The Influence of the HAFS-Vortex Initialization on the Systematic Low Bias of Tropical Cyclone Central Pressure in the HAFS Initial Condition

Vortex initialization (VI) was successfully implemented into the first operational Hurricane Analysis and Forecast System (HAFSv1.0A) in 2023 and underwent significant upgrades in 2024. These upgrades for HAFSv2.0A were a) the introduction of the cloud-vertical velocity relocation/cycling capability, and b) an update on the composite vortex. However, despite these VI upgrades, there have been several reports indicating that, at times, the central pressure of strong tropical cyclones (TC) at the model initial time is too low compared to the observation. Therefore, we analyzed the TC central pressure of 3 years of HAFS forecast results from 2021 to 2023. The analyses indicated that, in the North Atlantic basin, both HAFS-A and HAFS-B consistently exhibit a low bias in the central pressure at the model initial time when the TC is deeper than approximately 970 hPa. In contrast, the Hurricane Weather Research and Forecast model does not have such a noticeable bias. Also, the HAFS TC central pressure at the model initial time shows the same low bias in the Western Pacific basin where data assimilation is not used. It suggests that the low central pressure bias at the HAFS model initial time is primarily caused by VI. The analyses for the strong TC cases that have too low central pressure suggest that the vortex enhancement procedure in VI is the partial reason for this central pressure bias. Thus, the method that handles the composite vortex in the vortex enhancement procedure is modified. The new method improves the central pressure error and bias at the model initial time for both North Atlantic and Western Pacific basins.