Downscaling

anomalies.R

This script generates monthly climate anomalies geotiffs. anomalies.R uses the original annual geotiffs and the climatologies previously generated from these originals by climatologies.R. climatologies.R must be run before anomalies.R for any given data set.

This is done for each month and for the climate variables temperature and precipitation, and in some cases sea level pressure. Each mutli-band geotiff contains annual bands, the number of bands depending on the number of years in the original data set.

After anomalies are created, they may be downscaled to alternate climatologies of higher spatial resolution, e.g., PRISM 771-m, PRISM 2-km, or CRU 2.0 10-minute. Examples of downscaled data products include SNAP’s CMIP3 and CMIP5 global climate models and CRU 3.0, 3.1, and 3.22.

R code

library(raster)
library(parallel)

# Source functions
path.fun <- file.path("/workspace/UA/mfleonawicz/projects/Downscaling/code")
source(file.path(path.fun, "anomPrep.R"))
source(file.path(path.fun, "anomCalc.R"))
source(file.path(path.fun, "anomParSubFunc.R"))
anom.script.file <- file.path(path.fun, "anomCalcAsScript.R")  # source below

#### Choose datasets (current options: 'cru322' 'cru3101' 'cru31' 'cru30'
#### 'AR4gcm' 'AR5gcm') ####
dataset <- "cru322"

#### Set year ranges for reference climatology period and anomalies ####
yrs.clim <- c(1961, 1990)
yrs.anom <- c(1901, 2013)

#### Top level climatology and anomaly directories ####
dir.create(newDir <- paste("/Data/Base_Data/Climate/World/Anomalies/", yrs.clim[1], 
    "_", yrs.clim[2], "_Base_Climatology", sep = ""), showWarnings = F, recur = T)
climDir1 <- paste("/Data/Base_Data/Climate/World/Climatologies/", yrs.clim[1], 
    "_", yrs.clim[2], sep = "")

#### Set to TRUE is resampling outputs to a new grid and specify any raster
#### layer as the new grid ####
resample <- F
if (resample) r <- raster(file.path("/Data/Base_Data/Climate/World/ERA40/ERA40_historical_monthly_195709-200208.nc"), 
    varname = "t2m", layer = 1)

#### Prepare workspace. Specify senario name (options above) only if using a
#### projected climatology period ####
dp <- anomPrep(dataset, yrs.anom, yrs.clim, scenario = NULL)
for (i in 1:length(dp)) assign(names(dp)[i], dp[[i]])

#### Loop serially over variables, parallel processing anomalies by
#### models/datasets. Ideal for a collection of GCMs of similar file size. ####
for (k in 1:length(mainDir)) mclapply(1:length(files[[k]]), anomCalc, yr1 = yrs.anom[1], 
    yr2 = yrs.anom[2], var = varid[k], scen = scen[k], datID = datID, arID = arID, 
    pq = c(0.01, 0.99), tq = c(0.01, 0.99), setNAs = 0.5, interpNAs = T, pcdf = 0, 
    pr0maps = T)

#### Use below only for faster parallel processing by month of a single
#### model/dataset at a time. Ideal for CRU or a collection of GCMs one at a
#### time which differ substantially in file size. ####
yr1 = yrs.anom[1]
yr2 = yrs.anom[2]
scens <- scen
pq = c(0.01, 0.99)
tq = c(0.01, 0.99)
setNAs = 0.5
interpNAs = T
pcdf = 0
pr0maps = T
parallelizeMonths <- T
for (k in 1:length(mainDir)) {
    var = varid[k]
    scen = scens[k]
    for (i in 1:length(files[[k]])) {
        source(anom.script.file)
        print(paste("Completed model ", i, " of ", length(files[[k]]), " for variable-scenario combination ", 
            k, " of ", length(mainDir), sep = ""), quote = F)
        gc()
    }
}