1 |
eda4f37a
|
Benoit Parmentier
|
######################################## IBS 2013 POSTER #######################################
|
2 |
|
|
############################ Scripts for figures and analyses for the the IBS poster #####################################
|
3 |
|
|
#This script creates the figures used in the IBS 2013 poster.
|
4 |
|
|
#It uses inputs from interpolation objects created at earlier stages... #
|
5 |
|
|
#AUTHOR: Benoit Parmentier #
|
6 |
|
|
#DATE: 12/27/2012 #
|
7 |
|
|
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#491-- #
|
8 |
|
|
###################################################################################################
|
9 |
|
|
|
10 |
|
|
###Loading R library and packages
|
11 |
|
|
#library(gtools) # loading some useful tools
|
12 |
|
|
library(mgcv) # GAM package by Wood 2006 (version 2012)
|
13 |
|
|
library(sp) # Spatial pacakge with class definition by Bivand et al. 2008
|
14 |
|
|
library(spdep) # Spatial package with methods and spatial stat. by Bivand et al. 2012
|
15 |
|
|
library(rgdal) # GDAL wrapper for R, spatial utilities (Keitt et al. 2012)
|
16 |
|
|
library(gstat) # Kriging and co-kriging by Pebesma et al. 2004
|
17 |
|
|
library(automap) # Automated Kriging based on gstat module by Hiemstra et al. 2008
|
18 |
|
|
library(spgwr)
|
19 |
|
|
library(maptools)
|
20 |
|
|
library(graphics)
|
21 |
|
|
library(parallel) # Urbanek S. and Ripley B., package for multi cores & parralel processing
|
22 |
|
|
library(raster)
|
23 |
|
|
library(rasterVis)
|
24 |
|
|
library(plotrix) # Draw circle on graph and additional plotting options
|
25 |
|
|
library(reshape) # Data format and type transformation
|
26 |
|
|
## Functions
|
27 |
|
|
#loading R objects that might have similar names
|
28 |
|
|
load_obj <- function(f)
|
29 |
|
|
{
|
30 |
|
|
env <- new.env()
|
31 |
|
|
nm <- load(f, env)[1]
|
32 |
|
|
env[[nm]]
|
33 |
|
|
}
|
34 |
|
|
|
35 |
|
|
###Parameters and arguments
|
36 |
|
|
|
37 |
|
|
infile1<- "ghcn_or_tmax_covariates_06262012_OR83M.shp" #GHCN shapefile containing variables for modeling 2010
|
38 |
|
|
#infile2<-"list_10_dates_04212012.txt" #List of 10 dates for the regression
|
39 |
|
|
infile2<-"list_365_dates_04212012.txt" #list of dates
|
40 |
|
|
infile3<-"LST_dates_var_names.txt" #LST dates name
|
41 |
|
|
#infile4<-"models_interpolation_05142012.txt" #Interpolation model names
|
42 |
|
|
infile5<-"mean_day244_rescaled.rst" #mean LST for day 244
|
43 |
|
|
inlistf<-"list_files_05032012.txt" #list of raster images containing the Covariates
|
44 |
|
|
infile6<-"OR83M_state_outline.shp"
|
45 |
|
|
#stat_loc<-read.table(paste(path,"/","location_study_area_OR_0602012.txt",sep=""),sep=",", header=TRUE)
|
46 |
|
|
|
47 |
|
|
obj_list<-"list_obj_12272012.txt" #Results of fusion from the run on ATLAS
|
48 |
|
|
#obj_list<-"list_obj_08262012.txt" #Results of fusion from the run on ATLAS
|
49 |
|
|
path<-"/home/parmentier/Data/IPLANT_project/methods_interpolation_comparison_10242012" #Jupiter LOCATION on Atlas for kriging #Jupiter Location on XANDERS
|
50 |
|
|
setwd(path)
|
51 |
|
|
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";
|
52 |
|
|
#Number of kriging model
|
53 |
|
|
out_prefix<-"methods_comp_12272012_" #User defined output prefix
|
54 |
|
|
|
55 |
|
|
filename<-sub(".shp","",infile1) #Removing the extension from file.
