##################  Functions for use in the raster prediction stage   #######################################

############################ Interpolation in a given tile/region ##########################################

#This script contains 5 functions used in the interpolation of temperature in the specfied study/processing area:                             

# 1)predict_raster_model<-function(in_models,r_stack,out_filename)                                                             

# 2)fit_models<-function(list_formulas,data_training)           

# 3)runClimCAI<-function(j) : not working yet

# 4)runClim_KGFusion<-function(j,list_param)

# 5)runGAMFusion <- function(i,list_param) 

#

#AUTHOR: Benoit Parmentier                                                                       

#DATE: 03/12/2013                                                                                 

#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#363--   

##Comments and TODO:

#This script is meant to be for general processing tile by tile or region by region.

# Note that the functions are called from GAM_fusion_analysis_raster_prediction_mutlisampling.R.

# This will be expanded to other methods.

##################################################################################################

predict_raster_model<-function(in_models,r_stack,out_filename){

  #This functions performs predictions on a raster grid given input models.

  #Arguments: list of fitted models, raster stack of covariates

  #Output: spatial grid data frame of the subset of tiles

  list_rast_pred<-vector("list",length(in_models))

  for (i in 1:length(in_models)){

    mod <-in_models[[i]] #accessing GAM model ojbect "j"

    raster_name<-out_filename[[i]]

    if (inherits(mod,"gam")) {           #change to c("gam","autoKrige")

      raster_pred<- predict(object=r_stack,model=mod,na.rm=FALSE) #Using the coeff to predict new values.

      names(raster_pred)<-"y_pred"  

      writeRaster(raster_pred, filename=raster_name,overwrite=TRUE)  #Writing the data in a raster file format...(IDRISI)

      #print(paste("Interpolation:","mod", j ,sep=" "))

      list_rast_pred[[i]]<-raster_name

    }

  }

  if (inherits(mod,"try-error")) {

    print(paste("no gam model fitted:",mod[1],sep=" ")) #change message for any model type...

  }

  return(list_rast_pred)

}

fit_models<-function(list_formulas,data_training){

  #This functions several models and returns model objects.

  #Arguments: - list of formulas for GAM models

  #           - fitting data in a data.frame or SpatialPointDataFrame

  #Output: list of model objects 

  list_fitted_models<-vector("list",length(list_formulas))

  for (k in 1:length(list_formulas)){

    formula<-list_formulas[[k]]

    mod<- try(gam(formula, data=data_training)) #change to any model!!

    #mod<- try(autoKrige(formula, input_data=data_s,new_data=s_sgdf,data_variogram=data_s))

    model_name<-paste("mod",k,sep="")

    assign(model_name,mod) 

    list_fitted_models[[k]]<-mod

  }

  return(list_fitted_models) 

}

####

#TODO:

#Add log file and calculate time and sizes for processes-outputs

runClimCAI<-function(j){

  #Make this a function with multiple argument that can be used by mcmapply??

  #This creates clim fusion layers...still needs more code testing

  #Model and response variable can be changed without affecting the script

  prop_month<-0 #proportion retained for validation

  run_samp<-1

  list_formulas<-lapply(list_models,as.formula,env=.GlobalEnv) #mulitple arguments passed to lapply!!

  data_month<-dst[dst$month==j,] #Subsetting dataset for the relevant month of the date being processed

  LST_name<-lst_avg[j] # name of LST month to be matched

  data_month$LST<-data_month[[LST_name]]

  #TMax to model...

  data_month$y_var<-data_month$TMax #Adding TMax as the variable modeled

  mod_list<-fit_models(list_formulas,data_month) #only gam at this stage

  cname<-paste("mod",1:length(mod_list),sep="") #change to more meaningful name?

  names(mod_list)<-cname

  #Adding layer LST to the raster stack  

  pos<-match("elev",names(s_raster))

  layerNames(s_raster)[pos]<-"elev_1"

  pos<-match("LST",names(s_raster)) #Find the position of the layer with name "LST", if not present pos=NA

  s_raster<-dropLayer(s_raster,pos)      # If it exists drop layer

  LST<-subset(s_raster,LST_name)

  names(LST)<-"LST"

  #Screen for extreme values": this needs more thought, min and max val vary with regions

  #min_val<-(-15+273.16) #if values less than -15C then screen out (note the Kelvin units that will need to be changed later in all datasets)

