Page History

...

1. Explore the archive
   The first step is to find out if the data you are looking for like to retrieve is actually archived in MARS. To do that, explore the MARS content by directly browsing the MARS archive catalogue. Alternatively, browse the dataset descriptions on the main ECMWF website MARS Catalogue (authorisation required) or the  MARS content overview  which in turn contains contain links to the archive catalogueMARS Catalogue. Once you found the right data is foundfields, you can use Web-MARS to "Check for availability", to "Estimate download size" or to retrieve some sample data "Retrieve the selection", just a few fields, to verify that this is the data you expect to get. In the MARS catalogue Catalogue use Ctrl+Left-Mouse-Button to select/unselect multiple parameterslist items or Shift+Left-Mouse-Button to select a range. If now you like to retrieve larger data volumes (for more dates than you can get with one request in Web-MARS), to make the data retrieve part of a script etc. you need to create a MARS request and pass is to the MARS clientcommand.
   
   

2. Create a MARS request
   Once you found the right fields, let Web-MARS create a MARS request as a starting point . This can be done by clicking on "View the MARS request" on the last page of the archive tree. To turn Copy and paste this initial request from Web-MARS into a complete request you just need to file and add a line to specify your target file, e.g. target="myfile". Now this request should Your saved request file can now be used to retrieve the data by calling the mars command:

   Code Block
   % mars myrequestfile

   If you have access to ECMWF computers, you can call mars e.g. on ecgate. However, all MARS requests are preferably done in a batch job, see below. Web users with access to MARS (users without access to ECMWF computers) need to install and use the mars command on their local system to Access MARS via the Web API.
   
   

3. Retrieve the data in a batch job
   As MARS requests retrievals can use considerable resources, the general recommendation is to retrieve data via serial batch jobs on the ecgate server. A full ecgate batch job which retrieves fields from MARS, applies some computation and transfers the resulting data to a remote site, is provided with mars_ectrans.ksh. Once all this works fine, you might want to expand your script, e.g. to loop over several requests, write to multiple target files etc. For all modifications please consider the implications these might have reg. the resources like disk space and CPU time.

For further guidance please refer to other the more detailed pages in this user guide.