Ensemble data assimilation using passive and active microwave observations of precipitation in mountainous regions

The Goddard WRF ensemble data assimilation system has been developed to assimilate precipitation information into WRF model to improve QPF and QPE at high resolution. The flow-dependent forecast error covariance estimated in the assimilation procedure aims to capture the large temporal and spatial variability of precipitation and clouds. The microphysics at cloud-resolving scales and all-sky radiative transfer simulator serve as non-linear observation operators to link observables with model states. We present results of assimilating precipitation-affected microwave radiance and precipitation radar reflectivity from a pre-GPM constellation overland in the southeast US region. Observational bias correction for all-sky radiance is developed based on innovation statistics and a situation-dependent bias estimation model. The data impact is assessed with independent ground-based precipitation observations and evaluated in applications to dynamical downscaling and hydrological prediction.