The Development of WLRAS and Its Very Short-Term QPF Performance in Multiple Heavy Rainfall Events

Chih-Chien Tsai(1)* and Shu-Chih Yang(2), Chung-Yi Lin(3), Jia-Chyi Liou(3)
(1)Taiwan Typhoon and Flood Research Institute, National Applied Research Laboratories, Taipei, Taiwan
(2)Department of Atmospheric Sciences, National Central University, Taoyuan, Taiwan
(3)Taiwan Typhoon and Flood Research Institute, National Applied Research Laboratories, Taipei, Taiwan

This study develops a WRF-LETKF radar assimilation system (WLRAS) and investigates its very short-term QPF performance in multiple heavy rainfall events. These events include Typhoons Megi (2010), Nanmadol (2011), Nalgae (2011), Saola (2012), Soulik (2013), Kong-rey (2013) and a stationary front in 2012, all of which led to the closures of the scenic Suhua Highway along the east coast of Taiwan. We carry out 199 deterministic QPFs, each of which compares both runs with and without radar assimilation. Both runs are initialized with NCEP FNL data and the 3DVAR assimilation of GPS RO and GTS data. The radar assimilation run additionally undergoes the processes of ensemble perturbations, ensemble spin-up and three 30-min cycles of LETKF radar assimilation before the analysis mean is used to initialize the deterministic QPF. The radial velocity and reflectivity observations of CWB’s four operational S-band radars are assimilated, and the QPFs are verified with CWB’s rain gauge measurements.

The maxima, root-mean-square errors, biases and spatial correlation coefficients of 3-hour rainfall for the 199 QPFs compared with the rain gauge measurements are calculated for both runs with and without radar assimilation. All the four criteria show that the former outperforms the latter for the majority of the 199 QPFs; that is, both the intensity and spatial distribution of forecasted rainfall can mostly be improved by WLRAS. Moreover, in the event of Typhoon Saola (2012), we also engage a WRF-3DVAR system with a corresponding model setup. The results show that WLRAS provides more accurate QPFs than the 3DVAR system no matter from the rain maps or the ETS/Bias scores of 6-hour rainfall.


*email: cctsai@narlabs.org.tw
*Preference: Poster