Improving very short range prediction of high-impact weather using radar observations

Studies have shown that high-impact weather systems, especially severe storms, often contain fine-scale, rapidly-changing dynamical and microphysical structures. Failure to characterize the fine-scale features in the model initial fields will significantly reduce the model's capability and accuracy in forecasting these severe weather events. In this study, we examine how to improve the initial fields and thus enhance the capability of a mesoscale NWP model in very short range predictions by assimilating high-resolution, three-dimensional radial-wind and reflectivity observations from Doppler radars in addition to conventional data. A variational approach is used to assimilate radar observations with rapid-update-cycle (RUC) procedures, so the model fields can be updated hourly to better capture changes in the storm fields. The method has been successfully tested with several types of storms, including squall lines and a landfalling hurricane. Detailed results from our recent experiments will be presented to show the impact of radar data on very short range predictions of storms and to discuss several important issues regarding radar data assimilation at high resolution.