##################    Data preparation for interpolation   #######################################

############################ Covariate production for a given tile/region ##########################################

#This script produces covariates raster for a a specfied study area.                             

# It requires the following inputs:                                                              

# 1)list of modis tiles or shape file with region outline            

# 2)input names for global covariates:

# -SRTM CGIAR 1 km

# -Canopy heihgt (Simard)

# -land cover concensus (Jetz lab)

# -MODIS LST: mean and obs

#3) The output is a multiband file in tif format with projected covariates for the processing region/tile.             

#AUTHOR: Benoit Parmentier                                                                       

#DATE: 03/21/2013                                                                                 

#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.

#We must decide on a local projection. This can best set up from the tile/region extent: for now use

#- lcc with two standard paralells and one central meridian in the middle of the region.

#- produce a distance to ocean layer that is global.

#- do not keep output in memory??

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

create_modis_tiles_region<-function(modis_grid,tiles){

  #This functions returns a subset of tiles from the modis grdi.

  #Arguments: modies grid tile,list of tiles

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

  h_list<-lapply(tiles,substr,start=2,stop=3) #passing multiple arguments

  v_list<-lapply(tiles,substr,start=5,stop=6) #passing multiple arguments

  selected_tiles<-subset(subset(modis_grid,subset = h %in% as.numeric (h_list) ),

                         subset = v %in% as.numeric(v_list)) 

  return(selected_tiles)

}

#This function is very very slow not be used most likely

create_polygon_from_extent<-function(reg_ref_rast){

  #This functions returns polygon sp from input rast

  #Arguments: input ref rast

  #Output: spatial polygon

  set1f <- function(x){rep(1, x)}

  tmp_rast <- init(reg_ref_rast, fun=set1f, overwrite=TRUE)

  reg_outline_poly<-rasterToPolygons(tmp_rast)

  return(reg_outline_poly)

}

create_raster_region <-function(raster_name,reg_ref_rast){

  #This functions returns a subset of tiles from the modis grdi.

  #Arguments: raster name of the file,reference file with

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

  layer_rast<-raster(raster_name)

  new_proj<-proj4string(layer_rast)                  #Extract coordinates reference system in PROJ4 format

  region_temp_projected<-projectExtent(reg_ref_rast,CRS(new_proj))     #Project from current to region coord. system

  layer_crop_rast<-crop(layer_rast, region_temp_projected) #crop using the extent from teh region tile

  #layer_projected_rast<-projectRaster(from=layer_crop_rast,crs=proj4string(reg_outline),method="ngb")

  layer_projected_rast<-projectRaster(from=layer_crop_rast,to=reg_ref_rast,method="ngb")

  return(layer_projected_rast)

}

mosaic_raster_list<-function(mosaic_list,out_names,out_path){

  #This functions returns a subset of tiles from the modis grid.

  #Arguments: modies grid tile,list of tiles

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

  #Note that rasters are assumed to be in the same projection system!!

  rast_list<-vector("list",length(mosaic_list))

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

    # read the individual rasters into a list of RasterLayer objects

    # this may be changed so that it is not read in the memory!!!

    input.rasters <- lapply(as.character(mosaic_list[[i]]), raster)

    mosaiced_rast<-input.rasters[[1]]

    for (k in 2:length(input.rasters)){

      mosaiced_rast<-mosaic(mosaiced_rast,input.rasters[[k]], fun=mean)

      #mosaiced_rast<-mosaic(mosaiced_rast,raster(input.rasters[[k]]), fun=mean)

    }

    data_name<-paste("mosaiced_",sep="") #can add more later...

    raster_name<-paste(data_name,out_names[i],".tif", sep="")

    writeRaster(mosaiced_rast, filename=file.path(out_path,raster_name),overwrite=TRUE)  

    #Writing the data in a raster file format...  

    rast_list[[i]]<-file.path(out_path,raster_name)

  }

  return(rast_list)

}

covariates_production_temperature<-function(list_param){

  #This functions produce covariates used in the interpolation of temperature.

