Vortex Initialization through the High-Resolution Ensemble Kalman Filter Framework and Its Impact on Intensity Forecast: A Case Study of Typhoon Megi (2010)

Although tropical cyclone (TC) track prediction has been improved in the past decades, the intensity prediction remains challenging, especially for rapid intensification (RI) cases. Vortex initialization based on ensemble data assimilation (EDA) has the advantages of dynamical consistency and makes good use of observations, potentially giving more realistic initial conditions and better intensity forecasts.

In this study, initialization strategies based on EDA are applied to the case of Typhoon Megi (2010) under the framework of Weather Research and Forecasting model-TC Centered Local Ensemble Transform Kalman Filter (WRF-TCCLETKF) to explore the impact of assimilating inner-core observations, including dropsondes (DP) and the axisymmetric surface wind structure (VT). The former is from the Impact of Typhoons on the Ocean in Pacific (ITOP) field campaign and the latter follows Wu et al. (2010). The initialization time of the DA period was one or two days before the onset of RI (0000UTC 16 October, 2010) and the deterministic forecast was carried out. Results show that the earlier the forecast, the better the intensity prediction. The VT analysis leads to a TC intensity close to the observation; while the intensity in the DP analysis is much weaker. However, the TCs in the forecasts initialized from the DP analyses rapidly intensify with larger convective bursts (CBs) areas. By contrast, the axisymmetric surface wind structure can spin up the TCs effectively with a stronger TC circulation and a well-developed warm core; however, their TCs intensify slowly during the forecast period with less strong and penetrative convections. In conclusion, assimilating dropsonde better represents the characteristics of the TC structure, such as asymmetry, and leads to the process of RI.