|
56 |
|
|
ghcn<-readOGR(".", filename) #reading shapefile
|
57 |
|
|
|
58 |
|
|
### PREPARING RASTER COVARIATES STACK #######
|
59 |
|
|
|
60 |
|
|
#CRS<-proj4string(ghcn) #Storing projection information (ellipsoid, datum,etc.)
|
61 |
|
|
lines<-read.table(paste(path,"/",inlistf,sep=""), sep="") #Column 1 contains the names of raster files
|
62 |
|
|
inlistvar<-lines[,1] #column 3 the list of models to use...?
|
63 |
|
|
|
64 |
|
|
inlistvar<-paste(path,"/",as.character(inlistvar),sep="")
|
65 |
|
|
covar_names<-as.character(lines[,2]) #Column two contains short names for covaraites
|
66 |
|
|
|
67 |
|
|
s_raster<- stack(inlistvar) #Creating a stack of raster images from the list of variables.
|
68 |
|
|
layerNames(s_raster)<-covar_names #Assigning names to the raster layers
|
69 |
|
|
projection(s_raster)<-proj_str
|
70 |
|
|
|
71 |
|
|
#Create mask
|
72 |
|
|
pos<-match("LC10",layerNames(s_raster))
|
73 |
|
|
LC10<-subset(s_raster,pos)
|
74 |
|
|
LC10[is.na(LC10)]<-0 #Since NA values are 0, we assign all zero to NA
|
75 |
|
|
mask_land<-LC10<100
|
76 |
|
|
mask_land_NA<-mask_land
|
77 |
|
|
mask_land_NA[mask_land_NA==0]<-NA
|
78 |
|
|
|
79 |
|
|
data_name<-"mask_land_OR"
|
80 |
|
|
raster_name<-paste(data_name,".rst", sep="")
|
81 |
|
|
writeRaster(mask_land, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...(IDRISI)
|
82 |
|
|
#writeRaster(r2, filename=raster_name,overwrite=TRUE) #Writing the data in a raster file format...(IDRISI)
|
83 |
|
|
|
84 |
|
|
pos<-match("ELEV_SRTM",layerNames(s_raster))
|
85 |
|
|
ELEV_SRTM<-raster(s_raster,pos)
|
86 |
|
|
elev<-ELEV_SRTM
|
87 |
|
|
elev[-0.050<elev]<-NA #Remove all negative elevation lower than 50 meters...
|
88 |
|
|
|
89 |
|
|
mask_elev_NA<-elev
|
90 |
|
|
|
91 |
|
|
pos<-match("mm_01",layerNames(s_raster))
|
92 |
|
|
mm_01<-subset(s_raster,pos)
|
93 |
|
|
mm_01<-mm_01-273.15
|
94 |
|
|
mm_01<-mask(mm_01,mask_land_NA)
|
95 |
|
|
#mention this is the last... files
|
96 |
|
|
|
97 |
|
|
##################### METHODS COMPARISON ###########################
|
98 |
|
|
|
99 |
|
|
######################################################################
|
100 |
|
|
# PART 1 : USING ACCURACY METRICS FOR FIVE METHODS COMPARISON
|
101 |
|
|
# Boxplots and histograms
|
102 |
|
|
#start function here...
