Access MARS

Introduction

This Service will allow authorised users to retrieve and list MARS data from outside the ECMWF facilities. Users within ECMWF Member and Co-operating States may contact their Computing Representative to obtain access to MARS. All other users may request a username and password, following these instructions.

In this page you will find the step-by-step guide to install the ECMWF Web API client and create your own MARS Requests. We strongly recommend you that you get familiar with MARS.

Information

For those used to the MARS system, it is worth noting that the client that you will install in your system is a Web API service interface that forwards the request to our systems and expects a single output file. It has no knowledge of internal ECMWF libraries such as MIR or ecCodes, these operations are performed in our internal servers. Therefore, not all the functionality of the full MARS client is available, like multi-target, compute, etc...

This method is only supported on UNIX systems. It requires Python.

Step-by-step guide

Steps:

Download mars
Install ECMWF API client library
Install ECMWF KEY
Test installation
Check availability
MARS examples
Brief MARS request syntax
Documentation

(Optional) To follow this guide, we recommend you to have a directory in your $HOME space called "mars".

mkdir -p ~/mars/bin
mkdir -p ~/mars/lib

Download Mars

Download the mars python script: mars. You have to give execution permission to the file and put it in a location in your PATH.

If the file has been downloaded to ~/mars/bin, you have to execute:

Mars

chmod +x ~/mars/bin/mars
export PATH=~/mars/bin/:$PATH

Install ECMWF API client library

The client currently supports both Python 2.7.x and Python 3.

The ecmwf-api-client library was added to the Python Package Index (PYPI):

https://pypi.org/project/ecmwf-api-client/

Install via pip with:

pip install ecmwf-api-client

If you do not have admin privileges you can always install the client in your user space:

pip install --user ecmwf-api-client

Alternatively, if you have Anaconda installed you can install the Web API Python client library doing:

conda install -c conda-forge ecmwf-api-client

If you have Anaconda installed you can install the Web API Python client library doing:

conda install -c conda-forge ecmwf-api-client

Alternatively, you can first install Python (and pip) on your Windows environment.

Download windows installer exe from Python.org download page
Run the exe.
Screen will be shown to chose the installation option.
Uncheck "install for all user" option.
Go for the custom installation.
On next screen specify the directory path for which your user have full access on the computer (take a note of this as you will need it to set the environment variables).
Uncheck "create shortcuts for installed application" option.
Make sure "Add python to environment variable" option is Unchecked .
Complete the installation.

Then add the installation and Script folder path in PATH using set (temporary) or setx (permanent) in a Command Prompt window. This is to make sure that your computer knows where to find the Python interpreter. To do this you will have to modify a setting called PATH, which is a list of directories where Windows will look for programs. Further details and examples are available here. You may choose also to set your environment variables into a batch file (e.g. set-env.bat which you will need to run from the Command Prompt.

@echo off

set PY_HOME=C:\...\python37

set PATH=%PY_HOME%;%PY_HOME%\Scripts;%PATH%

You can now install it via pip (see instructions above).

For languages other than Python please see Web-API Downloads

Install ECMWF KEY

If you don't have an ECMWF account, please self register at https://api.ecmwf.int/v1/key ( click login and then register ) and then follow ONE of the steps below.

You can also try visiting https://api.ecmwf.int/v1/key/ to retrieve your credentials.

Note that the key expires in 1 year. You will receive an email to the registered email address 1 month before the expiration date with the renewal instructions.

Test installation

Example script

Retrieval of sea surface temperature for the first 10 days of May 2002, all synoptic times. It retrieves 40 fields.

retrieve,
  class   = od,
  stream  = oper,
  expver  = 1,
  date    = 20020501/to/20020510,
  time    = 00/06/12/18,
  type    = an,
  levtype = sfc,
  param   = sea surface temperature,
  target  = "sst.grib"

Save this script as test.req and execute:

mars test.req

The command should generate a sst.grib file containing the retrieved 40 fields in GRIB format.