  #It requires 15 arguments:

  #

  #

  #

  ###Loading R library and packages   

  library(RPostgreSQL)

  library(sp)                                             # Spatial pacakge with class definition by Bivand et al.

  library(spdep)                                          # Spatial pacakge with methods and spatial stat. by Bivand et al.

  library(rgdal)                                          # GDAL wrapper for R, spatial utilities

  library(raster)

  library(gtools)

  library(rasterVis)

  library(graphics)

  library(grid)

  library(lattice)

  ### Parameters and arguments

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

  ############ Main body: BEGIN--START OF THE SCRIPT ###################

  ##### STEP 1: PARSING ARGUMENTS

  var<-list_param$var

  in_path <-list_param$in_path

  out_path<- list_param$out_path

  lc_path<-list_param$lc_path 

  infile_modis_grid<-list_param$infile_modis_grid

  infile_elev<-list_param$infile_elev #this is the global file: replace later with the input produced by the DEM team

  infile_canheight<-list_param$infile_canheight #Canopy height

  list_tiles_modis<-list_param$list_tiles_modis #tile for Venezuela and surrounding area

  infile_reg_outline<-list_param$infile_reg_outline   #input region outline defined by polygon

  CRS_interp<-list_param$CRS_interp #local projection system

  CRS_locs_WGS84<-list_param$CRS_locs_WGS84 #

  out_region_name<-list_param$out_region_name  #generated on the fly

  out_suffix<-list_param$out_suffix 

  ref_rast_name<-list_param$ref_rast_name #local raster name defining resolution, exent, local projection--. set on the fly??

  covar_names<-list_param$covar_names 

  ##### SET UP STUDY AREA ####

  setwd(in_path)

  filename<-sub(".shp","",infile_modis_grid)       #Removing the extension from file.

  modis_grid<-readOGR(".", filename)     #Reading shape file using rgdal library

  #filename<-sub(".shp","",infile1)       #Removing the extension from file.

  #world_countries<-readOGR(".", filename)     #Reading shape file using rgdal library

  if (infile_reg_outline!=""){

    filename<-sub(".shp","",infile_reg_outline)   #Removing the extension from file.

    reg_outline<-readOGR(".", filename)

  }

  if (infile_reg_outline==""){

    reg_outline<-create_modis_tiles_region(modis_grid,list_tiles_modis) #problem...this does not 

    #align with extent of modis LST!!!

    writeOGR(reg_outline,dsn= ".",layer= paste("outline",out_region_name,"_",out_suffix,sep=""), 

             driver="ESRI Shapefile",overwrite_layer="TRUE")

  }

  #Should add option for a reference file here...

  #tmp<-extent(ref_rast)

  #modis_tiles<-create_modis_tiles_region(modis_grid,list_tiles_modis)

  ##Create covariates for the stuy area: pull everything from the same folder?

  #### STEP 2: process and/or produce covariates for the tile/region

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

  #1) LST climatology: project, mosaic

  i<-1

  tile<-list_tiles_modis[i]

  if (var=="TMIN"){

    lst_pat<-"LST_Night_1km"

  }

  if (var=="TMAX"){

    lst_pat<-"" #for the time being change at later stage...

    #day_pat<-"LST_Day_1km"

  }

  #Get list of files containing the LST averages

  pat_str2 <- glob2rx(paste("nobs","*",lst_pat,"*.tif",sep=""))

  tmp_str2<- mixedsort(list.files(pattern=pat_str2))

  pat_str1 <- glob2rx(paste("mean","*",lst_pat,"*.tif",sep=""))

  tmp_str1<- mixedsort(list.files(pattern=pat_str1))

  #add lines using grep to select tiles...

