You are here

Acquiring CMIP5 data from the World Data Center for Climate

This page is part of the project: Great Lakes Climate Ensemble

CMIP5 Data Listing

The most extensive archive of CMIP5 data that I have found available online is at the World Data Center for Climate portal.  Use the search filters to narrow the options.  Make sure "IPCC-AR5_CMIP5" is selected for the Project name.  The "Topic Name" filter allows you to narrow which variables are shown.  The "Keywords" filter can be used to find specific models and/or experiments (i.e., historical run and RCP runs).  Use these filters to narrow down the data options to those you need to download.  For example, select Project name: IPCC-AR5_CMIP5, Keywords: GFDL-CM3 and historical and one result should be available.  Click on the result "cmip5 output1 NOAA-GFDL GFDL-CM3 historical" and to search for specific variables, click on the magnifying glass at the bottom right (in the dataset_group section).  Then you can use the filters to find specific variables and the Aggregation filter to find the daily or monthly datasets.  Using our GFDL example, once narrowed by daily data for the precipitation_flux variable, five results should be available.  These are five different model runs, denoted by r1.... r2.... r3.... etc in the result names.  For our purposes we've been only using the first available run (run 1 typically).  If you click the result for run 1, you will see a dataset "Entry acronym" which is also the same as the dataset name.  This name corresponds to this particular dataset which you will need to know when calling the Jblob command to download the dataset.  

Setting up Jblob

You do not need an account to browse data, but you will need one to download the data.  Instructions for how to get an account can be found here

To be able to download data via the terminal (as opposed to point and click methods), you must download their Jblob program.  Jblob will give additional download options, like subsetting to a smaller region at the time of the download so the files do not have to be as large.  There are many other uses as well.

Step 1: Take note of where you put the Jblob-2.0.10 folder (or whichever version you have).  Mine is located at: Documents/programs/Jblob-2.0.10

Step 2: Edit Jblob-2.0.10/jblob to set the JBLOB_HOME path. I prefer to use terminal commands:

$ cd /Users/auraell/Documents/programs/Jblob-2.0.10
$ vi jblob
$ i

"i" allows you to insert text.  Use the arrow keys to navigate to the JBLOB_HOME line and edit the path to match where it exists on your machine. To save the file hit the "escape" key, then   :wq!

Step 3: You will need to copy the jblob.rc.example to your home directory, rename it to .netrc, and edit the top line:

$ cd $home
$ cp path/to/jblob.rc.example .netrc
$ vi .netrc

Edit the line and replace <myusername> and <mypassword> with your account information.  Remove the carrots.  Then save the file as in the steps above for saving.

Step 4: edit/create the .bash_profile to tell your machine where Jblob and Java exist:

$ vi .bash_profile
$ i

add the following 2 lines then save the file.

export JAVA_HOME=$(/usr/libexec/java_home)
export PATH=/path/to/Jblob-2.0.10:$PATH

"source" the file to allow changes to go into effect:

$ source .bash_profile

Test to make sure jblob is working.  

$ jblob --help

The following should be displayed:

jblob V2.0.10, 04/04/2011

USAGE: jblob --dataset | -d <acronym> [ options ]

       jblob --help

       jblob --version

where options include:

    --config   <filename>   Config file (username + password)

    --username <username>   Database username

    --password <password>   Database password

    --rmin     <rec_number> First record to transfer

    --rmax     <rec_number> Last record to transfer

    --dir      <directory>  Target directory

    --file     <filename>   Target filename

A complete listing of arguments/options is available at

Downloading Data via Jblob

See examples of ways to download or "precut" and download data.

For our purposes, we are downloading data for the region 20W to 170W and 10N to 80N.  In the terminal enter (do not include brackets):

jblob --dataset [dataset acroynm] --cdo "sellonlatbox,-170,-20,10,80" --dir [download directory]

[dataset acronym] can be found by navigating to the level of individual files in the data portal.  For example, BCB1hiDADclt111v1 is the dataset acroynm for cmip5 output1 BCC bcc-csm1-1 historical day atmos day r1i1p1 v1 clt listed in the table of entries for BCB1hi.