
|
651 |
|
|
#Clean out this part: make this a function call, should be done ine data preparation to retain the generality of the function
|
652 |
|
|
|
653 |
|
|
x<-as.data.frame(data_v)
|
654 |
|
|
d<-as.data.frame(data_s)
|
655 |
|
|
for (j in 1:nrow(x)){
|
656 |
|
|
if (x$value[j]== -999.9){
|
657 |
|
|
x$value[j]<-NA
|
658 |
|
|
}
|
659 |
|
|
}
|
660 |
|
|
for (j in 1:nrow(d)){
|
661 |
|
|
if (d$value[j]== -999.9){
|
662 |
|
|
d$value[j]<-NA
|
663 |
|
|
}
|
664 |
|
|
}
|
665 |
|
|
pos<-match("value",names(d)) #Find column with name "value"
|
666 |
|
|
names(d)[pos]<-y_var_name
|
667 |
|
|
pos<-match("value",names(x)) #Find column with name "value"
|
668 |
|
|
names(x)[pos]<-y_var_name
|
669 |
|
|
pos<-match("station",names(d)) #Find column with station ID
|
670 |
|
|
names(d)[pos]<-c("id")
|
671 |
|
|
pos<-match("station",names(x)) #Find column with name station ID
|
672 |
|
|
names(x)[pos]<-c("id")
|
673 |
|
|
|
674 |
|
|
data_s<-d
|
675 |
|
|
data_v<-x
|
676 |
|
|
|
677 |
|
|
data_s$y_var <- data_s[[y_var_name]] #Adding the variable modeled
|
678 |
|
|
data_v$y_var <- data_v[[y_var_name]]
|
679 |
|
|
|
680 |
|
|
#Adding back spatal definition
|
681 |
|
|
|
682 |
|
|
coordinates(data_s)<-cbind(data_s$x,data_s$y)
|
683 |
|
|
proj4string(data_s)<-proj_str
|
684 |
|
|
coordinates(data_v)<-cbind(data_v$x,data_v$y)
|
685 |
|
|
proj4string(data_v)<-proj_str
|
686 |
|
|
#### STEP3: NOW FIT AND PREDICT MODEL
|
687 |
|
|
|
688 |
|
|
list_formulas<-lapply(list_models,as.formula,env=.GlobalEnv) #mulitple arguments passed to lapply!!
|
689 |
|
|
#models names
|
690 |
|
|
cname<-paste("mod",1:length(list_formulas),sep="") #change to more meaningful name?
|
691 |
|
|
names(list_formulas) <- cname
|
692 |
|
|
#Now generate output file names for the predictions...
|
693 |
|
|
list_out_filename<-vector("list",length(list_formulas))
|
694 |
|
|
names(list_out_filename)<-cname
|
695 |
|
|
|
696 |
|
|
for (k in 1:length(list_out_filename)){
|
697 |
|
|
#i indicate which day is predicted, y_var_name indicates TMIN or TMAX
|
698 |
|
|
data_name<-paste(y_var_name,"_predicted_",names(list_formulas)[k],"_",
|
699 |
|
|
sampling_dat$date[i],"_",sampling_dat$prop[i],
|
700 |
|
|
"_",sampling_dat$run_samp[i],sep="")
|
701 |
|
|
raster_name<-file.path(out_path,paste(interpolation_method,"_",data_name,out_prefix,".tif", sep=""))
|
702 |
|
|
list_out_filename[[k]]<-raster_name
|
703 |
|
|
}
|
704 |
|
|
|
705 |
|
|
#now fit and predict values for raster image...
|
706 |
|
|
|
707 |
|
|
if (interpolation_method=="gam_daily"){
|
708 |
ca16094e
|
Benoit Parmentier
|
if(screen_data_training==TRUE){
|
709 |
|
|
col_names <-unlist(lapply(list_formulas,all.vars)) #extract all covariates names used in the models
|
710 |
|
|
col_names<-unique(col_names)
|
711 |
|
|
data_fit <- remove_na_spdf(col_names,data_s)
|
712 |
|
|
}else{
|
713 |
|
|
data_fit <- data_s
|
714 |
|
|
}
|
715 |
|
|
#mod_list<-fit_models(list_formulas,data_s) #only gam at this stage
|
716 |
|
|
mod_list<-fit_models(list_formulas,data_fit) #only gam at this stage
|
717 |
ab884b16
|
Benoit Parmentier
|
names(mod_list)<-cname
|
718 |
|
|
rast_day_list<-predict_raster_model(mod_list,s_raster,list_out_filename)
|
719 |
|
|
names(rast_day_list)<-cname
|
720 |
|
|
}
|
721 |
|
|
|
722 |
|
|
## need to change to use combined gwr autokrige function
|
723 |
|
|
if (interpolation_method=="kriging_daily"){
|
724 |
|
|
day_prediction_obj<-predict_auto_krige_raster_model(list_formulas,s_raster,data_s,list_out_filename)
|
725 |
|
|
mod_list <-day_prediction_obj$list_fitted_models
|
726 |
|
|
rast_day_list <-day_prediction_obj$list_rast_pred
|
727 |
|
|
names(rast_day_list)<-cname
|
728 |
|
|
}
|
729 |
|
|
|
730 |
|
|
if (interpolation_method=="gwr_daily"){
|
731 |
|
|
method_interp <- "gwr"
|
732 |
|
|
day_prediction_obj<-predict_autokrige_gwr_raster_model(method_interp,list_formulas,s_raster,data_s,list_out_filename)
|
733 |
|
|
mod_list <-day_prediction_obj$list_fitted_models
|
734 |
|
|
rast_day_list <-day_prediction_obj$list_rast_pred
|
735 |
|
|
names(rast_day_list)<-cname
|
736 |
|
|
}
|
737 |
|
|
#Some models will not be predicted...remove them
|
738 |
|
|
rast_day_list<-rast_day_list[!sapply(rast_day_list,is.null)] #remove NULL elements in list
|
739 |
|
|
|
740 |
|
|
#Prepare object to return
|
741 |
|
|
|
742 |
|
|
day_obj<- list(rast_day_list,data_s,data_v,sampling_dat[i,],mod_list,list_models)
|
743 |
|
|
obj_names<-c(y_var_name,"data_s","data_v","sampling_dat","mod","formulas")
|
744 |
|
|
names(day_obj)<-obj_names
|
745 |
|
|
save(day_obj,file= file.path(out_path,paste("day_obj_",interpolation_method,"_",var,"_",sampling_dat$date[i],"_",sampling_dat$prop[i],
|
746 |
|
|
"_",sampling_dat$run_samp[i],out_prefix,".RData",sep="")))
|
747 |
|
|
return(day_obj)
|
748 |
|
|
|
749 |
|
|
}
|