1 |
b8e2ba85
|
Benoit Parmentier
|
################## Interpolation of Tmax Using Kriging #######################################
|
2 |
|
|
########################### Kriging and Cokriging ###############################################
|
3 |
|
|
#This script interpolates station values for the Oregon case study using Kriging and Cokring. #
|
4 |
|
|
#The script uses LST monthly averages as input variables and loads the station data #
|
5 |
|
|
#from a shape file with projection information. #
|
6 |
|
|
#Note that this program: #
|
7 |
|
|
#1)assumes that the shape file is in the current working. #
|
8 |
|
|
#2)relevant variables were extracted from raster images before performing the regressions #
|
9 |
|
|
# and stored shapefile #
|
10 |
|
|
#This scripts predicts tmax using autokrige, gstat and LST derived from MOD11A1. #
|
11 |
|
|
#also included and assessed using the RMSE,MAE,ME and R2 from validation dataset. #
|
12 |
|
|
#TThe dates must be provided as a textfile. #
|
13 |
|
|
#AUTHOR: Benoit Parmentier #
|
14 |
|
|
#DATE: 07/15/2012 #
|
15 |
|
|
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#364-- #
|
16 |
|
|
##################################################################################################
|
17 |
|
|
|
18 |
|
|
###Loading R library and packages
|
19 |
|
|
#library(gtools) # loading some useful tools
|
20 |
|
|
library(mgcv) # GAM package by Wood 2006 (version 2012)
|
21 |
|
|
library(sp) # Spatial pacakge with class definition by Bivand et al. 2008
|
22 |
|
|
library(spdep) # Spatial pacakge with methods and spatial stat. by Bivand et al. 2012
|
23 |
|
|
library(rgdal) # GDAL wrapper for R, spatial utilities (Keitt et al. 2012)
|
24 |
|
|
library(gstat) # Kriging and co-kriging by Pebesma et al. 2004
|
25 |
|
|
library(automap) # Automated Kriging based on gstat module by Hiemstra et al. 2008
|
26 |
|
|
library(spgwr)
|
27 |
|
|
library(gpclib)
|
28 |
|
|
library(maptools)
|
29 |
|
|
library(graphics)
|
30 |
|
|
library(parallel) # Urbanek S. and Ripley B., package for multi cores & parralel processing
|
31 |
|
|
library(raster)
|
32 |
|
|
|
33 |
|
|
###Parameters and arguments
|
34 |
|
|
|
35 |
|
|
infile1<- "ghcn_or_tmax_covariates_06262012_OR83M.shp" #GHCN shapefile containing variables for modeling 2010
|
36 |
|
|
#infile2<-"list_10_dates_04212012.txt" #List of 10 dates for the regression
|
37 |
|
|
infile2<-"list_365_dates_04212012.txt"
|
38 |
|
|
infile3<-"LST_dates_var_names.txt" #LST dates name
|
39 |
|
|
infile4<-"models_interpolation_05142012.txt" #Interpolation model names
|
40 |
|
|
infile5<-"mean_day244_rescaled.rst"
|
41 |
|
|
inlistf<-"list_files_05032012.txt" #Stack of images containing the Covariates
|
42 |
|
|
|
43 |
|
|
path<-"/home/parmentier/Data/IPLANT_project/data_Oregon_stations_07152012" #Jupiter LOCATION on Atlas for kriging
|
44 |
|
|
#path<-"H:/Data/IPLANT_project/data_Oregon_stations" #Jupiter Location on XANDERS
|
45 |
|
|
|
46 |
|
|
setwd(path)
|
47 |
|
|
prop<-0.3 #Proportion of testing retained for validation
|
48 |
|
|
seed_number<- 100 #Seed number for random sampling
|
49 |
|
|
models<-7 #Number of kriging model
|
50 |
|
|
out_prefix<-"_07202012_auto_krig1" #User defined output prefix
|
51 |
|
|
|
52 |
|
|
source("krigingUK_function_07192012b.R")
|
53 |
|
|
|
54 |
|
|
###STEP 1 DATA PREPARATION AND PROCESSING#####
|
55 |
|
|
|
56 |
|
|
###Reading the station data and setting up for models' comparison
|
57 |
|
|
filename<-sub(".shp","",infile1) #Removing the extension from file.