  #Format list for mosaicing: mosaic for every month the relevant number of files

  #out_rastnames<-paste("_lst_","nobs",out_suffix,sep="")

  out_rastnames<-paste("_",lst_pat,"_","nobs",out_suffix,sep="")

  list_date_names<-c("jan","feb","mar","apr","may","jun","jul","aug","sep","oct","nov","dec")

  mosaic_list<-split(tmp_str2,list_date_names)

  new_list<-vector("list",length(mosaic_list))

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

    j<-grep(list_date_names[i],mosaic_list,value=FALSE)

    names(mosaic_list)[j]<-list_date_names[i]

    new_list[i]<-mosaic_list[j]

  }

  mosaic_list<-new_list

  out_rastnames<-paste(list_date_names,out_rastnames,sep="")

  #reproject and crop if necessary

  #nobs_m_list<-list.files(pattern='mosaiced_.*._lst_nobs_VE_03182013.tif')

  nobs_m_list<-mosaic_raster_list(mosaic_list,out_rastnames,out_path)

  ##Now mosaic for mean: should reorder files!!

  pat_str1 <- glob2rx(paste("mean","*.tif",sep=""))

  tmp_str1<- mixedsort(list.files(pattern=pat_str1))

  out_rastnames<-paste("_",lst_pat,"_","mean",out_suffix,sep="")

  list_date_names<-c("jan","feb","mar","apr","may","jun","jul","aug","sep","oct","nov","dec")

  mosaic_list<-split(tmp_str1,list_date_names)

  new_list<-vector("list",length(mosaic_list))

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

    j<-grep(list_date_names[i],mosaic_list,value=FALSE)

    names(mosaic_list)[j]<-list_date_names[i]

    new_list[i]<-mosaic_list[j]

  }

  mosaic_list<-new_list

  out_rastnames<-paste(list_date_names,out_rastnames,sep="")

  #mean_m_list<-list.files(pattern='mosaiced_.*._lst_mean_VE_03182013.tif')

  mean_m_list<-mosaic_raster_list(mosaic_list,out_rastnames,out_path)

  #Use this as ref file for now?? Ok for the time being: this will need to change to be a processing tile.

  ref_rast<-raster(mean_m_list[[1]])

  #ref_rast <-raster("mosaiced_dec_lst_mean_VE_03182013.tif")

  #Modis shapefile tile is slighly shifted:

  # +proj=sinu +lon_0=0 +x_0=0 +y_0=0 +a=6370997 +b=6370997 +units=m +no_defs for ref_rast

  #"+proj=sinu +lon_0=0 +x_0=0 +y_0=0 +a=6371007.181 +b=6371007.181 +units=m +no_defs" ??

  #reassign proj from modis tile to raster? there is a 10m diff in semi-axes...(a and b)

  ##Write function to screen data values...

  #Screen LST for extreme values?

  #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)

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

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

  ##2) Crop and reproject Canopy height data

  #Make it a function?

  #canopy_rast<-raster("Simard_Pinto_3DGlobalVeg_JGR.tif")

  canopy_name<-file.path(in_path,infile_canheight)

  CANHEIGHT<-create_raster_region(canopy_name,ref_rast)

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

  #3) Creating elev, aspect, slope from STRM

  SRTM_reg<-create_raster_region(infile_elev,ref_rast)

  #Call a function to reproject the data in a local projection defined on the fly using the processing tile

  #extent...For the time being just use sinusoidal projection.

  ###calculate slope and aspect

  terrain_rast<-terrain(SRTM_reg, opt=c("slope","aspect"),unit="degrees", neighbors=8) #, filename=\u2019\u2019, ...)

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

  r1<-raster(terrain_rast,layer=pos)             #Select layer from stack

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

  r2<-raster(terrain_rast,layer=pos)             #Select layer from stack

  N<-cos(r1)

  E<-sin(r1)

  Nw<-sin(r2)*cos(r1)   #Adding a variable to the dataframe

  Ew<-sin(r2)*sin(r1)   #Adding variable to the dataframe.

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

  #4) LCC land cover

  oldpath<-getwd()

  setwd(lc_path)

  #lc_name<-"con_1km_class_1.tif"

  lc_list<-list.files(pattern="con_1km_class_.*.tif")

  #lc<-raster(file.path(lc_path,lc_names))

  lc_reg_list<-vector("list",length(lc_list))

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

    lc_name<-lc_list[[i]]

    lc_reg<-create_raster_region(lc_name,ref_rast)

    data_name<-paste("reg_",sub(".tif","",lc_name),"_",sep="") #can add more later...

