Download monthly-mean TMP2m for all ensemble members

This data is available on the NERSC web portal.

There is one 3.1Mb file for each of 80 ensemble menbers for each of 206 years (1806-2015). About 50Gb of data in total, and it’s kept on tape at NERSC; so it will take some time to download (several hours at best).

#!/usr/bin/env python

# Download the 20CRv3 monthly-mean T2m.

import os
import subprocess
import tarfile


local_dir="%s/20CR/version_3/monthly_means" % os.getenv('SCRATCH')

for year in range(1806,2016):
    tf = "%s/%04d/" % (local_dir,year,year)
    if os.path.isfile(tf): continue
    remote = "%s/TMP2m_%04d_mnmean.tar" % (base_url,year)
    local = "%s/TMP2m_%04d_mnmean.tar" % (local_dir,year)
    if not os.path.exists(os.path.dirname(local)):
    if not os.path.isfile(local):
        cmd="wget -O %s %s" % (local,remote),shell=True)
        if wg_retvalue!=0:
            raise Exception("Failed to retrieve data")
    tf =
    # Update the modification times on the file or SCRATCH
    #  will just delete them all again
    ncfs=os.listdir("%s/%04d" % (local_dir,year))
    for ncf in ncfs:
        os.utime("%s/%04d/%s" % (local_dir,year,ncf), None)