spiPRISM.R

# SPI/SPEI for monthly PRISM
# MAC 08/08/18

library(SPEI)
library(raster)
library(zoo)
library(rasterVis)

# set rasteroptions
rasterOptions(progress = 'text')
# map layers
states <- getData('GADM', country='United States', level=1)

# dates 1895-2017 PRISM data
dates=seq(as.Date("1895-01-01"), as.Date("2017-12-01"), by="month")

# load RESAMPLED data
prec<-stack("/scratch/crimmins/PRISM/monthly/processed/west/resampled/resampledWESTmonthlyPRISM_prec_1895_2017.grd")
prec<-prec[[which(dates=="1915-01-01"):which(dates=="2015-12-01")]] # breaks at 1918

# calculate SPI
funSPI <- function(x, scale=12, na.rm=TRUE,...) as.numeric((spi(x, scale=scale, na.rm=na.rm, ...))$fitted)
#rstSPEI <- calc(wtrBal, fun = funSPEI)
# parallell calc
ptm <- proc.time()
beginCluster(7)
    spiTemp <- clusterR(prec, calc, args=list(fun=funSPI))
endCluster()
proc.time() - ptm

# plot grids
# my.at <- seq(-3, 3, 0.5)
# mapTheme <- rasterTheme(region=brewer.pal(11,"Spectral"))
# levelplot(spiTemp[[1212]], par.settings = mapTheme, at=my.at, margin=FALSE, main="SPI", colorkey = list(space='right'))+
#   layer(sp.polygons(states))
# # check stats of dist fitting
# meanSPI<-calc(spiTemp, mean, na.rm=TRUE)
# sdSPI<-calc(spiTemp, sd, na.rm=TRUE) 

# write out files
writeRaster(spiTemp,filename="/scratch/crimmins/PRISM/monthly/processed/west/resampled/resampledWESTmonthlyPRISM_SPI12_1915_2015.grd", overwrite=TRUE )