Assessment of an Experimental Version of fvGFS for TC Genesis Forecasting Ability in the Western North Pacific

Abstract We evaluated the ability of the fvGFS with a 13-km resolution in simulating tropical cyclone genesis (TCG) by conducting hindcast experiments for 42 TCG events over 2018–2019 in the Western North Pacific (WNP). We observed an improved hit rate with a lead time of between 5 and 4 days; however, from 4 to 3 days lead time, no consistent improvement in the temporal and spatial errors of TCG was obtained. More Fail cases occurred when and where a low-level easterly background flow prevailed: from mid-August to September 2018 and after October 2019 and mainly in the eastern WNP. In Hit cases, 850-hPa stream function and divergence, 200-hPa divergence, and genesis potential index (GPI) provided favorable TCG conditions. However, the Hit–Fail case differences in other suggested factors (vertical wind shear, 700-hPa moisture, and SST) were nonsignificant. By contrast, the reanalysis used for validation showed only significant difference in 850-hPa stream function. We stratified the background flow of TCG into four types. The monsoon trough type (82%) provided the most favorable environmental conditions for successful hindcasts, followed by the subtropical high (45%), easterly (17%), and others (0%) types. These results indicated that fvGFS is more capable of enhancing monsoon trough circulation and provides a much better environment for TCG development but is less skillful in other types of background flow that provides weaker large-scale forcing. The results suggest that the most advanced high-resolution weather forecast models such as the fvGFS warrants further improvement to properly simulate the subtle circulation features (e.g., mesoscale convection system) that might provide seeds for TCG.