[A-AS04] Advances in Tropical Cyclone Research: Past, Present, and Future

[AAS04-08] High-resolution 1000-member Ensemble Simulations of Typhoon Hagibis (2019)

*Pin-Ying Wu¹, Takuya Kawabata², Le Duc³ (1.Japan Meteorological Business Support Center, 2.Meteorological Research Institute, 3.The University of Tokyo)

Keywords:ensemble simulation, probability prediction

Ensemble simulations are widely used for dealing with uncertainty in weather forecasts, but the estimates using ensemble simulations would suffer from sampling errors with an insufficient ensemble size. Recently, benefiting from the improvement of computing power, conducting large-size ensemble simulations has become possible. This study investigated the feasibility of flow-dependent probability predictions of strong winds using 1000-member ensembles during the strike of typhoons. The 1000-member ensemble simulations of typhoon Hagibis (2019) are performed using the non-hydrostatic model (NHM) of the Japan Meteorological Agency (JMA) with 5- and 1-km horizontal grid spacing. The results suggest the benefits of producing probability forecasts of winds using ensemble simulations, which provide additional information to the warning of strong winds announced by JMA. In addition, high probabilities of strong winds are estimated along rivers and basins with the 1-km ensemble, while the same feature is not seen with the 5-km one. To make best use of the ensemble on estimating strong wind probability prediction over the land, we further re-centered the 1-km ensemble to a higher resolution simulation with 200-m grid spacing, which provides a better agreement of wind speed with observations. Beside the strong wind predictions, we also explored the non-Gaussian distribution of the wind fields and the impact of boundary conditions for constructing ensemble simulations.