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

    writeRaster(lc_reg, filename=file.path(out_path,raster_name),overwrite=TRUE)  

    lc_reg_list[[i]]<-file.path(out_path,raster_name)

  }

  setwd(out_path)

  lc_reg_list<-mixedsort(list.files(pattern=paste("^reg_con_1km_class_.*.",out_suffix,".tif",sep="")))

  lc_reg_s<-stack(lc_reg_list)

  #Now combine forest classes...in LC1 forest, LC2, LC3, LC4 and LC6-urban...??

  #create a local mask for the tile/processing region

  #LC12<-raster(paste("reg_con_1km_class_12_",out_suffix,".tif",sep="")) #this is open water

  LC12<-raster(lc_reg_s,layer=nlayers(lc_reg_s)) #this is open water

  LC_mask<-LC12

  LC_mask[LC_mask==100]<-NA

  LC_mask <- LC_mask > 100

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

  #5) DISTOC, distance to coast: Would be useful to have a distance to coast layer ready...

  #This does not work...clump needs igraph. I'll look into this later...for now I used IDRISI to clump pixels.

  #rc<-clump(LC12)

  #tab_freq<-freq(rc)

  #Modify at a later stage:

  #raster<-"DISTOC_VE_01292013.rst"  

  #ocean_rast<-raster(file.path(in_path,"lc12_tmp_grouped_rec.rst"))

  #ocean_rast[ocean_rast==0]<-NA

  #Distance calculated in a global layer??

  #distoc_reg<-distance(ocean_rast,doEdge=TRUE) #this is very slow: more than 35 min use GRASS instead??

  #load DISTOC produced from IDRISI: automate the process later

  distoc_reg<-raster(file.path(in_path,"DISTOC_VE_01292013.rst"))

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

  #6) X-Y coordinates and LAT-LONG: do not keep in memory?

  #ref_rast <-raster("mosaiced_dec_lst_mean_VE_03182013.tif")

  r1 <-ref_rast

  xy <-coordinates(r1)  #get x and y projected coordinates...

  CRS_interp<-proj4string(r1)

  xy_latlon<-project(xy, CRS_interp, inv=TRUE) # find lat long for projected coordinats (or pixels...)

  x <-init(r1,v="x")

  y <-init(r1,v="y")

  lon <-x

  lat <-lon

  lon <-setValues(lon,xy_latlon[,1]) #longitude for every pixel in the processing tile/region

  lat <-setValues(lat,xy_latlon[,2]) #latitude for every pixel in the processing tile/region

  rm(r1)

  #coord_s<-stack(x,y,lat,lon)

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

  ##Step 3: combine covariates in one stack for the next work flow stage

  r<-stack(x,y,lon,lat,N,E,Nw,Ew,SRTM_reg,terrain_rast,CANHEIGHT,distoc_reg)

  #rnames<-c("x","y","lon","lat","N","E","N_w","E_w","elev","slope","aspect","CANHEIGHT","DISTOC")

  s_raster<-r

  #Add landcover layers

  #lc_names<-c("LC1","LC2","LC3","LC4","LC5","LC6","LC7","LC8","LC9","LC10","LC11","LC12")

  s_raster<-addLayer(s_raster, lc_reg_s)

  lst_s<-stack(c(as.character(mean_m_list),as.character(nobs_m_list)))

  s_raster<-addLayer(s_raster, lst_s)

  #covar_names<-c(rnames,lc_names,lst_names)

  names(s_raster)<-covar_names

  #Write out stack of number of change 

  data_name<-paste("covariates_",out_region_name,"_",sep="")

  raster_name<-paste(data_name,var,"_",out_suffix,".tif", sep="")

  #writeRaster(s_raster, filename=raster_name,NAflag=-999,bylayer=FALSE,bandorder="BSQ",overwrite=TRUE)  #Writing the data in a raster file format...

  s_raster_m<-mask(s_raster,LC_mask,filename=raster_name,

                 overwrite=TRUE,NAflag=-999,bylayer=FALSE,bandorder="BSQ")

  #using bil format more efficient??

  return(raster_name)

}

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

################### END OF SCRIPT/FUNCTION #####################