Evaluation of a simplified dynamical rainfall forecasting model from rain events simulated using a meteorological model

This paper presents an evaluation of a simplified dynamical model from simu lated rain events. This model is based on conservation of the rainwater con tent of an atmospheric column and uses multi-scan radar data and surface me teorological observations. A test of the model using radar data from Oklaho ma City (USA) and the Cevennes region (France) has not clearly shown the mo del efficiency compared to the simple extrapolation of radar observations. This statement can be explained by the limited number of rain events availa ble for testing the model and errors inherent in radar measurements. The pr oposed evaluation is based on a mono-dimensional microphysical and meteorol ogical model, which serves to simulate reference rainfall. The outputs of t he meteorological model: rainwater content, cloud top temperature, surface temperature, pressure and dew point are used for initialising the simplifie d dynamical model as in the case of observed rain events. The evaluation of the simplified dynamical model is performed by comparing forecasts to refe rence rainrates. This comparison shows that the simplified dynamical model performs better than the simple extrapolation method for short lead-times u p to 30 min. This time is related to the response time of the rainwater atm ospheric column. For longer lead-times, the dynamics of the simplified mode l tends to a steady state characterised by a constant rainfall rate control led by the source term. This result which has still to be confirmed makes t he model adapted to requirements of urban hydrology. (C) 1999 Elsevier Scie nce Ltd. All rights reserved.