Data assimilation for very short-range forecasting in COSMO

Christoph Schraff

DWD, Offenbach, Germany

This contribution will summarise the status and developments related to the data assimilation in the COSMO consortium with regard to very short-range forecasting.
The current operational system for the COSMO model is based on an observation-nudging type technique. For convection-permitting model configurations, it includes a latent heat nudging scheme for the assimilation of radar-derived precipitation rates at full temporal and horizontal resolution. The positive impact on precipitation is very large at the beginning and decreases rapidly during the first 6 hours of the forecast.
To replace the nudging scheme, a novel ensemble data assimilation system for convection-permitting NWP using mesh widths of typically less than 3 km is being developed in the framework of a COSMO Priority Project called KENDA (Km-scale ENsemble-based Data Assimilation). The main objective of this system is to provide suitable perturbed initial conditions for an ensemble prediction system, and to use a situation-dependent description of the background errors for combining short-range forecast information with new observations. The development focuses currently on the LETKF (Local Ensemble Transform Kalman Filter) for model resolutions of 2 – 3 km. A special focus will be on the use of 3-dimensional radar reflectivity and radial wind, GNSS slant path delay data, and satellite-derived cloud information which are all not used in the current nudging scheme. Since the development of the LETKF system is still in a rather early phase, results obtained so far are only preliminary. The presentation will therefore focus on an overview of the current and planned development.