1D+3DVar assimilation of radar reflectivity data: a proof of concept

An original one-dimensional (1-D) retrieval followed by a three-dimensional variational (1D+3DVar) assimilation technique is being developed to assimilate volumes of radar reflectivity data in the high-resolution numerical weather prediction Arome model. The good performance of the 1-D retrieval is shown for an isolated storm over southwestern France through an observing system simulation experiment. The full method is applied with real data to a flash-flood event, which occurred in a mountainous area. For this complex case, the assimilation of reflectivity data improves short-term precipitation forecasts. The assimilation of reflectivity data has a positive impact on the convective system's dynamics by feeding the cold pool under the storm, which controls the intensity and location of the updrafts. A one-hourly update cycle of 3 h further improves these results. A sensitivity study is also presented to evaluate the assimilation method for this flash-flood event in different conditions. The smoothing coefficient involved in the 1-D retrieval is shown to have a very small impact on analyses and quantitative precipitation forecasts. The assimilation of reflectivity data is found to be able to cause the creation of a cold pool, which modifies favourably the precipitation quantitative forecast. Finally, results from an 8-d-long assimilation cycle are presented.