Characteristics of Tropical Cyclone Genesis Forecasts and Underdispersion in High-Resolution Ensemble Forecasting with a Stochastic Kinetic Energy Backscatter Scheme

This study evaluates ensemble forecasts with a stochastic kinetic energy backscatter scheme (SKEBS) to predict tropical cyclone (TC) genesis and also to characterize the related ensemble underdispersion. Several sets of ensemble forecasts are generated using an advanced research version of the Weather Research and Forecasting model at 5 km horizontal resolution to predict the genesis of Hurricane Ernesto (2006) and Typhoon Nuri (2008). Ensemble forecasts with SKEBS are compared against a control ensemble forecast with the WRF model using downscaled initial conditions derived from the NCEP Global Ensemble Forecasting System. It is found that ensemble forecasts with SKEBS are able to generate probabilistic forecasts for TC genesis and also capable of indicating the forecast uncertainties. Compared with the deterministic forecast that fails to predict the genesis of Typhoon Nuri, the ensemble forecast with SKEBS is able to produce the genesis forecast. However, the underdispersion of ensemble forecasts with SKEBS is also present in all cases in terms of the simulation period and over the whole model domain, TC environment, and inner core regions, although it is reduced near the TC inner core region. In addition, the initial perturbation–based ensemble forecasts shows slightly less underdispersion compared with the SKEBS ensembles.