Elisabeth Bauernschubert, German Weather Service, DWD

Assimilation of Radial Winds in an Ensemble Kalman Filter on the Convective Scale


With the Kilometer Scale Ensemble Data Assimilation System KENDA the German Weather Service (DWD) and the Consortium for Small Scale Modeling (COSMO) have developed an Ensemble Kalman Filter (EnKF) for Data Assimilation on the convective scale (cf. Schraff et al, QJRMS 2016). The system has been successfully tested in an operational setup to provide deterministic forecasts as well as initial states for the operational convective-scale ensemble prediction system COSMO-DE-EPS (cf. Keane et al, MZ 2016). It is currently under preparation for operational use.

The operational setup of the KENDA system consists of the ensemble data assimilation (EDA) system for conventional observations over central Europe in combination with the assimilation of RADAR data based on latent heat nudging (LHN, Stephan et al 2012). This hybrid approach has proven to be an efficient method to assimilate radar composits into the system.

The new weather radar network of the DWD includes 17 dual-polarimetric C-Band Doppler radars distributed throughout Germany. The radars offer unique 3-dimensional information about dynamical and microphysical characteristics of precipitating clouds in high spatial and temporal resolutions. The assimilation of radar reflectivities within the KENDA system has been successfully tested by Bick et al, QJRMS 2016. The assimilation is based on the EMVORADO 3D-RADAR forward operator developed by Zeng and Blahak (Zeng et al 2013).

We will present ongoing work on the assimilation of 3D-RADAR radial winds by the ensemble Kalman filter of the KENDA system, based on the EMVORADO forward operator of Zeng and Blahak (see above). Tests with different types of localization, superobbing/resolution, observation errors and quality control have been carried out. We will discuss the impact of the radial wind data on the precipitation forecasts of the COSMO model, in particular for higher thresholds. Challenges and next steps towards operational use of the data will be presented.

File available here