R/calcGridPop.R

In pik-piam/mrplayground: MadRat playground

#' calcGridPop
#' 
#' @description Past and future (SSP1-5) population based on HYDE3.2 and Jones & O'Neill (2016)
#' Data is scaled to match WDI data from calcPopulation
#' NOTE that some scaling factors for the future (for small countries...Gambia and Djibouti) are somewhat off, data read in is 50% of WDI data, most likely due to large resolution
#' @param subtype past (1965-2005), future (2005-2010) or all (divergence changed to start at 2015)
#' @param cellular only cellular
#' @param FiveYear TRUE for 5 year time steps, otherwise yearly from source
#' @param harmonize_until 2005 default divergence of SSPs
#' @return Population in millions.
#' @author David Chen
#' @importFrom magclass add_columns collapseNames
#' @importFrom magpiesets findset
#' @importFrom madrat calcOutput toolGetMapping toolAggregate


calcGridPop <- function(subtype="all", cellular=TRUE,FiveYear=TRUE, harmonize_until=2015) {
  if(!cellular)(stop("Run calcPopulation instead"))
  ##past 
  if (subtype=="past"){
    
  gridpop <- readSource("GridPop",subtype="past",convert=F)
  CountryToCell     <- toolGetMapping("CountryToCellMapping.csv", type = "cell")
  agg   <- toolAggregate(gridpop, rel=CountryToCell, from="celliso", to="iso", partrel=TRUE)
  
  ## scale to match WDI country-level pop 
  
  pop <- calcOutput("Population",aggregate=F)
 pop <- time_interpolate(pop, interpolated_year=getYears(agg))
  
  #scaling factor sc_f applied to every cell
  sc_f <- (agg/1e6)/pop[getRegions(agg),,"pop_SSP2"]
  
  gridpop1 <- gridpop/sc_f[getRegions(gridpop),,]
  
  #mauritius and ATF have 0, but the cells exist, so get filled. 
  #the single MUS cell gets the pop, the ATF cells get even division of pop
  gridpop1["MUS",,] <- 1e6*pop["MUS",,"pop_SSP2"]
  gridpop1["ATF",,] <- 1e6*pop["ATF",,"pop_SSP2"]/length(getCells(gridpop["ATF",,]))
  gridpop1["SGS",,] <- 1e6*pop["SGS",,"pop_SSP2"]/length(getCells(gridpop["SGS",,]))
  x <- collapseNames(gridpop1,collapsedim=2)
  x <- add_columns(x, dim=3.1,addnm=c(getNames(pop)[1:5]))
  x[,,2:6] <- x[,,1]
  x <- (x[,,-1])/1e6
  }
  
  ##future
  if (subtype=="future"){
   
    gridpop <- readSource("GridPop",subtype="future",convert=F)
    CountryToCell     <- toolGetMapping("CountryToCellMapping.csv", type = "cell")
    agg   <- toolAggregate(gridpop, rel=CountryToCell, from="celliso", to="iso", partrel=TRUE)
    
    ## scale to match madrat country-level pop 
    pop <- calcOutput("Population",aggregate=F)[,,1:5]
    pop <- time_interpolate(pop, interpolated_year=getYears(agg))
    
    #scaling factor sc_f applied to every cell
    sc_f <- (agg/1e6)/pop[getRegions(agg),,]
    gridpop1 <- gridpop/sc_f[getRegions(gridpop),,]
    
    #ATF, SJM, SGS have 0, but the cells exist, so get filled. 
    #the few cells for each island get even division of total pop
    for (i in c("SGS","ATF","SJM")){
      gridpop1[i,,] <- 1e6*pop[i,,]/length(getCells(gridpop[i,,]))
    }
    
    x <- gridpop1/1e6
  }
  
if (subtype=="all"){
    past <- calcOutput("GridPop",subtype="past", aggregate=F, FiveYear=F)
    future <- calcOutput("GridPop", subtype="future", aggregate=F, FiveYear=F)
    #harmonize future SSPs to divergence year
    
    harm_y <- getYears(future, as.integer = T)[1:(harmonize_until-2009)]
    for (i in 1:5){
         future[,harm_y,i] <- future[,harm_y,"pop_SSP2"]} 
    x <- mbind(past,future)
    x <- time_interpolate(x, interpolated_year = 1965:2100)
    }

  
if (FiveYear==TRUE){
  years <- findset("time")
  x <- x[,intersect(years,getYears(x)),]
}
  
  return(list(x=x,
              weight=NULL,
              unit="million",
              description="Population in millions",
              isocountries=FALSE))
}