Dependence of Hurricane Track Forecasts on the Spectral Representation of Cumulus Convection

Abstract

This study investigates the sensitivity of hurricane track and intensity forecasts to various cumulus parameterization schemes with scale-awareness, including the Simplified Arakawa-Schubert (SAS), Relaxed Arakawa-Schubert (RAS), and Chikira-Sugiyama Arakawa-Wu (CSAW), within the NOAA Global Forecast System (GFS) version 17 framework. Using a set of nine initial conditions for Hurricane Ian (2022), we demonstrate that schemes utilizing a spectrum of cloud types (RAS and CSAW) produce westward-shifted trajectories compared to the single-cloud SAS scheme, which exhibits an eastward track bias. For intensity forecasts, SAS was more closely aligned with observations over RAS and CSAW. When testing the sensitivity of the CSAW convective parameterization with Wb (cloud-base vertical velocity) spectrum and entrainment rates, a critical inverse relationship was revealed between parameterized convective strength and resolved storm intensity. Specifically, reducing the Wb range or increasing entrainment attenuates the sub-grid scale convective response, facilitating a compensatory intensification of the resolved-scale vortex and kinetic energy. Spectral analysis indicates that cloud-spectrum schemes exhibit reduced energy variance in low-frequency modes, which directly impacts large-scale steering. Furthermore, potential vorticity (PV) analysis at 500 hPa confirms that track divergence is governed by the spatial orientation of PV anomalies, with the storm propagating toward regions of maximum PV gradient. These findings underscore that the representation of the cloud spectrum and the subsequent energy partitioning between parameterized and resolved scales are fundamental controls on tropical cyclone evolution in high-resolution numerical weather prediction models.