  #r1[r1 < (min_val)]<-NA

  s_raster<-addLayer(s_raster,LST)            #Adding current month

  #Now generate file names for the predictions...

  list_out_filename<-vector("list",length(mod_list))

  names(list_out_filename)<-cname  

  for (k in 1:length(list_out_filename)){

    #j indicate which month is predicted

    data_name<-paste("clim_month_",j,"_",cname[k],"_",prop_month,

                     "_",run_samp,sep="")

    raster_name<-paste("fusion_",data_name,out_prefix,".tif", sep="")

    list_out_filename[[k]]<-raster_name

  }

  #now predict values for raster image...

  rast_clim_list<-predict_raster_model(mod_list,s_raster,list_out_filename)

  names(rast_clim_list)<-cname

  #Some modles will not be predicted...remove them

  rast_clim_list<-rast_clim_list[!sapply(rast_clim_list,is.null)] #remove NULL elements in list

  #Adding Kriging for Climatology options

  clim_xy<-coordinates(data_month)

  fitclim<-Krig(clim_xy,data_month$TMax,theta=1e5) #use TPS or krige 

  mod_krtmp1<-fitclim

  model_name<-"mod_kr"

  clim_rast<-interpolate(LST,fitclim) #interpolation using function from raster package

  #Saving kriged surface in raster images

  #data_name<-paste("clim_month_",j,"_",model_name,"_",prop_month,

  #                 "_",run_samp,sep="")

  #raster_name_clim<-paste("fusion_",data_name,out_prefix,".tif", sep="")

  #writeRaster(clim_rast, filename=raster_name_clim,overwrite=TRUE)  #Writing the data in a raster file format...(IDRISI)

  #now climatology layer

  #clim_rast<-LST-bias_rast

  data_name<-paste("clim_month_",j,"_",model_name,"_",prop_month,

                   "_",run_samp,sep="")

  raster_name_clim<-paste("fusion_",data_name,out_prefix,".tif", sep="")

  writeRaster(clim_rast, filename=raster_name_clim,overwrite=TRUE)  #Writing the data in a raster file format...(IDRISI)

  #Adding to current objects

  mod_list[[model_name]]<-mod_krtmp1

  #rast_bias_list[[model_name]]<-raster_name_bias

  rast_clim_list[[model_name]]<-raster_name_clim

  #Prepare object to return

  clim_obj<-list(rast_clim_list,data_month,mod_list,list_formulas)

  names(clim_obj)<-c("clim","data_month","mod","formulas")

  save(clim_obj,file= paste("clim_obj_month_",j,"_",out_prefix,".RData",sep=""))

  return(clim_obj) 

}

#

runClim_KGFusion<-function(j,list_param){

  #Make this a function with multiple argument that can be used by mcmapply??

  #Arguments: 

  #1)list_index: j 

  #2)covar_rast: covariates raster images used in the modeling

  #3)covar_names: names of input variables 

  #4)lst_avg: list of LST climatogy names, may be removed later on

  #5)list_models: list input models for bias calculation

  #6)dst: data at the monthly time scale

  #7)var: TMAX or TMIN, variable being interpolated

  #8)y_var_name: output name, not used at this stage

  #9)out_prefix

  #

  #The output is a list of four shapefile names produced by the function:

  #1) clim: list of output names for raster climatogies 

  #2) data_month: monthly training data for bias surface modeling

  #3) mod: list of model objects fitted 

  #4) formulas: list of formulas used in bias modeling

  ### PARSING INPUT ARGUMENTS

  #list_param_runGAMFusion<-list(i,clim_yearlist,sampling_obj,var,y_var_name, out_prefix)

  index<-list_param$j

  s_raster<-list_param$covar_rast

  covar_names<-list_param$covar_names

  lst_avg<-list_param$lst_avg

  list_models<-list_param$list_models

  dst<-list_param$dst #monthly station dataset

  var<-list_param$var

  y_var_name<-list_param$y_var_name

  out_prefix<-list_param$out_prefix

  #Model and response variable can be changed without affecting the script

  prop_month<-0 #proportion retained for validation

  run_samp<-1 #This option can be added later on if/when neeeded

  #### STEP 2: PREPARE DATA

  data_month<-dst[dst$month==j,] #Subsetting dataset for the relevant month of the date being processed

  LST_name<-lst_avg[j] # name of LST month to be matched

  data_month$LST<-data_month[[LST_name]]

  #Adding layer LST to the raster stack  

  covar_rast<-s_raster

  #names(s_raster)<-covar_names

  pos<-match("LST",names(s_raster)) #Find the position of the layer with name "LST", if not present pos=NA

  s_raster<-dropLayer(s_raster,pos)      # If it exists drop layer

  LST<-subset(s_raster,LST_name)

  names(LST)<-"LST"

  s_raster<-addLayer(s_raster,LST)            #Adding current month

  #LST bias to model...