|
103 |
|
|
|
104 |
|
|
lines<-read.table(paste(path,"/",obj_list,sep=""), sep=",") #Column 1 contains the names RData objects
|
105 |
|
|
inlistobj<-lines[,1]
|
106 |
|
|
tinlistobj<-paste(path,"/",as.character(inlistobj),sep="")
|
107 |
|
|
obj_names<-as.character(lines[,2]) #Column two contains short names for obj. model
|
108 |
|
|
|
109 |
|
|
tb_metrics_fun<-function(list_obj,path_data,names_obj){
|
110 |
|
|
nel<-length(inlistobj)
|
111 |
|
|
#method_mod <-vector("list",nel) #list of one row data.frame
|
112 |
|
|
method_tb <-vector("list",nel) #list of one row data.frame
|
113 |
|
|
for (k in 1:length(inlistobj)){
|
114 |
|
|
#obj_tmp<-load_obj(as.character(inlistobj[i]))
|
115 |
|
|
#method_mod[[i]]<-obj_tmp
|
116 |
|
|
#names(method_mod[[i]])<-obj_names[i]
|
117 |
|
|
mod_tmp<-load_obj(as.character(inlistobj[k]))
|
118 |
|
|
tb<-mod_tmp[[1]][[3]][0,] #copy of the data.frame structure that holds the acuary metrics
|
119 |
|
|
#mod_tmp<-method_mod[[k]]
|
120 |
|
|
for (i in 1:365){ # Assuming 365 days of prediction
|
121 |
|
|
tmp<-mod_tmp[[i]][[3]]
|
122 |
|
|
tb<-rbind(tb,tmp)
|
123 |
|
|
}
|
124 |
|
|
rm(mod_tmp)
|
125 |
|
|
for(i in 4:(ncol(tb))){ # start of the for loop #1
|
126 |
|
|
tb[,i]<-as.numeric(as.character(tb[,i]))
|
127 |
|
|
}
|
128 |
|
|
method_tb[[k]]<-tb
|
129 |
|
|
}
|
130 |
|
|
names(method_tb)<-names_obj
|
131 |
|
|
return(method_tb)
|
132 |
|
|
}
|
133 |
|
|
|
134 |
|
|
tmp44<-tb_metrics_fun(as.character(inlistobj),path,obj_names)
|
135 |
|
|
#Condensed, and added other comparison, monthly comparison...:ok
|
136 |
|
|
|
137 |
|
|
plot_model_boxplot_combined_fun<-function(tb_list,path_data,obj_names,mod_selected,out_prefix,layout_m){
|
138 |
|
|
|
139 |
|
|
method_stat<-vector("list",length(obj_names)) #This contains summary information based on accuracy metrics (MAE,RMSE)
|
140 |
|
|
names_method<-obj_names
|
141 |
|
|
metrics<-c("MAE","RMSE")
|
142 |
|
|
tb_metric_list<-vector("list",length(metrics))
|
143 |
|
|
tb_metric_list_na<-vector("list",length(metrics))
|
144 |
|
|
mean_list<-vector("list",length(metrics))
|
145 |
|
|
sd_list<-vector("list",length(metrics))
|
146 |
|
|
na_mod_list<-vector("list",length(metrics))
|
147 |
|
|
|
148 |
|
|
for(i in 1:length(metrics)){ # Reorganizing information in terms of metrics
|
149 |
|
|
#for(k in 1:length(tb_list)){ # start of the for main loop to all methods
|
150 |
|
|
#tb<-tb_list[[k]]
|
151 |
|
|
#metrics<-as.character(unique(tb$metric)) #Name of accuracy metrics (RMSE,MAE etc.)
|
152 |
|
|
metric_name<-paste("tb_t_",metrics[i],sep="")
|
153 |
|
|
png(paste("boxplot",metric_name,out_prefix,"_combined.png", sep="_"),height=480*layout_m[1],width=480*layout_m[2])
|
154 |
|
|
par(mfrow=layout_m)
|
155 |
|
|
for(k in 1:length(tb_list)){ # start of the for main loop to all methods
|
156 |
|
|
#}#for(i in 1:length(metrics)){ # Reorganizing information in terms of metrics
|
157 |
|
|
tb<-tb_list[[k]]
|
158 |
|
|
#metric_name<-paste("tb_t_",metrics[i],sep="")
|
159 |
|
|
tb_metric<-subset(tb, metric==metrics[i])
|
160 |
|
|
assign(metric_name,tb_metric)
|
161 |
|
|
tb_metric_list[[i]]<-tb_metric
|
162 |
|
|
tb_processed<-tb_metric
|
163 |
|
|
mod_pat<-glob2rx("mod*")
|
164 |
|
|
var_pat<-grep(mod_pat,names(tb_processed),value=FALSE) # using grep with "value" extracts the matching names
|
165 |
|
|
#mod_pat<-mod_selected
|
166 |
|
|
#var_pat<-grep(mod_pat,names(tb_processed),value=FALSE) # using grep with "value" extracts the matching names
|
167 |
|
|
na_mod<-colSums(!is.na(tb_processed[,var_pat]))
|
168 |
|
|
for (j in 1:length(na_mod)){
|
169 |
|
|
if (na_mod[j]<183){
|
170 |
|
|
tmp_name<-names(na_mod)[j]
|
171 |
|
|
pos<-match(tmp_name,names(tb_processed))
|
172 |
|
|
tb_processed<-tb_processed[,-pos] #Remove columns that have too many missing values!!!