|
58 |
|
|
ghcn<-readOGR(".", filename) #reading shapefile
|
59 |
|
|
|
60 |
|
|
CRS<-proj4string(ghcn) #Storing projection information (ellipsoid, datum,etc.)
|
61 |
|
|
|
62 |
|
|
mean_LST<- readGDAL(infile5) #Reading the whole raster in memory. This provides a grid for kriging
|
63 |
|
|
proj4string(mean_LST)<-CRS #Assigning coordinate information to prediction grid.
|
64 |
|
|
|
65 |
|
|
##Extracting the variables values from the raster files
|
66 |
|
|
|
67 |
|
|
lines<-read.table(paste(path,"/",inlistf,sep=""), sep=" ") #Column 1 contains the names of raster files
|
68 |
|
|
inlistvar<-lines[,1]
|
69 |
|
|
inlistvar<-paste(path,"/",as.character(inlistvar),sep="")
|
70 |
|
|
covar_names<-as.character(lines[,2]) #Column two contains short names for covaraites
|
71 |
|
|
|
72 |
|
|
s_raster<- stack(inlistvar) #Creating a stack of raster images from the list of variables.
|
73 |
|
|
layerNames(s_raster)<-covar_names #Assigning names to the raster layers
|
74 |
|
|
projection(s_raster)<-CRS
|
75 |
|
|
|
76 |
|
|
#stat_val<- extract(s_raster, ghcn3) #Extracting values from the raster stack for every point location in coords data frame.
|
77 |
|
|
pos<-match("ASPECT",layerNames(s_raster)) #Find column with name "value"
|
78 |
|
|
r1<-raster(s_raster,layer=pos) #Select layer from stack
|
79 |
|
|
pos<-match("slope",layerNames(s_raster)) #Find column with name "value"
|
80 |
|
|
r2<-raster(s_raster,layer=pos) #Select layer from stack
|
81 |
|
|
N<-cos(r1*pi/180)
|
82 |
|
|
E<-sin(r1*pi/180)
|
83 |
|
|
Nw<-sin(r2*pi/180)*cos(r1*pi/180) #Adding a variable to the dataframe
|
84 |
|
|
Ew<-sin(r2*pi/180)*sin(r1*pi/180) #Adding variable to the dataframe.
|
85 |
|
|
#r<-stack(N,E,Nw,Ew)
|
86 |
|
|
#rnames<-c("Northness","Eastness","Northness_w","Eastness_w")
|
87 |
|
|
#layerNames(r)<-rnames
|
88 |
|
|
#s_raster<-addLayer(s_raster, r)
|
89 |
|
|
#s_sgdf<-as(s_raster,"SpatialGridDataFrame") #Conversion to spatial grid data frame
|
90 |
|
|
xy<-coordinates(r1) #get x and y projected coordinates...
|
91 |
|
|
xy_latlon<-project(xy, CRS, inv=TRUE) # find lat long for projected coordinats (or pixels...)
|
92 |
|
|
tmp<-raster(xy_latlon) #, ncol=ncol(r1), nrow=nrow(r1))
|
93 |
|
|
ncol(tmp)<-ncol(r1)
|
94 |
|
|
nrow(tmp)<-nrow(r1)
|
95 |
|
|
extent(tmp)<-extent(r1)
|
96 |
|
|
projection(tmp)<-CRS
|
97 |
|
|
tmp2<-tmp
|
98 |
|
|
values(tmp)<-xy_latlon[,1]
|
99 |
|
|
values(tmp2)<-xy_latlon[,2]
|
100 |
|
|
|
101 |
|
|
r<-stack(N,E,Nw,Ew,tmp,tmp2)
|
102 |
|
|
rnames<-c("Northness","Eastness","Northness_w","Eastness_w", "lon","lat")
|
103 |
|
|
layerNames(r)<-rnames
|
104 |
|
|
s_raster<-addLayer(s_raster, r)
|
105 |
|
|
rm(r)
|
106 |
|
|
### adding var
|
107 |
|
|
ghcn = transform(ghcn,Northness = cos(ASPECT*pi/180)) #Adding a variable to the dataframe
|
108 |
|
|
ghcn = transform(ghcn,Eastness = sin(ASPECT*pi/180)) #adding variable to the dataframe.
