## Process precip data for Nuuk

## Packages
library("tibble")
library("readr")
library("tidyr")
library("curl")

## Data - data are at ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2/v2.prcp.Z
## file is compressed
url <- "ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2/v2.prcp.Z"
tmp <- "./v2.prcp.Z"
curl_download(url, tmp, quiet = FALSE)

## At this point you'll need to uncompress the x-compress file you just downloaded,
## and extract the `v2.prcp` file to the working (or other) directory.
## There is an R package `uncompress` that could do this for you, but it was removed
## from CRAN at some point; the last version archived there is from 2012.
path.2.v2.prcp <- "./v2.prcp"

## data have 16 chars, then twelve groups of 5 chars
## the first 16 chars contain relevant info and need extracting into separate variables on reading
precip <- read_fwf(path.2.v2.prcp,
                   fwf_widths(c(3, 5, 3, 1, 4, rep(5, 12)),
                              col_names = c("Country","WMOID","Modifier","Duplicate","Year", month.abb)),
                   na = c("-9999","-8888")) # -8888 means a trace, which I ignore (naughty!)
                                            # -9999 is straight up missing

## subset only the bits we want
precip <- precip[precip$WMOID == 4250L, ] # 4250 is the WMO station ID for Nuuk (Godthaab)

## gather the month cols into a single key value pair
precip <- gather(precip, Month, Precip, -Country, -WMOID, -Modifier, -Duplicate, -Year)

## add a numeric column for month
months <- 1:12
names(months) <- month.abb
precip <- add_column(precip, nMonth = unname(months[precip$Month]), .after = "Month")

## Drop some other stuff
precip <- precip[, c("WMOID", "Year", "Month", "nMonth", "Precip")]

## Save this to RDS
write_rds(precip, "./nuuk-precip-data.rds")