You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 4 Next »

Page under construction - please ignore content

MARS is ECMWF's Meteorological Archival and Retrieval System. The system has the following features:

  • Facilities to archive and retrieve environmental data
  • 24/7 service
  • Batch and interactive modes
  • Large amount of data, both in size and number of items stored
  • Large number of users with different requirements.

Access to the MARS archive requires being an ECMWF registered user. To request an account, please send an email to efas@ecmwf.int.

A full documentation of the MARS archive can be found here.

You can access through the ECMWF Web API. For support and how to use this this service, please see documentation here. It is strongly recommended that you familiarize yourself with the use of MARS coding and the data available. The Copernicus Emergency Management Service EFAS data in MARS can be browsed through here: https://apps.ecmwf.int/mars-catalogue/. For a short tutorial and guide to the data, please see the step-by-step guide below

Decoding the grib file

Data in MARS are stored as grib files. Grib files cannot be read directly and need first to be decoded using ECMWF's free software ecCodes . Please see the general help page for more information on how to read and use the grib files. Note that you have to use ecCodes 2.8.0 or later versions the read EFAS data.

The how to read EFAS data pages provide more detailed information, useful links and tutorials.

A step by step guide to accessing data in the MARS archive

The MARS data catalogue and associated data files can be browsed through the web page https://apps.ecmwf.int/mars-catalogue/, which also lists the available CEMS-EFAS data.

Navigation is done by clicking on links on each pages.


EFAS data is stored under the special datasets 'Copernicus Emergency Management Service (CEMS).

Within the Copernicus Emergency Management Service (CEMS) page, EFAS data is organised in two 'streams':

  • Climatology, which corresponds to historical hydrological simulation forced with observation. It contains historical simulation forced with the using the same model and configuration as the operational system. Currently the data is available from 1991 to 2018-12-11
  • Operational model, which corresponds to EFAS operational medium-range hydrological forecasts time series starting from 2018-10-10

Each stream is accessible by clicking on the associated link; for example here 'operational model'

In MARS, data are coded as 'stream' and 'class'. After selected the type of data of interest, the associated code is highlighted in bold under the “Current selection” section.  The selection can also be modified clicking on an alternative code.

Within the EFAS operational model section, two versions are available:

  • Version 1, which is the operational model version; i.e. it corresponds to the EFAS forecasts that have been published
  • Version 9001, which is the forecasts produced by the test system


Data published within version 1 always correspond to the latest operational forecasts and should be the default choice. Only very special cases would require accessing data from the version 9001/ test system, and only version 1 is recommended to use.


EFAS operational model version 1 page contains different types of data:

  • Observations: this contains the hydrological simulations forced with observations
  • Assimilations: this contains the data used to bridge the gap between latest observations received and the time when forecasts starts (fill Up); it is generated using the most recent high resolution meteorological forecast as forcing
  • Forecasts: this contains the deterministic EFAS forecasts runs
  • Ensemble: this contains the ensemble EFAS forecast runs. For simluations forced by ECMWF ensemble forecasts, they are organised as control forecast (single member forced by ECMWF-ENS control forecast input) and all other ensemble simulations (perturbed forecast)

More details on the type of data can be found in the GloFAS data terminology page.



Once selecting a data type (here Forecast), the list of years with available data is shown, then when selecting a specific year, the list of months with available data is shown.

Note the type corresponding to 'Forecast' is fc (shown in bold).

Once a month has been selected (here March), the list of 'parameters' (or variable type) available is shown.

EFAS available variables are organised in two categories:

  • Surface variables, corresponding to variables simulated on a single level, the land surface. For EFAS they include river discharge and snow water equivalent
  • Soil levels, corresponding to variables simulated on different levels. For EFAS, data is available for three soil levels

Once the type of level has been selected (here soil levels), the list of meteorological forecast models used to produce the forecast (or 'origin') is shown.

In EFAS, simulations from two deterministic meteorological forecast models are available: DWD and ECMWF

Once the 'origin' has been selected (here ECMWF), the different attributes of the available data can be selected. They include:

  • Date of forecast: day of the year when the forecast has been released
  • Time of forecast: time of the day of the forecast released
  • Forecast lead time: number of hours between the start of the forecast and the forecasted value
  • Level: level code number, 1 starting at the top (Fredrik Wetterhall  to check)

After selecting the attributes (no selection will obtain the entire dataset), the data can be downloaded directly by clicking on 'Retrieve the selection in GRIB' or through a MARS request as code by clicking 'View the MARS request'.

Note that retrieving EFAS data in netCDF is not possible directly from MARS.


The "View the MARS request" page shows all MARS codes corresponding to the selection. Similar codes can be used for automatic retrievals/ customised requests using scripting languages.

For more information on how to use the MARS scripting to customize your retrievals please refer directly to the MARS user guide.

To get access through the ECMWF Web API, please see documentation here.

  • No labels