|
173 |
|
|
}
|
174 |
|
|
}
|
175 |
|
|
tb_metric_list_na[[i]]<-tb_processed
|
176 |
|
|
mod_pat<-glob2rx("mod*")
|
177 |
|
|
var_pat<-grep(mod_pat,names(tb_processed),value=FALSE)
|
178 |
|
|
#Plotting box plots
|
179 |
|
|
|
180 |
|
|
#png(paste("boxplot",metric_name,names_methods[k],out_prefix,".png", sep="_"))
|
181 |
|
|
boxplot(tb_processed[,var_pat],main=names_methods[k], ylim=c(1,5),
|
182 |
|
|
ylab= metrics[i], outline=FALSE) #ADD TITLE RELATED TO METHODS...
|
183 |
|
|
|
184 |
|
|
#Add assessment of missing prediction over the year.
|
185 |
|
|
mean_metric<-sapply(tb_processed[,var_pat],mean,na.rm=T)
|
186 |
|
|
sd_metric<-sapply(tb_processed[,var_pat],sd,na.rm=T)
|
187 |
|
|
mean_list[[i]]<-mean_metric
|
188 |
|
|
sd_list[[i]]<-sd_metric
|
189 |
|
|
na_mod_list[[i]]<-na_mod_list
|
190 |
|
|
#Now calculate monthly averages and overall averages over full year
|
191 |
|
|
method_stat<-list(mean_list,sd_list,na_mod_list)
|
192 |
|
|
method_stat[[k]]<-list(mean_list,sd_list,na_mod_list)
|
193 |
|
|
names(method_stat[[k]])<-c("mean_metrics","sd_metrics","na_metrics")
|
194 |
|
|
names(mean_list)<-metrics
|
195 |
|
|
method_mean[[k]]<-mean_list
|
196 |
|
|
names_methods<-obj_names
|
197 |
|
|
#names(method_stat)<-obj_names
|
198 |
|
|
}
|
199 |
|
|
dev.off() #Close file where figures are drawn
|
200 |
|
|
}
|
201 |
|
|
return(method_stat)
|
202 |
|
|
}
|
203 |
|
|
|
204 |
|
|
tb_list<-tmp44
|
205 |
|
|
mod_selected<-""
|
206 |
|
|
layout_plot<-c(1,5)
|
207 |
|
|
mean_methods<-plot_model_boxplot_fun(tb_list,path,obj_names,mod_selected,out_prefix)
|
208 |
|
|
mean_methods_2<-plot_model_boxplot_combined_fun(tb_list,path,obj_names,mod_selected,out_prefix,layout_m=layout_plot)
|
209 |
|
|
|
210 |
|
|
##################### PART II #######################
|
211 |
|
|
|
212 |
|
|
##PLOTTING OF ONE DATE TO COMPARE METHODS!!!
|
213 |
|
|
|
214 |
|
|
lf_raster_fus<-"_365d_GAM_fusion_all_lstd_12272012.rst"
|
215 |
|
|
lf_raster_cai<-"_365d_GAM_CAI4_all_12272012.rst"
|
216 |
|
|
date_selected<-"20100103"
|
217 |
|
|
titles<-list(cai=c("cai mod1","cai mod4","cai mod7"),
|
218 |
|
|
fusion=c("fusion mod1"," fusion mod4"," fusion mod7"))
|
219 |
|
|
|
220 |
|
|
mask_rast<-mask_elev_NA
|
221 |
|
|
mod_selected1<-c(1,4,7)
|
222 |
|
|
mod_selected2<-c(1,4,7)
|
223 |
|
|
#lf_raster_fus<-file_pat1
|
224 |
|
|
#lf_raster_cai<-file_pat2
|
225 |
|
|
file_pat1<-glob2rx(paste("*tmax_predicted*",date_selected,"*",lf_raster_cai,sep="")) #Search for files in relation to fusion
|
226 |
|
|
#lf_cai<-list.files(pattern=file_pat) #Search for files in relation to fusion
|
227 |
|
|
file_pat2<-glob2rx(paste("*tmax_predicted*",date_selected,"*",lf_raster_fus,sep="")) #Search for files in relation to fusion
|
228 |
|
|
#lf_fus<-list.files(pattern=file_pat) #Search for files in relation to fusion
|
229 |
|
|
