Assimilation of Doppler Weather Radar Data with a Regional WRF-3DVAR System: Influence of Data Assimilation Volume on Precipitation Forecast

In this study, the Weather Research & Forecast (WRF) model is adopted to provide mesoscale rainfall forecasts of four typical 24h duration rainstorms in North China and the DWR (Doppler Weather Radar) data is assimilated through the three-dimensional variational system (3DVar) in a cycling run mode to update the initial conditions of the WRF model. Global Telecommunication System (GTS) data produced by the National Center for Atmospheric Research (NCAR) are also assimilated as a benchmark. Before starting the assimilation process, the model spins up for 30h with the initial condition provided by the Global Forecast System. In the 1h WRF-3DVAR assimilation cycle, which means the DWR data are assimilated with a time interval of 1 h, the forecasts from the previous assimilation run serve as the background for the next run. A total of 24 runs in a cycle are performed for each 24h rainstorm. In order to highlight the contribution of the data assimilation volume, a 6h WRF-3DV AR assimilation cycle is also employed, which means, the DWR data are also assimilated with a time interval of 6 h. The results show that DWR data assimilation has improved the skill of high-intensity rainfall forecasts from the WRF model. The assimilation effect on the WRF model is not entirely decided by the amount of the assimilated data, but is closely related to the data quality.