  #if (var==TMAX): will need to modify to take into account TMAX, TMIN and LST_day,LST_night

  data_month$LSTD_bias<-data_month$LST-data_month$TMax

  data_month$y_var<-data_month$LSTD_bias #Adding bias as the variable modeled

  #### STEP3:  NOW FIT AND PREDICT  MODEL

  list_formulas<-lapply(list_models,as.formula,env=.GlobalEnv) #mulitple arguments passed to lapply!!

  mod_list<-fit_models(list_formulas,data_month) #only gam at this stage

  cname<-paste("mod",1:length(mod_list),sep="") #change to more meaningful name?

  names(mod_list)<-cname

  #Now generate file names for the predictions...

  list_out_filename<-vector("list",length(mod_list))

  names(list_out_filename)<-cname  

  for (k in 1:length(list_out_filename)){

    #j indicate which month is predicted

    data_name<-paste("bias_LST_month_",j,"_",cname[k],"_",prop_month,

                     "_",run_samp,sep="")

    raster_name<-paste("fusion_",data_name,out_prefix,".tif", sep="")

    list_out_filename[[k]]<-raster_name

  }

  #now predict values for raster image...

  rast_bias_list<-predict_raster_model(mod_list,s_raster,list_out_filename)

  names(rast_bias_list)<-cname

  #Some modles will not be predicted...remove them

  rast_bias_list<-rast_bias_list[!sapply(rast_bias_list,is.null)] #remove NULL elements in list

  mod_rast<-stack(rast_bias_list)  #stack of bias raster images from models

  rast_clim_list<-vector("list",nlayers(mod_rast))

  names(rast_clim_list)<-names(rast_bias_list)

  for (k in 1:nlayers(mod_rast)){

    clim_fus_rast<-LST-subset(mod_rast,k)

    data_name<-paste("clim_LST_month_",j,"_",names(rast_clim_list)[k],"_",prop_month,

                     "_",run_samp,sep="")

    raster_name<-paste("fusion_",data_name,out_prefix,".tif", sep="")

    rast_clim_list[[k]]<-raster_name

    writeRaster(clim_fus_rast, filename=raster_name,overwrite=TRUE)  #Wri

  }

  #### STEP 4:Adding Kriging for Climatology options

  bias_xy<-coordinates(data_month)

  fitbias<-Krig(bias_xy,data_month$LSTD_bias,theta=1e5) #use TPS or krige 

  mod_krtmp1<-fitbias

  model_name<-"mod_kr"

  bias_rast<-interpolate(LST,fitbias) #interpolation using function from raster package

  #Saving kriged surface in raster images

  data_name<-paste("bias_LST_month_",j,"_",model_name,"_",prop_month,

                   "_",run_samp,sep="")

  raster_name_bias<-paste("fusion_",data_name,out_prefix,".tif", sep="")

  writeRaster(bias_rast, filename=raster_name_bias,overwrite=TRUE)  #Writing the data in a raster file format...(IDRISI)

  #now climatology layer

  clim_rast<-LST-bias_rast

  data_name<-paste("clim_LST_month_",j,"_",model_name,"_",prop_month,

                   "_",run_samp,sep="")

  raster_name_clim<-paste("fusion_",data_name,out_prefix,".tif", sep="")

  writeRaster(clim_rast, filename=raster_name_clim,overwrite=TRUE)  #Writing the data in a raster file format...(IDRISI)

  #Adding to current objects

  mod_list[[model_name]]<-mod_krtmp1

  rast_bias_list[[model_name]]<-raster_name_bias

  rast_clim_list[[model_name]]<-raster_name_clim

  #### STEP 5: Prepare object and return

  clim_obj<-list(rast_bias_list,rast_clim_list,data_month,mod_list,list_formulas)

  names(clim_obj)<-c("bias","clim","data_month","mod","formulas")

  save(clim_obj,file= paste("clim_obj_month_",j,"_",out_prefix,".RData",sep=""))

  return(clim_obj)