|
109 |
|
|
ghcn = transform(ghcn,Northness_w = sin(slope*pi/180)*cos(ASPECT*pi/180)) #Adding a variable to the dataframe
|
110 |
|
|
ghcn = transform(ghcn,Eastness_w = sin(slope*pi/180)*sin(ASPECT*pi/180)) #adding variable to the dataframe.
|
111 |
|
|
|
112 |
|
|
#Remove NA for LC and CANHEIGHT
|
113 |
|
|
ghcn$LC1[is.na(ghcn$LC1)]<-0
|
114 |
|
|
ghcn$LC3[is.na(ghcn$LC3)]<-0
|
115 |
|
|
ghcn$CANHEIGHT[is.na(ghcn$CANHEIGHT)]<-0
|
116 |
|
|
|
117 |
|
|
dates <-readLines(paste(path,"/",infile2, sep=""))
|
118 |
|
|
LST_dates <-readLines(paste(path,"/",infile3, sep=""))
|
119 |
|
|
#models <-readLines(paste(path,"/",infile4, sep=""))
|
120 |
|
|
|
121 |
|
|
#Model assessment: specific diagnostic/metrics for GAM
|
122 |
|
|
results_AIC<- matrix(1,1,models+3)
|
123 |
|
|
results_GCV<- matrix(1,1,models+3)
|
124 |
|
|
results_DEV<- matrix(1,1,models+3)
|
125 |
|
|
#results_RMSE_f<- matrix(1,length(models)+3)
|
126 |
|
|
|
127 |
|
|
#Model assessment: general diagnostic/metrics
|
128 |
|
|
results_RMSE <- matrix(1,1,models+3)
|
129 |
|
|
results_MAE <- matrix(1,1,models+3)
|
130 |
|
|
results_ME <- matrix(1,1,models+3) #There are 8+1 models
|
131 |
|
|
results_R2 <- matrix(1,1,models+3) #Coef. of determination for the validation dataset
|
132 |
|
|
|
133 |
|
|
results_RMSE_f<- matrix(1,1,models+3) #RMSE fit, RMSE for the training dataset
|
134 |
|
|
results_MAE_f <- matrix(1,1,models+3)
|
135 |
|
|
#Screening for bad values: value is tmax in this case
|
136 |
|
|
#ghcn$value<-as.numeric(ghcn$value)
|
137 |
|
|
ghcn_all<-ghcn
|
138 |
|
|
ghcn_test<-subset(ghcn,ghcn$value>-150 & ghcn$value<400)
|
139 |
|
|
ghcn_test2<-subset(ghcn_test,ghcn_test$ELEV_SRTM>0)
|
140 |
|
|
ghcn<-ghcn_test2
|
141 |
|
|
#coords<- ghcn[,c('x_OR83M','y_OR83M')]
|
142 |
|
|
|
143 |
|
|
###CREATING SUBSETS BY INPUT DATES AND SAMPLING
|
144 |
|
|
set.seed(seed_number) #Using a seed number allow results based on random number to be compared...
|
145 |
|
|
ghcn.subsets <-lapply(dates, function(d) subset(ghcn, ghcn$date==as.numeric(d))) #Producing a list of data frame, one data frame per date.
