Impact of tropical cyclone initialization on its convection development and intensity: A case study of Typhoon megi (2010)

This study investigates the impact of tropical cyclone (TC) initialization methods on TC intensity prediction under a framework coupling the Weather Research and Forecasting Model with the TC Centered-Local Ensemble Transform Kalman Filter (WRF TCC-LETKF). While the TC environments are constrained by assimilating the same environmental observations, two different initialization strategies, assimilating real dropsonde observations (the DP experiment) and synthetic axisymmetric surface wind structure (the VT experiment), are employed to construct the TC inner-core structure. These two experiments have distinct results on predicting the rapid intensification (RI) of Typhoon Megi (2010), which can be attributed to their different convective burst (CB) development. In DP, the assimilation of the dropsondes helps establish a realistic TC structure with asymmetry information, leading to scattered CB distribution and persistent RI with abundant moisture supply. In VT, assimilating the axisymmetric surface wind structure spins up the TC efficiently. However, the initially excessive CB coverage causes a too-early high-level warm core, and the reduced moisture supply hinders RI. The forecast results imply that if the TC structure is initialized using a scheme considering only the axisymmetric vortex structure, the RI potential can possibly be underestimated due to the inability to represent the realistic asymmetric structure. Finally, assimilation of both the real and synthetic data can be complementary, giving a strong TC initially that undergoes a longer RI period.