}

## Run function for kriging...?

#runGAMFusion <- function(i) {            # loop over dates

runGAMFusion <- function(i,list_param) {            # loop over dates

    #### Change this to allow explicitly arguments...

  #Arguments: 

  #1)index: loop list index for individual run/fit

  #2)clim_year_list: list of climatology files for all models...(12*nb of models)

  #3)sampling_obj: contains, data per date/fit, sampling information

  #4)dst: data at the monthly time scale

  #5)var: variable predicted -TMAX or TMIN

  #6)y_var_name: name of the variable predicted - dailyTMax, dailyTMin

  #7)out_prefix

  #

  #The output is a list of four shapefile names produced by the function:

  #1) list_temp: y_var_name

  #2) rast_clim_list: list of files for temperature climatology predictions

  #3) delta: list of files for temperature delta predictions

  #4) data_s: training data

  #5) data_v: testing data

  #6) sampling_dat: sampling information for the current prediction (date,proportion of holdout and sample number)

  #7) mod_kr: kriging delta fit, field package model object

  ### PARSING INPUT ARGUMENTS

  #list_param_runGAMFusion<-list(i,clim_yearlist,sampling_obj,var,y_var_name, out_prefix)

  rast_clim_yearlist<-list_param$clim_yearlist

  sampling_obj<-list_param$sampling_obj

  ghcn.subsets<-sampling_obj$ghcn_data_day

  sampling_dat <- sampling_obj$sampling_dat

  sampling <- sampling_obj$sampling_index

  var<-list_param$var

  y_var_name<-list_param$y_var_name

  out_prefix<-list_param$out_prefix

  dst<-list_param$dst #monthly station dataset

  ##########

  # STEP 1 - interpolate delta across space

  ############# Read in information and get traiing and testing stations

  date<-strptime(sampling_dat$date[i], "%Y%m%d")   # interpolation date being processed

  month<-strftime(date, "%m")          # current month of the date being processed

  LST_month<-paste("mm_",month,sep="") # name of LST month to be matched

  proj_str<-proj4string(dst) #get the local projection information from monthly data

  ###Regression part 1: Creating a validation dataset by creating training and testing datasets

  data_day<-ghcn.subsets[[i]]

  mod_LST <- ghcn.subsets[[i]][,match(LST_month, names(ghcn.subsets[[i]]))]  #Match interpolation date and monthly LST average

  data_day$LST <- as.data.frame(mod_LST)[,1] #Add the variable LST to the dataset

  dst$LST<-dst[[LST_month]] #Add the variable LST to the monthly dataset

  ind.training<-sampling[[i]]

  ind.testing <- setdiff(1:nrow(data_day), ind.training)

  data_s <- data_day[ind.training, ]   #Training dataset currently used in the modeling

  data_v <- data_day[ind.testing, ]    #Testing/validation dataset using input sampling

  ns<-nrow(data_s)

  nv<-nrow(data_v)

  #i=1

  date_proc<-sampling_dat$date[i]

  date_proc<-strptime(sampling_dat$date[i], "%Y%m%d")   # interpolation date being processed

  mo<-as.integer(strftime(date_proc, "%m"))          # current month of the date being processed

  day<-as.integer(strftime(date_proc, "%d"))

  year<-as.integer(strftime(date_proc, "%Y"))

  ##########

  # STEP 2 - JOIN DAILY AND MONTHLY STATION INFORMATION

  ##########

  modst<-dst[dst$month==mo,] #Subsetting dataset for the relevant month of the date being processed

  #Change to y_var...could be TMin

  #modst$LSTD_bias <- modst$LST-modst$y_var

  modst$LSTD_bias <- modst$LST-modst$TMax; #That is the difference between the monthly LST mean and monthly station mean

  #Clearn out this part: make this a function call

  x<-as.data.frame(data_v)

  d<-as.data.frame(data_s)

  #x[x$value==-999.9]<-NA

  for (j in 1:nrow(x)){

    if (x$value[j]== -999.9){

      x$value[j]<-NA

    }

  }

  for (j in 1:nrow(d)){

    if (d$value[j]== -999.9){

      d$value[j]<-NA

    }

  }

  #x[x$value==-999.9]<-NA

  #d[d$value==-999.9]<-NA

  pos<-match("value",names(d)) #Find column with name "value"