|
146 |
|
|
sampling<-vector("list",length(dates))
|
147 |
|
|
|
148 |
|
|
for(i in 1:length(dates)){
|
149 |
|
|
n<-nrow(ghcn.subsets[[i]])
|
150 |
|
|
ns<-n-round(n*prop) #Create a sample from the data frame with 70% of the rows
|
151 |
|
|
nv<-n-ns #create a sample for validation with prop of the rows
|
152 |
|
|
ind.training <- sample(nrow(ghcn.subsets[[i]]), size=ns, replace=FALSE) #This selects the index position for 70% of the rows taken randomly
|
153 |
|
|
ind.testing <- setdiff(1:nrow(ghcn.subsets[[i]]), ind.training)
|
154 |
|
|
sampling[[i]]<-ind.training
|
155 |
|
|
}
|
156 |
|
|
|
157 |
|
|
|
158 |
|
|
kriging_mod<-mclapply(1:length(dates), runKriging, mc.cores = 8)#This is the end bracket from mclapply(...) statement
|
159 |
|
|
|
160 |
|
|
#for(i in 1:length(dates)){ # start of the for loop #1
|
161 |
|
|
#i<-3 #Date 10 is used to test kriging
|
162 |
|
|
|
163 |
|
|
## Plotting and saving diagnostic measures
|
164 |
|
|
accuracy_tab_fun<-function(i,f_list){
|
165 |
|
|
tb<-f_list[[i]][[3]]
|
166 |
|
|
return(tb)
|
167 |
|
|
}
|
168 |
|
|
|
169 |
|
|
tb<-kriging_mod[[1]][[3]][0,] #empty data frame with metric table structure that can be used in rbinding...
|
170 |
|
|
tb_tmp<-kriging_mod #copy
|
171 |
|
|
|
172 |
|
|
for (i in 1:length(tb_tmp)){
|
173 |
|
|
tmp<-tb_tmp[[i]][[3]]
|
174 |
|
|
tb<-rbind(tb,tmp)
|
175 |
|
|
}
|
176 |
|
|
rm(tb_tmp)
|
177 |
|
|
|
178 |
|
|
for(i in 4:(models+3)){ # start of the for loop #1
|
179 |
|
|
tb[,i]<-as.numeric(as.character(tb[,i]))
|
180 |
|
|
}
|
181 |
|
|
|
182 |
|
|
tb_RMSE<-subset(tb, metric=="RMSE")
|
183 |
|
|
tb_MAE<-subset(tb,metric=="MAE")
|
184 |
|
|
tb_ME<-subset(tb,metric=="ME")
|
185 |
|
|
tb_R2<-subset(tb,metric=="R2")
|
186 |
|
|
tb_RMSE_f<-subset(tb, metric=="RMSE_f")
|
187 |
|
|
tb_MAE_f<-subset(tb,metric=="MAE_f")
|
188 |
|
|
|
189 |
|
|
tb_diagnostic1<-rbind(tb_RMSE,tb_MAE,tb_ME,tb_R2)
|
190 |
|
|
#tb_diagnostic2<-rbind(tb_,tb_MAE,tb_ME,tb_R2)
|
191 |
|
|
|
192 |
|
|
mean_RMSE<-sapply(tb_RMSE[,4:(models+3)],mean)
|
193 |
|
|
mean_MAE<-sapply(tb_MAE[,4:(models+3)],mean)
|
194 |
|
|
mean_R2<-sapply(tb_R2[,4:(models+3)],mean)
|
195 |
|
|
mean_ME<-sapply(tb_ME[,4:(models+3)],mean)
|
196 |
|
|
mean_MAE_f<-sapply(tb_MAE[,4:(models+3)],mean)
|
197 |
|
|
mean_RMSE_f<-sapply(tb_RMSE_f[,4:(models+3)],mean)
|
198 |
|
|
|
199 |
|
|
write.table(tb_diagnostic1, file= paste(path,"/","results2_fusion_Assessment_measure1",out_prefix,".txt",sep=""), sep=",")
|
200 |
|
|
write.table(tb, file= paste(path,"/","results2_fusion_Assessment_measure_all",out_prefix,".txt",sep=""), sep=",")
|
201 |
|
|
save(kriging_mod,file= paste(path,"/","results2_fusion_Assessment_measure_all",out_prefix,".RData",sep=""))
|
202 |
|
|
|
203 |
|
|
#### END OF SCRIPT #####
|