Tropical Cyclone Intensity Forecasting: Predictability and Uncertainty

In spite of recent efforts and advancements in numerical weather prediction (NWP) models and data assimilation, forecasting tropical cyclone (TC) intensity remains a major challenge. The lack of a significant progress in TC intensity prediction has motivated studies seeking for better understanding of fundamental questions about the predictability limit of TC intensity and forecast uncertainty. Some studies, using idealized model simulations, have focused on the control of TC intensity by internal variability within an idealized storm. The conclusions of these studies suggest that the predictability limit of TC intensity is less than 3 days, mainly because the intrinsic dynamics are dominated by small-scale circulations driven by convective processes. This study, however, takes a different approach by allowing TCs to interact explicitly with their environment in realistic ensemble forecasts. A set of high-resolution (1.3 km grid resolution) ensemble experiments using the weather research and forecasting (WRF) model and the stochastic kinetic energy backscatter (SKEBS) perturbation method show that the predictability of TC intensity in a realistic framework is quite different than in the idealized simulations. The forecast uncertainty with respect to TC intensity is strongly related to uncertainty in the environmental conditions, e.g., vertical wind shear in Hurricane Earl (2010). Even though there is large uncertainty in terms of timing of rapid intensification, the vortex scale forecast error grows significantly slower compared to errors on smaller scales. An error growth analysis in spectral space reveals that the low wavenumber component (wavenumbers 0 and 1) of the cyclone vortex features predictability limits up to 7 days (and possibly beyond) for a long-lasting, non-landfalling TC. These results show that environmental control over the main feature of the TC vortex can lead to significantly longer upper limits of TC intensity predictability. Predictive skill can be gained by improved prediction of storm structure and environmental flow. However, a cautionary note is that the increased uncertainty during rapid intensification and at longer forecast times is a sign of the fundamental limitation of a deterministic TC intensity forecast.