Impact of radar assimilation on forecasts of a heavy rainfall case
Wenxue Tong1, Juanzhen Sun2, Gang Li1, Hongli Wang3
1Nanjing University of Information Science and Technology, Nanjing, China
2National Center for Atmospheric Research, Boulder, Colorado
3National Oceanic and Atmospheric Administration, Boulder, Colorado
The impact of radar assimilation on quantitative precipitation forecast (QPF) of a heavy precipitation case in a mesoscale convective system was investigated in this study. The heavy rain, which occurred during 8 and 9 August 2008 over Colorado and Kansas, also caused flash flood over south Denver and its nearby suburbs. A new 1-hourly assimilation-forecast cycle strategy was implemented with WRF and its 3DVAR system. This new strategy is composed of two independent techniques: two-step assimilation and Frequent Restart Rapid Cycle (FRRC) configuration. The two-step assimilation technique is used to incorporate radar and synoptic-scale observations, and the FRRC configuration was introduced to suppress the forecast noise common in convective-scale simulation.
Experiments with different configurations were performed, including rapidly cycled 3DVAR with and without radar data assimilation. The comparison of different forecasts showed that the new strategy of 1-hourly cycled assimilation with radar data is able to improve QPF up to 12 h, and successfully captures the development of the convective system, also improves the prediction of the location and intensity of the flash flood embedded in the mesoscale convective system. The diagnosis of the analysis fields shows that two-step assimilation is able to incorporate synoptic-scale information from GTS observations as well as convective-scale features from radar observations in the 3DVAR analysis. The FRRC scheme is effective in suppressing the noise produced by the model integration. The use of FRRC leads to a much lower false alarm rate.