27A Ensemble-based Assimilation of Radial Velocity of a Coastal Doppler Radar in China for Typhoon Vicente (2012) near Peak-intensity Stage

Monday, 3 August 2015
Back Bay Ballroom (Sheraton Boston )
Lei Zhu, Peking University, Beijing, China; and Z. Meng

This work explores the impact of a WRF-based ensemble Kalman filter (EnKF) by continuously assimilating ground-based Doppler radar data in south China on the prediction of a high-impact landfalling TC Vicente (2012) near peak intensity stage. After more than ten years' development, the Doppler radar network in China has become mature. However, its potential contribution to analyses and forecasts of landfalling TCs have not been examined using ensemble-based data assimilation method. Besides, the performance of EnKF during the peak-intensity stage of a TC still remains an unknown question worldwide.

Vicente was an intense western North Pacific TC that made landfall around 2000 UTC 23 July 2012 near the Pearl River Delta region of Guangdong Province, China with a peak 10-m wind speed of 44 m s-1 along with considerable inland flooding after a rapid intensification process. Results showed that the WRF-EnKF efficiently assimilated coastal radar radial velocity and apparently improved the depiction of the inner-core structure of Vicente which further improved the forecasts of its track and intensity, and especially the associated inland heavy precipitation of various thresholds. The EnKF forecast successfully captured the inland intense precipitation and the spiral rain bands. Despite some overestimation of the most intense rainfall, the 24-h accumulated rainfall forecast initialized from the EnKF analyses was much improved over no data assimilation experiment (NoDA) after only four volumes of radial velocity being assimilated. The 1-h precipitation forecast also compared favorably to the observed rainfall in terms of both location and intensity than NoDA. Besides, this work also demonstrated the capability of the EnKF in effectively ingesting radar data near the peak stage of a TC, which complemented previous works on assimilating radar data when TCs are not very strong.

Sensitivity analyses were further performed to explore the leading dynamics that controlled the prediction and predictability of TC track, intensity and rainfall during and after TC landfall based on ensemble forecasts started from EnKF analyses. Correlation analyses showed that TC's intensity and precipitation were closely tied to TC track features. A more northward track was associated with a more northward initial location and a weaker intensity. TC forecast latitude was more correlated with its initial position than initial intensity. On the other hand, a stronger forecast intensity was associated with a more southward initial location and a stronger intensity. TC forecast intensity was more correlated with its initial intensity than initial position.

Relative to previous works that mostly focused on improvement in TC track and intensity forecast, this work demonstrated the possible contribution of assimilating coastal land-based radar data to the TC-associated inland rainfall forecast. It could be also helpful for the development of an operational EnKF system in China to make full use of the newly-matured Doppler radar network.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner