Changes between Version 78 and Version 79 of udg/ecoms/dataserver/catalog

Oct 27, 2015 7:43:23 PM (6 years ago)



  • udg/ecoms/dataserver/catalog

    v78 v79  
    4747'''Data Homogeneization:''' The different nature of the datasets, and the idiosyncratic naming and storage conventions often applied by the modelling centres, makes necessary an homogenization across datasets in order to implement a truly user-friendly toolbox for data access.  To this aim, the [wiki:RPackage R package for data access] has been developed. Data homogenization is achieved through the creation of a common vocabulary. The particular variables of each dataset are then translated -and transformed if necessary- into the common vocabulary by means of a ''dictionary''. Both features -vocabulary and dictionary- are described [wiki:RPackage/homogeneization here]. In particular, some typical transformations performed by the `loadECOMS` interface are deaccumulation of initialization-accumulated variables to daily accumulated (i.e.: '''DAr''' --> '''DA''') and scaling and/or offset of variables to match standard units (e.g. -273.15 for conversion K --> ºC).
     49= DATASETS =
     51== System4 (provided by ECMWF) ==
     52The [ System 4] seasonal forecasting system became operational in November 2011. The corresponding hindcast is archived in the Meteorological Archival and Retrieval System ([ MARS]), the main data repository at the ECMWF, as a colection of GRIB-1 files at 0.75º spatial resolution. The downloaded data has been exposed as three different virtual datasets  (see the [wiki:../listofvariables available variables] for these datasets):
     53* '''[ System4_seasonal_15]''': There are twelve initializations (hereafter called `runtimes`) per year (the first of January, February, ...), each with 15 members running for 7 months (hereafter called simply `times`). Period: 1981-2010.
     54* '''[ System4_seasonal_51]''': There are only four `runtimes` per year (the first of  February, May, August and November), each with 51 members running for 7 months. Period: 1981-2010.
     55* '''[ System4_annual_15]''': There are four `runtimes` per year each with 15 members, but the forecasts run for 13 months. Period: 1981-2010.
     57A preliminary [ validation report] produced in SPECS (milestone MS22) is available for precipitation (System4_seasonal_15). The reports for all datasets and variables will be produced after feedback with end-users.
     59== CFSv2 (provided by NCEP) ==
     60The [ CFS version 2] seasonal forecasting model became operational at NCEP in March 2011. The corresponding [ retrospective CFSv2] forecast dataset is stored in the [ NOMADS server] as a collection of GRIB-2 files at 1º spatial resolution.  The downloaded data is exposed as a single virtual dataset  (see the [wiki:../listofvariables available variables] for this dataset):
     61* '''[ CFSv2_seasonal]'''. There are four initializations (4 cycles) from every 5th day (thus providing on average 24 members per month) running for 9 moths (see [wiki:./CFSv2 CFSv2 members] for more detailed information of members' construction for this dataset). Period: 1982-2010. '''Note:''' For better comparability with other hindcasts, the [wiki:RPackage R data access package] defines by default an ensemble of 15 members for each lead month and forecast season.
     63== WATCH/WFDEI ==
     64The WATCH-Forcing-Data-ERA-Interim: [ WFDEI] was produced post-WATCH using WFD methodology applied to ERA-Interim data. It is a meteorological forcing dataset extending into early 21st C (1979 – 2012). Eight meteorological variables at 3-hourly time steps, and as daily averages, for the global land surface at 0.5º x 0.5º resolution.
     66 * '''[ WFDEI]''' Daily pseudo-observations
     68See the [ UDG catalog] for additional datasets provided by the UDG.