Cloud and Vertical Velocity Field Improvement in the HAFS-Vortex Initialization

Vortex Initialization (VI) is widely used in regional hurricane models, including the Hurricane Analysis and Forecast System (HAFS). However the VI (i.e., old VI), which was used in the operational HAFS (HAFSv1.0), does not have a capability to improve cloud variables, and thus there is an inconsistency between cloud and other variables (e.g., horizontal wind and pressure) in the model initial condition. In the 2023 real-time HAFSv1.1 experiment, VI was significantly improved as a relocation/cycling capability for cloud and vertical velocity fields was introduced into HAFS-VI. In this new VI, cloud and vertical velocity fields in the initial condition are relocated (cold start) or are cycled from the 6-h HAFS forecast of the previous cycle (warm start), which yields improved initial conditions.

To examine the impact of the new VI, two sensitivity experiments (i.e., HFNN and HFNW) for 13 hurricanes were conducted based on the HAFSv1.1 experiment that uses the relocation/cycling capability for both cloud and vertical velocity fields. The HFNN experiment turns off the cloud and vertical velocity relocation/cycling in the new VI (i.e., equivalent to the old VI configuration), and the HFNW experiment applies the relocation/cycling capability to cloud variables but not to the vertical velocity field. A comparison of the HAFSv1.1 and HFNN experiments reveals that the new VI not only improves the intensity forecast to a certain degree but also generates more realistic cloud field in the initial condition than the old VI.

Interestingly, the HFNW experiment shows that when the vertical velocity is not cycled or relocated the intensity forecast is worse than the HFNN and HAFSv1.1 experiments, despite using the cloud relocation/cycling capability. This finding strongly suggests that the vertical velocity field in the initial condition is another important factor that might affect the intensity forecast.