You can use ECMWFService Python class to execute a retrieve command:

server = ECMWFService("mars")
server.execute({ "python":"query" }, "target")
# Alternatively the function execute can also accept a text variable with the request
server.execute(req,target)

Example:

#!/usr/bin/env python
from ecmwfapi import ECMWFService
 
server = ECMWFService("mars")
server.execute(
    {
    "class": "od",
    "date": "20150101",
    "expver": "1",
    "levtype": "sfc",
    "param": "167.128",
    "step": "0/to/240/by/12",
    "stream": "oper",
    "time": "00",
    "type": "fc"
    },
    "target.grib")

Check availability

To check the Data availability you should have Access to the MARS Catalogue. Please note that the MARS catalogue is a super-set of the Archive catalogue.

Archive Catalogue

MARS Catalogue (restricted to authorised users)

you will view all the experiments that are available from our archive. We strongly recommend you to to navigate through MARS Catalogue to become familiar with the availability of our archive. You may select a class and start navigate trough the contents.

Please take into account some considerations about the content:

Different class and stream include different parameters, times, and steps
In each class not all the parameters are available from all steps
In each class and stream not all the steps are available from all times

The web-application above will help you to check and understand the availability. For any kind of selections the system will update the attributes in a dynamic way to reflect the current availability. (i.e. if you change the steps some parameters will be added or removed).

Tip

We encourage users to use the " View MARS request " feature once the selection has been done.

You can also use list action with output=cost keyword to see the available data and its cost without need to retrieve it. This is useful to check the amount of data and number of tapes that your request will need to access. If your request takes longer than expected, it is a good practice to check its cost.

list,
  class   = od,
  stream  = oper,
  expver  = 1,
  date    = 20020501/to/20020510,
  time    = 00/06/12/18,
  type    = an,
  levtype = sfc,
  param   = sea surface temperature,
  output  = cost,
  target  = "sst.list"

#!/usr/bin/env python

from ecmwfapi import ECMWFService
 
server = ECMWFService("mars")
req = '''
list,
  class   = od,
  stream  = oper,
  expver  = 1,
  date    = 20020501/to/20020510,
  time    = 00/06/12/18,
  type    = an,
  levtype = sfc,
  param   = sea surface temperature,
  output  = cost
'''
server.execute(req, "sst.list")

This would produce something like:

size=21599920;
number_of_fields=40;
online_size=21599920;
off_line_size=0;
number_of_tape_files=0;
number_of_disk_files=1;
number_of_online_fields=40;
number_of_offline_fields=0;
number_of_tapes=0;

Indicating that the size of this request will be 20.59 Mb (output is in Kb) and that it is fully located in the online disk (no need to access any tape).

Bear in mind that the maximum value of number_of_tapes recommended is 1 tape per request and should not exceed 3. Otherwise we strongly encourage you to split it in several different requests iterating by date.

If you do not set the "output=cost" keyword on the list action, the output will show you information about the type of requests and the entries:

sst.list

class   = od
expver  = 1
file[0] = hpss:/mars/prod/od/o/oper/an/sfc/marsodoper/1/an/20020501/sfc/126665.20031108.113424
id      = 126665
levtype = sfc
month   = 200205
stream  = oper
type    = an
year    = 2002
date       file length missing offset     param  time     
2002-05-01 0    539998 .       4729126    34.128 00:00:00 
2002-05-01 0    539998 .       37233582   34.128 06:00:00 
2002-05-01 0    539998 .       69738038   34.128 12:00:00 
2002-05-01 0    539998 .       102242494  34.128 18:00:00 
2002-05-02 0    539998 .       134746950  34.128 00:00:00 
2002-05-02 0    539998 .       167251406  34.128 06:00:00 
2002-05-02 0    539998 .       199755862  34.128 12:00:00 
2002-05-02 0    539998 .       232260318  34.128 18:00:00 
2002-05-03 0    539998 .       264764774  34.128 00:00:00 
2002-05-03 0    539998 .       297269230  34.128 06:00:00 
2002-05-03 0    539998 .       329773686  34.128 12:00:00 
2002-05-03 0    539998 .       362278142  34.128 18:00:00 
2002-05-04 0    539998 .       394782598  34.128 00:00:00 
2002-05-04 0    539998 .       427287054  34.128 06:00:00 
2002-05-04 0    539998 .       459791510  34.128 12:00:00 
2002-05-04 0    539998 .       492295966  34.128 18:00:00 
2002-05-05 0    539998 .       524800422  34.128 00:00:00 
2002-05-05 0    539998 .       557304878  34.128 06:00:00 
2002-05-05 0    539998 .       589809334  34.128 12:00:00 
2002-05-05 0    539998 .       622313790  34.128 18:00:00 
2002-05-06 0    539998 .       654818246  34.128 00:00:00 
2002-05-06 0    539998 .       687322702  34.128 06:00:00 
2002-05-06 0    539998 .       719827158  34.128 12:00:00 
2002-05-06 0    539998 .       752331614  34.128 18:00:00 
2002-05-07 0    539998 .       784836070  34.128 00:00:00 
2002-05-07 0    539998 .       817340526  34.128 06:00:00 
2002-05-07 0    539998 .       849844982  34.128 12:00:00 
2002-05-07 0    539998 .       882349438  34.128 18:00:00 
2002-05-08 0    539998 .       914853894  34.128 00:00:00 
2002-05-08 0    539998 .       947358350  34.128 06:00:00 
2002-05-08 0    539998 .       979862806  34.128 12:00:00 
2002-05-08 0    539998 .       1012367262 34.128 18:00:00 
2002-05-09 0    539998 .       1044871718 34.128 00:00:00 
2002-05-09 0    539998 .       1077376174 34.128 06:00:00 
2002-05-09 0    539998 .       1109880630 34.128 12:00:00 
2002-05-09 0    539998 .       1142385086 34.128 18:00:00 
2002-05-10 0    539998 .       1174889542 34.128 00:00:00 
2002-05-10 0    539998 .       1207393998 34.128 06:00:00 
2002-05-10 0    539998 .       1239898454 34.128 12:00:00 
2002-05-10 0    539998 .       1272402910 34.128 18:00:00 

Grand Total:
============

Entries       : 40
Total         : 21,599,920 (20.5993 Mbytes)

Spend some time to understand how it works, make some selections try to execute the MARS scripts.

MARS examples

40 year re-analysis

Retrieval of snow depth from the ERA-40 archive for November 1993, for all analysis base times. It retrieves 120 fields. See in MARS Catalogue

retrieve,
  class   = e4,
  stream  = oper,
  expver  = 1,
  date    = 19931101/to/19931130,
  time    = 00/06/12/18,
  type    = an,
  levtype = sfc,
  param   = sd,
  repres  = gg,
  target  = "era40.199311.sd"

Operational Analysis

Retrieval of sea surface temperature for the first 10 days of May 2002, all synoptic times. It retrieves 40 fields. See in MARS Catalogue

retrieve,
  class   = od,
  stream  = oper,
  expver  = 1,
  date    = 20020501/to/20020510,
  time    = 00/06/12/18,
  type    = an,
  levtype = sfc,
  param   = sea surface temperature,
  target  = "sst"

Ensemble Forecast

Retrieval of surface temperature an 10m wind components (U and V), 20 first members of the EPS for 2nd January 2001 for time-steps 12, 36 and 60. It retrieves 180 fields. See in MARS Catalogue

retrieve,
  class   = od,
  stream  = enfo,
  expver  = 1,
  date    = 20010102,
  time    = 12,
  step    = 12/36/60,
  type    = pf,
  levtype = sfc,
  param   = st/10u/10v,
  number  = 1/to/20,
  target  = "perturbed.sfc"

Snow depth 1 month

Retrieval of snow depth from the ERA-Interim archive for December 2007, for all analysis base times. It retrieves 124 fields. See in MARS Catalogue

retrieve,
class   = ei,
stream  = oper,
expver  = 1,
date    = 20071201/to/20071231
time    = 00/06/12/18,
type    = an,
levtype = sfc,
param   = sd,
target  = "era-int.200712.sd"

You can find more advanced examples here: MARS example requests

We strongly suggest users to split their requests by date e.g. one month at a time.

If you want to get big chunks of data, please have a look to the Retrieval efficiency page.

Brief MARS request syntax

See Brief MARS request syntax page.

Documentation

Training material (slides, pdf): MARS introduction and basic concepts

Mars manual

MARS FAQ

Web API FAQ