  #names(d)[pos]<-c("dailyTmax")

  names(d)[pos]<-y_var_name

  names(x)[pos]<-y_var_name

  #names(x)[pos]<-c("dailyTmax")

  pos<-match("station",names(d)) #Find column with name "value"

  names(d)[pos]<-c("id")

  names(x)[pos]<-c("id")

  names(modst)[1]<-c("id")       #modst contains the average tmax per month for every stations...

  dmoday <-merge(modst,d,by="id",suffixes=c("",".y2"))  

  xmoday <-merge(modst,x,by="id",suffixes=c("",".y2"))  

  mod_pat<-glob2rx("*.y2")   

  var_pat<-grep(mod_pat,names(dmoday),value=FALSE) # using grep with "value" extracts the matching names

  dmoday<-dmoday[,-var_pat]

  mod_pat<-glob2rx("*.y2")   

  var_pat<-grep(mod_pat,names(xmoday),value=FALSE) # using grep with "value" extracts the matching names

  xmoday<-xmoday[,-var_pat] #Removing duplicate columns

  data_v<-xmoday

  #dmoday contains the daily tmax values for training with TMax being the monthly station tmax mean

  #xmoday contains the daily tmax values for validation with TMax being the monthly station tmax mean

  ##########

  # STEP 3 - interpolate daily delta across space

  ##########

  #Change to take into account TMin and TMax

  daily_delta<-dmoday$dailyTmax-dmoday$TMax

  daily_delta_xy<-as.matrix(cbind(dmoday$x,dmoday$y))

  fitdelta<-Krig(daily_delta_xy,daily_delta,theta=1e5) #use TPS or krige

  mod_krtmp2<-fitdelta

  model_name<-paste("mod_kr","day",sep="_")

  data_s<-dmoday #put the 

  data_s$daily_delta<-daily_delta

  #########

  # STEP 4 - Calculate daily predictions - T(day) = clim(month) + delta(day)

  #########

  rast_clim_list<-rast_clim_yearlist[[mo]]  #select relevant month

  rast_clim_month<-raster(rast_clim_list[[1]])

  daily_delta_rast<-interpolate(rast_clim_month,fitdelta) #Interpolation of the bias surface...

  #Saving kriged surface in raster images

  data_name<-paste("daily_delta_",sampling_dat$date[i],"_",sampling_dat$prop[i],

                   "_",sampling_dat$run_samp[i],sep="")

  raster_name_delta<-paste("fusion_",data_name,out_prefix,".tif", sep="")

  writeRaster(daily_delta_rast, filename=raster_name_delta,overwrite=TRUE)  #Writing the data in a raster file format...(IDRISI)

  #Now predict daily after having selected the relevant month

  temp_list<-vector("list",length(rast_clim_list))  

  for (j in 1:length(rast_clim_list)){

    rast_clim_month<-raster(rast_clim_list[[j]])

    temp_predicted<-rast_clim_month+daily_delta_rast

    data_name<-paste(y_var_name,"_predicted_",names(rast_clim_list)[j],"_",

                     sampling_dat$date[i],"_",sampling_dat$prop[i],

                     "_",sampling_dat$run_samp[i],sep="")

    raster_name<-paste("fusion_",data_name,out_prefix,".tif", sep="")

    writeRaster(temp_predicted, filename=raster_name,overwrite=TRUE) 

    temp_list[[j]]<-raster_name

  }

  ##########

  # STEP 5 - Prepare output object to return

  ##########

  mod_krtmp2<-fitdelta

  model_name<-paste("mod_kr","day",sep="_")

  names(temp_list)<-names(rast_clim_list)

  coordinates(data_s)<-cbind(data_s$x,data_s$y)

  proj4string(data_s)<-proj_str

  coordinates(data_v)<-cbind(data_v$x,data_v$y)

  proj4string(data_v)<-proj_str

  delta_obj<-list(temp_list,rast_clim_list,raster_name_delta,data_s,

                  data_v,sampling_dat[i,],mod_krtmp2)

  obj_names<-c(y_var_name,"clim","delta","data_s","data_v",

               "sampling_dat",model_name)

  names(delta_obj)<-obj_names

  save(delta_obj,file= paste("delta_obj_",sampling_dat$date[i],"_",sampling_dat$prop[i],

                                "_",sampling_dat$run_samp[i],out_prefix,".RData",sep=""))

  return(delta_obj)

}