In mdsumner/CMIP5files:

knitr::opts_chunk$set(echo = TRUE)

R gives extremely powerful access to database facilities, and also general computational tools for accessing model output. Here we take a quick look at data from CMIP5 for ACCESS1-3.

run on raijin

Open the file listing database and have a look around. We can use the dplyr verbs filter and select to easily scan through a table.

## this was mentioned in discussions about what we were after: 
##/g/data/ua6/authoritative/IPCC/CMIP5/CSIRO-BOM/ACCESS1-3/rcp85/mon/seaIce/OImon/r1i1p1/files/sic_20120413/sic_OImon_ACCESS1-3_rcp85_r1i1p1_200601-210012.nc

library(dplyr)
db <- src_sqlite("/g/data1/ua6/unofficial-ESG-replica/tmp/tree/cmip5_raijin_latest.db")
cmip5 <- tbl(db, "cmip5")
cmip5 %>% filter(model == "ACCESS1-3") %>% select(-id)

Take a closer filter down into the model, ensemble, experiment etc.

 cmip5 %>% filter(model == "ACCESS1-3", variable == "sic", ensemble == "r1i1p1", experiment == "rcp85", mip == "OImon") %>% select(-id)

From here we'll go for the right matching file, and divorce from the db.

files <- cmip5 %>% filter(model == "ACCESS1-3", variable == "sic", ensemble == "r1i1p1", experiment == "rcp85", mip == "OImon", version == "v20121203") %>% 
     collect()

## seems to be one-file per id, sometimes - do better matching in time
f <- list.files(files$id[1], full.names = TRUE)[1]

print(f)

Tool-up and have a look.

library(angstroms)  ## not really about roms, but it works here too
library(rancid)

ncdb <- NetCDF(f)

ncdb$variable

It's "sic" that we are after.

coords <- romscoords(f, c("lon", "lat"))
library(tidyraster)
library(rbgm)
b <- bgmfile(system.file("extdata","bgm",  "antarctica_28.bgm", package = "bgmfiles"))
map <- romsmap(boxSpatial(b), coords)

mdsumner/CMIP5files documentation built on May 22, 2019, 4:25 p.m.