Assessing the Impact of Initial Condition Errors on Intensity Forecasts of Hurricane Katia (2011)

Why is predicting the strongest winds within tropical cyclones (TCs) such a challenging task? One possible reason is the uncertainty in the forecasts that originates from poor representations of TC structure in the initial conditions of numerical models. To understand how initial-time errors can lead to large intensity forecast uncertainty, this study examined ensemble forecasts from the Advanced Hurricane Weather Research and Forecasting (AHW) model. Specifically, the research efforts were focused on the poorly forecasted Hurricane Katia (2011) by using 96 ensemble forecasts from the AHW model. Two distinct subgroups were identified and extensively studied: 1) 10 members that predicted a strong hurricane (named strongest members) and 2) 10 members that predicted a weak storm (named weakest members). Composites of these two subgroups were created to diagnose how differences in the initial conditions of these members resulted in substantially different forecast scenarios. Results indicated that both subsets were initialized with a relatively strong vertical wind shear, but the strongest members had more water vapor in the downshear and right-of-shear quadrants. The moist environment in these quadrants aided TC intensification by maintaining and enhancing the shear-organized convection. Initial vorticity asymmetries were also substantially different. In the strongest members, a cyclonic vorticity feature located downshear seemed to have played an important role in strengthening the TC vortex. These findings suggest that sampling and assimilating observations from the downshear and right-of-shear flanks of TCs could help reduce the uncertainty in the initial conditions, thus increasing the accuracy of numerical TC intensity forecasts.