Public data server documentation

WARNING (6/5/2004)

The Met Office data for the coupled model hindcast (GloSea) started in February 1968 were corrupted. The data have been corrected and users are requested to make a new retrieval to substitute their previous files.

WARNING (6/9/2006)

Some of the Met Office data (total precipitation and total soil moisture) for the coupled model hindcast (GloSea) started in November 1992 were corrupted. The data have been corrected and users are requested to make a new retrieval to substitute their previous files.

WARNING (10/12/2006)

The Met Office maximum temperature data are the same as the minimum temperature data for all start dates. The data will be corrected soon and a new notice will appear here to request users to make a new retrieval to substitute their previous files.

Note that you can now retrieve DAILY and MONTHLY data. Check the third menu on the left column to select the type you are interested in. Below there is an brief description of the different parameters used in the site for DEMETER data retrieval:

• Experiment: several coupled models have been run to produce the hindcasts
• Starting date: initial date of the hindcast; each hindcast is initialised on the first day of the month at 00 GMT
• Parameters: pressure level and surface variables are made available; soil temperature level 1 contains surface temperature over land and sea surface temperature over the oceans
• Forecast month (for monthly mean data): each hindcast is run for six months
• Forecast day (for daily data): each hindcast has 180 days available
• Ensemble member: each ensemble contains nine members

VERY IMPORTANT !!!

Take into account the following pieces of advice before retrieving the data:

• The DEMETER data distribution system is part of the ECMWF public data server. Thus, the same rules as to the general public data distribution apply.
• Files in both GRIB and NetCDF formats are available. The NetCDF format follows ECMWF conventions, and it might change with time till a final consensus and complete compatibility with CF conventions are reached.
• Don't forget to check the units of the variables.
• The variable "Soil level temperature 1" contains the sea surface temperature (obtained from the atmosphere model) over the oceans and the temperature of the first soil layer over the continents.
• Daily fluxes (precipitation and radiation) are daily accumulations from 00 GMT of the day indicated until 24 hours later.
• The monthly mean ERA-40 data are distributed for the purpose of verification in a similar format to the model hindcasts, though they are analyses (except for precipitation, which is calculated from 12 to 36 hours 24-hour forecasts). The structure of the retrieval has been made the same as for the monthly mean hindcast data. The GRIB headers (different from the ones appearing in the ERA-40 section of the public data server) had to be changed accordingly, so that they look as if the ERA-40 data were seasonal forecasts, although they obviously are not.
• The fields for the different DEMETER hindcasts as well as for the corresponding ERA-40 data are global, starting at 87.5°N and going to 87.5°S. They have been interpolated into a 2.5° regular grid.
• Retrievals should be quite fast. You can customize the areas you are interested in and avoid transferring files too big through the line. For technical reasons the maximum number of fields that can be retrieved in one go has been limited to 20,000.
• Remember that you can plot any data, though the server is not intended to be a plotting system. Try to use it to just check that the variables you want to retrieve look as you expect. When plotting, you can select different options. Use the "geography", "coastlines" and "contour" tools that appear after your selection has been made.