Tools and datasets

In PANDOWAE several partners work with given tools or datasets. By the use of operational models and datasets there is an intensive collaboration with operational centres.

COSMO: The Consortium for Small-Scale Modelling (COSMO) model is a non-hydrostatic limited area model based on the Lokal-Modell (LM) (Steppeler et al. 2003) originally developed at the DWD and run operationally at DWD with 7 km horizontal resolution (COSMO-EU, formerly LME) and 2.8 km horizontal resolution (COSMO-DE, formerly LMK). A number of other European weather services have contributed to the development of the COSMO system. The COSMO model is installed at several partner institutes (Karlsruhe, Mainz, DLR) and has been used for several years with horizontal resolutions ranging from 40 km to 1 km.

COSMO-LEPS: In the COSMO-LEPS system (Molteni et al. 2001; Tibaldi et al. 2003) the COSMO model is nested in 10-15 members of the ECMWF ensemble prediction system (EPS) that are selected using a clustering algorithm.

GME: The GME (Global Model Europe) is the operational global weather forecast model of the German Weather Service (Majewski et al. 2002). It is hydrostatic and based on an icosahedral grid.

Different products of the ECMWF:

ECMWF operational analyses: The current model version T799L91 and the 4d-var data assimilation scheme provide global analyses every 3 hours with a horizontal resolution of approximately 25 km.

ECMWF IFS Deterministic forecasts: The same model version is used for twice daily 10-day global forecasts, started at 00 and 12 UTC.

ECMWF IFS EPS: the operational ECMWF ensemble prediction system is run at a T399L62 with a control forecast and 51 perturbed members. The initial perturbations are calculated using singular vectors (Molteni et al. 1996; Leutbecher et al. 2007) and a stochastic physics scheme is included (Buizza et al. 1999).

ECMWF IFS Data assimilation system: ECMWF currently uses a 12 hour window four-dimensional variational (4D-Var) data assimilation system, where the full model resolution (T799) is used for comparing observation and model equivalents and a reduced resolution (T255/T95) for their minimization. For research applications, the ECMWF system can be applied to reproduce the analysis of any given date with a large number of options: incorporating additional observations, ignoring particular observation types or observations in particular regions, reducing the resolution and many others changes.

ECMWF ERA40 und ERAinterim: The ERA40 data set was produced with ECMWF model version T159L60 (corresponding to a horizontal resolution of about 100 km) and the 3d-var data assimilation technique for the time period 1958-2002. It provides a reasonably consistent meteorological data set for this extended period, allowing climatological investigations of various weather phenomena. Currently, a new reanalysis data set is produced, for a shorter time period of about 1980 till today, based upon model version T255L60 and the 4d-var data assimilation technique.

TIGGE: Nine operational Numerical Weather Prediction centres have agreed to deliver a selection of data from global ensemble forecasts in near real time to the THORPEX Interactive Grand Global Ensemble data base. CMA, ECMWF and NCAR have agreed to become Archive and Distribution centres (Bougeault, 2006). TIGGE data will be available for research purposes with a delay of 48 h after the initial time of the forecast.

ECHAM: The GCM ECHAM is a global climate model which was developed at the Max-Planck-Institut für Meteorologie in Hamburg and Universität Hamburg based on the ECMWF forecast model.

MM5: The fifth-generation Pennsylvania State University - National Center for Atmospheric Research (PSU-NCAR) Mesoscale Model. This is a non hydrostatic mesoscale circulation model involving an Arakawa-Lamb B-staggered grid on sigma surfaces.

Piecewise PV inversion: Potential vorticity inversion code (Davis and Emanuel 1991) has been provided by Dr. Christopher Davis and used for previous studies in Karlsruhe and Mainz. This code will be modified so it can be applied to the COSMO model output.

EOF/Cluster analysis: This methodology uses a combination of an empirical orthogonal function analysis and a fuzzy clustering applied to the principal components (Harr et al. 2007). Thus ensemble members with similar contributions to the variability patterns are identified through the EOF analysis.