Utilisation des données d'un radar météorologique pour la simulation des écoulements en rivière à l'aide d'un modèle hydrologique matriciel

Abstract

A good knowledge of the spatial and temporal distribution of precipitation is required for streamflow forecasting with a mathematical model. Thus, an operational method for correcting and integrating radar and station data with a minimum of operator intervention is under development. A total of 297 CAPPI maps at a height of 2 km have been used, each map representing a 2 h accumulation for 43 rainy days in the summer of 1978. Preliminary results have shown that it is necessary to detect and correct ground clutter as well as echoes coming from anomalous propagation. Thus, an algorithm has been developed to do so. In order to obtain a better idea on the necessity of correcting those echoes and on the efficiency of the algorithm, estimations of the mean areal daily rainfalls on the basin calculated from station data, are compared to estimations computed from (a) raw radar data, (b) radar data corrected with a ground clutter mask, and (c) radar data corrected with the developed algorithm, before and after calibration with station data. Similar comparisons between measured streamflows and streamflows computed by the CEQUEAU model, are presented, except for raw radar data. Actual results show that the identification and correction algorithm behaves very satisfactorily and could permit the operational use of radar data with hydrological models. Moreover, the sensitivity of the calibration factor on the number of meteorological stations is studied. It is shown that a relatively small number of stations may be used to calibrate the radar data. However, this number should be greater for convective than for continuous rain.