P2.6 a Bayesian Approach to Retrieve Raindrop Size Distribution from Polarimetric Radar Data

The retrieval of raindrop size distribution (DSD) from polarimetric radar data (PRD) has the potential to improve the quantitative precipitation estimation (QPE), understanding precipitation microphysics and quantitative precipitation forecasts (QPF) (e.g., Brandes et al. 2004). Radar measurements and the retrieval model contain errors, which lead to errors in the retrieved DSDs. For example, if the retrieval model is based on the assumption of pure rain while the radar echoes sometimes includes snow, hail or non-hydrometeor effects, the retrieved DSD will be incorrect. In practice, however, it is difficult to know measurement and model errors. The Bayesian theory offers a promising method of optimizing the use of PRD for DSD retrievals. Using the prior information of measured DSDs, the Bayesian approach provides not only the mean values of DSD parameters, but also their standard deviations, which determine the reliability of retrievals. Two factors are required for the success of the Bayesian approach. The first one is the use of appropriate state parameters and conditional/forward model. The other one is the correct prior information of the rain.