netCDF test with R for a job interview (1st part)

netCDF_test.r

library(curl)
url = "ftp://.../tas_201711.nc"
curl_download(url = url, destfile = "/tmp/tas_201711.nc")


# Write a simple R script that 
## 1. reads the netcdf file

as.mon <- function(posixlt){
  posixlt <- unclass(posixlt)
  paste(posixlt[["year"]] + 1900, posixlt[["mon"]] + 1, sep = "-")
}



library(RNetCDF)
netcdfData <- open.nc(con = "/tmp/tas_201711.nc")
temp <- var.get.nc(netcdfData, variable = "tas")
time_coord2 <- var.get.nc(netcdfData, variable = "time")
time_unit2 <- att.get.nc(netcdfData, "time", "units")
time_posixct2 <- utcal.nc(time_unit2, time_coord2, "c")
time_posixlt2 <- as.POSIXlt(time_posixct2)
time_month <- sapply(time_posixlt2, as.mon)


# or


library(ncdf4)
nc_data <- nc_open("/tmp/tas_201711.nc")
library(ncdf4.helpers)
tas_time <- nc.get.time.series(nc_data, v = "tas", time.dim.name = "time")
temp <- ncvar_get(nc_data, "tas")
library(PCICt)
time_posixlt2 <- as.POSIXlt.PCICt(tas_time)
time_month <- sapply(time_posixlt2, as.mon)

## 2. computes the monthly mean of the temperature

library(parallel)
cl <- makeCluster(detectCores() - 2)
mean_temp <- parApply(cl, temp, c(1, 2), by, time_month, mean)



# Could you create a random R array object with the function array() with 4 dimensions “latitude”, “longitude”, “time”, and “member”, and create an R function to compute the ensemble mean? 
# The function should work with random order of the dimensions (i.e., “member” can be either of the 1st/2nd/3rd/4th dimension). 
# If possible, including some initial checks of the inputs in the function.

netarray <- array(rnorm(2*3*5*7), dim = c(2,3,5,7), dimnames = list("latitude" = NULL, "longitude" = NULL, "time" = NULL, "member" = NULL))

ensemble_mean <- function(arry, ensemble){
  ix <- which(ensemble == names(attr(arry, "dimnames")))
  apply(arry, setdiff(1:4, ix), mean)
}

ensemble_mean(netarray, "member")