How to download CORDEX data from the ESGF


Summary: It’s not easy and you need the software wget.

The way the ESGF works is to generate a wget script based on input you give. So you select which domain, which variable, which scenario, which regional model, what time step etc then ESGF generates a wget script that you then run to do the actual data retrieval.

You need to have the software wget and I only know how to make this work in a linux environment so if you are in a windows environment you will need to google wget for windows and install it.


Here are the steps I’d recommend to download data.


There are a few to choose from: Sweden, Germany or UK linked below are my preference.  or or

You will need to create an account. This is done by clicking the Create Account link in the top right. This will provide you with an OpenId (like a username) and a password you need to execute the wget script at the end of the process.

Once you have an account, Login, and you can continue.


2. Which CORDEX do you want

On ESGF is data for CORDEX “Phase 1” (0.44 degree data, which has by far the largest GCM-RCM downscaled data available), CORDEX-Adjust if you want bias-corrected data (smaller GCM-RCM dataset). In the Search block enter your choice and then click the “Search” button. I will use “CORDEX” as the example

Now you have narrowed the search to CORDEX data, you  will see this in the Search Constraints just under the Search button:

Search Constraints: CORDEX


3. Select the Africa domain….

Click on “Domain”  in the special CORDEX box on the left and tick AFR-44 (not AFR-44i).

If you want the CORDEX-CORE data (the new 0.22 degree downscaling, but a very limited GCM-RCM matrix – 3 RCMs downscaled 4 GCMs) select AFR-22.

Then click the search button.

Now you will see just under the Search button

Search Constraints:   CORDEX | AFR-44


4. Now select the variable… I would say if you are wanting multiple variables download one variable at a time because you can only download 1000 files at a time in a wget script and sometimes there are more than this number of files. So you can think you are downloading all the files but then some will be missing if there are over 1000.

You can select variables in one of 2 ways – I will use precipitation as an example:

1 – If you are familiar with the abbreviations used for variables on ESGF you can select the “Variable” tab and select “pr” from the list

2 – If you are not familiar with the variable conventions on ESGF you can select the “Variable Long Name” tab and select “Precipitation” from the list

Using (1) as an example, tick “pr” and click the “Search” button again and now you will see you have selected

Search Constraints: CORDEX|AFR-44|pr

Some other popular variables:

tas = average surface temperature, tasmin = average surface minimum temperature,
tasmax = average surface maximum temperature, pr = precipitation)


5. Now choose your experiment…
Click “Experiment” on the left menu and you will see:
Evaluation – CORDEX RCMS downscaled ERA-Interim data
historical – CORDEX RCMS downscaled GCM 1950-2005
rcp26, rcp 45, rcp85 – CORDEX RCMS downscaled GCM 2006-2100 for each RCP

In my view you always need the evaluation period to evaluate the bias in each RCM. Always download the evaluation run.

But I will use rcp85 as an example…so tick rcp85 and click the “Search” button again and now you will see you have selected.

Search Constraints:   CORDEX | AFR-44 | pr | rcp85


6. Now you can select the time frequency you are interested in, daily or monthly data
Click “Time Frequency” on the left menu and select daily or monthly, I’d suggest monthly until you are confident in using ESGF because the data files are smaller. Remember you can only download 1000 files at a time in a wget script and sometimes there are more than this number of files, the excess of which will not be downloaded.

CORDEX-CORE also has 3- and 6-hourly data, there are lots of files if you select this option so me sure of what you are trying to do before you download these.
Using monthly data – tick “mon” and Search again to give

Search Constraints:   CORDEX | AFR-44 | pr | rcp85 | mon


7. Next choose your Ensemble

We generally use r1i1p1 so click “Ensemble” and tick “r1i1p1” then Search to give

Search Constraints:   CORDEX | AFR-44 | pr | rcp85 | mon | r1i1p1


8. Then in the CORDEX box choose your “Downscaling Realization” on the left menu and select “v1” then Search to give

Search Constraints:   CORDEX | AFR-44 | pr | rcp85 | mon | r1i1p1 | v1


9. Now in the CORDEX box click “RCM Model” to select the RCMs you want the data from. I would do them one at a time because the you can only download 1000 files at a time in a wget script and sometimes there are more than this number of files. This is also why I suggest downloading 1 variable at a time.

For e.g. the RCA4 model, click on “RCM Model” in the CORDEX boc and select a model then click Search to give

Search Constraints:   CORDEX | AFR-44 | pr | rcp85 | mon | r1i1p1 | v1 | RCA4



To find out which GCMs each RCM has downloaded select a RCM from the “RCM Model”, then “Search”. Once this is done click “Driving Model” just above the CORDEX box and you will see which GCMs are downscaled. E.g. Select “RCA4” from the “RCM Model” tab and “Search”, then select  “Driving Model” and you will see 9 GCMS here.

To look at other RCMS, e.g. CCLM, go up to the “Search Constraints” filter and delete RCA4 by clicking on the red “X”, wait for the results and then select “CCLM” from the “RCM Model” tab and repeat.

RCA4 has downscaled 9 GCMs, CCLM4 downscaled 3 GCMs, REMO 4  and RACMO 1

If you want to have RCMs that downscaled the same GCM you can select the GCM in the “Driving Model” part of the left menu before you do step 9.


10. Making the wget script

There are two options again (1) make a series of wget script for each of the results you see in the results window or (2) make a wget script that will download all the files in your search window:

(1) Simply click on the link [WGET Script] for each line of your search return page and you will be prompted to save a wget script for each search return result. This downloads the wget script that you will need to execute (see 11 below). The “Show Data” and “Metadata” links are useful to see the files you will be downloading and also the metadata associated with the files. If you have followed my example there will be 9 lines so 9 links.

(2) Above the list of results you will see a link “Add all displayed results to Data Cart“. Clicking this will add all your search results to your “My Data cart”. Once this is done you will see in your “Data Cart” on the top right the results have been added. Click on the “My Data Cart” link to be taken to your Data Cart page.

Click the “Select All Databases” tab and then the [WGET Script] link – this then generates a box that says “For better performance, WGET scripts are generated for each Data Center separately. Click on each link below to retrieve the script for each Data Center.” and a link to download the wget script e.g. “WGET Script for” click the link and download the wget script.

Save the wget scripts to where you will be downloading the data.


11. Execute the wget script (See here for lots more info…

Then you need to make the wget script executable, so on the command line do
chmod 744
Then execute the script:
./ -H

You will be asked to enter your ESGF OpenId and Password

NOTE: This is how I do it in linux, I don’t know how to do wget in windows, you will need to find out if this is your working environment.

If you don’t know about wget someone will have to help you

Hope this helps.