Data Assimilation Seminar
Prof. Shu-Chih Yang (September 10, 2019, 16:00-)
|Affiliation||National Central University, Taiwan|
|Title||Recent developments and challenges of the regional ensemble data assimilation system for high-impact weather prediction in Taiwan|
|Abstract||The high-impact weather prediction in Taiwan is very challenging due to the nature of multi-scale interactions and the complex topography. A regional data assimilation system that combines high-resolution numerical model and high-spatiotemporal resolution observations become the key to improve the high-impact weather in Taiwan. In this talk, I will cover our recent developments for predicting two types of high-impact weather events, heavy precipitation and air-pollution.
The WRF-Local Ensemble Transform Kalman Filter Radar data assimilation system (WLRAS) has been developed as the main component of convective-scale ensemble data assimilation and for very short-term precipitation prediction in Taiwan. Although this system has demonstrated very useful forecast skill in terms of quantitative precipitation nowcasting, we are still facing many challenges with the limitations of the radar observations and EnKF methodology. Recently, we have made several improvements for WLRAS, including (1) a strategy to deal with sampling errors when assimilating the radar radial velocity, (2) additional assimilation of fast moisture information provided by the ground-based GNSS zenith total delay and (3) implementation of the dual-localization method to consider multi-scale interactions. All these developments show significant improvement for heavy rainfall prediction.
The WRF coupled with the Community Multiscale Air Quality Model (CMAQ) has been applied to air-pollution (PM2.5) prediction. Improving the accuracy of the model initial condition and the description of emission can both improve the air-pollution prediction. Taking the first route, the WRF-LETKF system assimilates the fine particulate matter profile retrieved from Micro-Pulse lidar. The lidar retrieved PM2.5 profile is able to reflect the dynamic and thermodynamic structures of the planetary boundary layer (PBL). Our results show that assimilation of the lidar data can correct the near-surface overestimated wind speed and further improve the vertical structure and transportation of PM2.5 concentration during the 12-h forecast. As a conclusion, lidar data assimilation is potential for improving the air-pollution prediction.