layout_plot<-c(2,3)
|
230 |
|
|
raster_plots_interpolation_fun<-function(file_pat1,file_pat2,mod_selected1,mod_selected2,titles,mask_rast,
|
231 |
|
|
layout_m,out_suffix){
|
232 |
|
|
layout_m<-layout_plot
|
233 |
|
|
lf_cai<-list.files(pattern=file_pat1) #Search for files in relation to fusion
|
234 |
|
|
lf_fus<-list.files(pattern=file_pat2) #Search for files in relation to fusion
|
235 |
|
|
|
236 |
|
|
r1<-stack(lf_cai[mod_selected1]) #CAI
|
237 |
|
|
r2<-stack(lf_fus[mod_selected2])#FUS
|
238 |
|
|
predictions<-stack(r1,r2)
|
239 |
|
|
predictions<-mask(predictions,mask_rast)
|
240 |
|
|
layerNames(predictions)<-unlist(titles)
|
241 |
|
|
|
242 |
|
|
s.range <- c(min(minValue(predictions)), max(maxValue(predictions)))
|
243 |
|
|
col.breaks <- pretty(s.range, n=50)
|
244 |
|
|
lab.breaks <- pretty(s.range, n=5)
|
245 |
|
|
temp.colors <- colorRampPalette(c('blue', 'white', 'red'))
|
246 |
|
|
X11(height=6,width=12)
|
247 |
|
|
#plot(predictions, breaks=col.breaks, col=rev(heat.colors(length(col.breaks)-1)),
|
248 |
|
|
# axis=list(at=lab.breaks, labels=lab.breaks))
|
249 |
|
|
plot(predictions, breaks=col.breaks, col=temp.colors(length(col.breaks)-1),
|
250 |
|
|
axis=list(at=lab.breaks, labels=lab.breaks))
|
251 |
|
|
#plot(reg_outline, add=TRUE)
|
252 |
|
|
savePlot(paste("comparison_one_date_CAI_fusion_tmax_prediction_",date_selected,out_prefix,".png", sep=""),type="png")
|
253 |
|
|
#png(paste("boxplot",metric_name,out_prefix,"_combined.png", sep="_"),height=480*layout_m[1],width=480*layout_m[2])
|
254 |
|
|
#par(mfrow=layout_m)
|
255 |
|
|
png(paste("comparison_one_date_CAI_fusion_tmax_prediction_levelplot_",date_selected,out_prefix,".png", sep=""),
|
256 |
|
|
height=480*layout_m[1],width=480*layout_m[2])
|
257 |
|
|
levelplot(predictions,main="comparison", ylab=NULL,xlab=NULL,par.settings = list(axis.text = list(font = 2, cex = 1.5),
|
258 |
|
|
par.main.text=list(font=2,cex=2),strip.background=list(col="white")),par.strip.text=list(font=2,cex=1.5),
|
259 |
|
|
#col.regions=temp.colors,at=seq(-1,1,by=0.02))
|
260 |
|
|
col.regions=temp.colors(25))
|
261 |
|
|
dev.off()
|
262 |
|
|
#savePlot(paste("comparison_one_date_CAI_fusion_tmax_prediction_levelplot_",date_selected,out_prefix,".png", sep=""),type="png")
|
263 |
|
|
}
|
264 |
|
|
|
265 |
|
|
raster_plots_interpolation_fun(file_pat1,file_pat2,
|
266 |
|
|
mod_selected1,mod_selected2,titles,mask_rast,layout_plot,out_prefix)
|
267 |
|
|
|
268 |
|
|
|
269 |
|
|
#### FIGURE 3: Transect map
|
270 |
|
|
|
271 |
|
|
### FIGURE 4: transect plot
|
272 |
|
|
|
273 |
|
|
|
274 |
|
|
|
275 |
|
|
#### END OF THE SCRIPT #########
|
276 |
|
|
|
277 |
|
|
|
278 |
|
|
#This can be entered as textfile or option later...ok for running now on 12/07/2012
|
279 |
|
|
|
280 |
|
|
|
281 |
|
|
#Figure 1: Boxplots for all methods and models...
|