In addition to our data assimilation and forecasting work at the scale of a radar, we also validate numerical precipitation forecasts at larger scales using radar composites in order to understand WHY do models have difficulties in predicting precipitation, and HOW could these difficulties be overcome.

For this purpose, several projects have been and are carried out, focusing on evaluating the ability of NWP to forecast the diurnal cycle of precipitation, on the importance of horizontal resolution for forecast quality, and on the advantages of ensemble prediction and radar data assimilation. The conclusions of these studies are then communicated to the modeling community as guidelines for choosing future research directions.