Assimilation of radar data in mesoscale models: Physical initialization and latent heat nudging

The increasing availability of quality controlled remotely sensed data (e.g . products from weather radar networks) as well as enhanced computer capaci ty allows an efficient use of these data in numerical weather prediction (N WP) models. In this paper two different assimilation techniques for radar m easurements in mesoscale models are presented: a physical initialization (P I), currently under development at the University of Bonn, and a latent hea t nudging (LHN) method implemented at the German Weather Service (DWD). Bot h algorithms are designed for the non-hydrostatic limited-area model LM (Lo kal-Modell) of DWD. Input data are standard DWD measurements: national rada r composites (reflectivities) and synoptical observations (temperature and dew point). Within the PI scheme the LM profiles of vertical wind, specific water vapor, and cloud water content are adjusted in such a way, that the model reproduces the radar derived precipitation. In the LHN scheme the pro gnostic variables temperature and specific humidity are changed. In PI and LHN runs with synthetic radar data the life-cycle of a single convective st orm was well represented. (C) 2000 Elsevier Science Ltd. All rights reserve d.