Evaluation of Hurricane Analysis and Forecast System (HAFS) Error Statistics Stratified by Internal Structure and Environmental Metrics

Abstract This study quantifies tropical cyclone (TC) error statistics from the Hurricane Analysis and Forecast System (HAFS) across different environmental conditions (e.g., vertical wind shear) and inner-core structural metrics. A particular focus is the evolution of poorly understood aspects of internal TC structure, including vortex tilt, and their impact on forecast errors. Although previous studies have demonstrated that vortex tilt, vertical wind shear, and precipitation processes impact TC intensity and track, this is the first known study to stratify these cooperative interactions to gain insights into their relationships with forecast errors. A 3-yr retrospective sample of forecasts in the North Atlantic basin from two HAFS configurations (HAFS-A and HAFS-B) demonstrates that TCs with larger tilt magnitudes have larger forecast track errors on average than smaller tilt TCs. Smaller tilt magnitudes have larger absolute intensity errors in short-range forecasts, whereas larger tilt magnitudes tend to have larger negative intensity biases at medium range. TCs with a tilted vortex are shown to have both left-of-shear [maximizing downshear left (DSL)] and left-of-tilt-oriented positional track biases. Furthermore, those cases with greater downshear biases tend to have more convection and larger positive intensity biases, highlighting the importance of the interplay between inner-core characteristics and forecast errors.