#MODIFIED ON: 07/18/2014            

library(gridExtra)                           # Combining lattice plots

library(colorRamps)                          # Palette/color ramps for symbology

library(ncf)                                 # No paramtric covariance function

out_prefix<-"_07182014"

table_ac <- round(table_ac,digit=3) #roundto three digits the differences

file_name<-paste("table4_comparisons_interpolation_methods_avg_paper","_",out_prefix,".txt",sep="")

l_s_table_term_tb <- vector("list",length(l_obj))

  #extract models for every day

  #s.table_term_tb <-extract_from_list_obj(list_models_info,"s.table_term")

  s.table_term_list <- extract_list_from_list_obj(list_models_info,"s.table_term")

  names(s.table_term_list) <- dates #adding names to the list

  #Adding dates to df

  s.table_term_list <- add_rownames_list_df(s.table_term_list)  

  s.table_term_tb <- do.call(rbind.fill,s.table_term_list) 

  #Adding month to df

  #s.table_term_tb <- add_month_tag(s.table_term_tb)

  s.table_term_tb$month <- add_month_tag(s.table_term_tb,"rownames")

  l_s_table_term_tb[[k]] <- s.table_term_tb

##Add new figure:

s.table_term_tb <- l_s_table_term_tb[[2]] #baseline 2 (lat*lon)

s.table_term_tb <- l_s_table_term_tb[[1]] #baseline 1 (lat*lon+elev_s)

tt <- aggregate(s.table_term_tb$p_val_rec2~s.table_term_tb$month,FUN=mean)

tt <- aggregate(s.table_term_tb$p_val_rec2~s.table_term_tb$month,FUN=mean)

plot(tt)

test <- subset(s.table_term_tb,mod_name=="mod4" & term_name=="s(LST)")

tt<-aggregate(test$p_val_rec3~test$month,FUN=mean)

plot(tt,type="l",ylim=c(0.2,1))

test2 <- subset(s.table_term_tb,mod_name=="mod4" & term_name=="s(elev_s)")

tt2<-aggregate(test2$p_val_rec2~test2$month,FUN=mean)

plot(tt2)

lines(tt2)

test1 <- subset(s.table_term_tb,mod_name=="mod1" & term_name=="s(elev_s)")

tt1<-aggregate(test1$p_val_rec2~test1$month,FUN=mean)

plot(tt1)

#model with LST and elev

tb1_mod4_month <- raster_prediction_obj_1$summary_month_metrics_v[[4]] #note that this is for model1

#model with lev only

b1_mod1_month <- raster_prediction_obj_1$summary_month_metrics_v[[1]] #note that this is for model1

tb1_mod1_month<-raster_prediction_obj_1$summary_month_metrics_v[[1]] #note that this is for model1

plot((tb1_mod4_month$rme) - tb1_mod1_month$rmse)

#now get correlation with LST and elev

list_data_s <-extract_list_from_list_obj(raster_prediction_obj_1$validation_mod_obj,"data_s")

list_data_v <-extract_list_from_list_obj(raster_prediction_obj_1$validation_mod_obj,"data_v")

data_v_combined <-convert_spdf_to_df_from_list(list_data_v) #long rownames

#LSTD_bias_avgm<-aggregate(LSTD_bias~month,data=data_month_all,mean)

#LSTD_bias_sdm<-aggregate(LSTD_bias~month,data=data_month_all,sd)

data_v_combined

LST_avgm <- aggregate(LST~month+id,data=data_v_combined,mean)

#test <- aggregate(LST+dailyTmax~month+id,data=data_v_combined,mean)

TMax_avgm <- aggregate(dailyTmax~month+id,data=data_v_combined,mean)

elev_id <- aggregate(elev~id,data=data_v_combined,mean)

l_df_cor<- vector("list",length=12)

for(i in 1:12){

  df_LST <- subset(LST_avgm,month==i)

  df_TMax <- subset(TMax_avgm,month==i)

  df <-merge(df_LST,df_TMax,by="id",all=F)

  df_tmp <- merge(df,elev_id,by="id",all=F)

  cor_LST_dailyTmax <- cor(df_tmp$LST,df_tmp$dailyTmax)

  cor_elev_dailyTmax <- cor(df_tmp$elev,df_tmp$dailyTmax)

  df_cor <- data.frame(LST_tmax=cor_LST_dailyTmax,elev_tmax=cor_elev_dailyTmax)

  l_df_cor[[i]] <- df_cor

}

#cor elev~ tmax by month

#cor LST tmax by month

df_cor <- do.call(rbind,l_df_cor)

plot(df_cor$LST_tmax,ylim=c(-1,1),col="red",type="l")

lines(df_cor$elev_tmax,ylim=c(-1,1),col="blue")