Evaluating Improvements in PBL and Cumulus Schemes in HAFS for Forecasts of TC Structure and Large-Scale Steering

Planetary boundary layer (PBL) and convective physics are both key for accurate prediction of tropical cyclone structure and evolution. A key question is how physics behave at different scales, as tropical cyclones are complex systems affected by processes ranging from cloud-scale to synoptic scale. In this study, we evaluate tropical cyclone forecasts across a variety of scales and applications, to examine how the physics parameterizations perform in these different settings, specifically focusing on the PBL and cumulus schemes. An examination of Hurricane Laura (2020) shows how both Hurricane Analysis and Forecast System (HAFS) and high-resolution UFS runs had a left bias in track, while a lower resolution run was more realistic, indicating that the physics were not optimal at higher resolution for this case. Building on this scaling idea, we examined a new prognostic closure scheme for the scale-aware SAS in forecasts of Hurricane Ian. Track results were mixed, but generally positive for the HAFS-B configuration, although some runs showed excessive deepening, indicating a need for further adjustment of mass flux calculations. Finally, to examine how TC-specific physics impact large-scale forecast skill, we examined 25-km runs of the Global Forecast System (GFS) using TC-specific modifications to the EDMF-TKE PBL scheme. For one month of runs, we found that the PBL changes led to improvement in some surface-based parameters (including the dew point) but slightly degraded the large-scale forecasts. Together, these results show the need to better understand the performance of parameterizations at different scales, and also how parameterizations interact in various applications.