Impacts of GNSS Radio Occultation Data on Predictions of Two Super-Intense Typhoons with WRF Hybrid Variational-Ensemble Data Assimilation

The impact study of global navigation satellite system (GNSS) radio occultation (RO) data was investigated for two super-intense typhoon events, Typhoon Megi (2010) and Typhoon Haiyan (2013), both leading to significant disasters in several countries. The Weather Research and Forecasting (WRF) model with hybridvariational-ensemble data assimilation (DA) system was used to assimilate both FORMOSAT-3/COSMIC RO refractivity and global telecommunication system (GTS) observations. The background error covariance (BEC) in the hybrid DA system is contributed by three-dimensional variational (3DVAR) for static BEC, and ensemble adjustment Kalman filter (EAKF) for flow-dependent BEC. The model simulations show skillful track predictions on the two typhoons, and the assimilations with GNSS-RO data give better performances than the RO-data denied experiments, due to the reduction on the initial analysis errors. In addition, verifications against ERA-Interim show that the model simulations with assimilated RO data result in smaller biases. For both cases, better track predictions are obtained using the hybrid WRFDA than the WRF 3DVAR. A horizontal scale of 800 km for the covariance localization is found to give the best performance that significantly reduces the eastward bias in later track predictions of Typhoon Megi after departing from northwestern Philippines.