HAFS Ensemble Data Assimilation and Prediction - A Real Time Experiment for North Atlantic Basin in 2023

The Hurricane Analysis and Forecast System (HAFS) regional ensemble data assimilation (DA) and prediction real-time experiment has been running for the 2023 Atlantic hurricane season as one of the real-time demo projects supported by the Hurricane Forecast Improvement Program (HFIP). The real-time system has one 3-km resolution deterministic forecast (DF) and 20 6-km resolution ensemble forecast (EF) members. The DA system utilizes the hybrid three-dimensional ensemble-variational (3DEnVar) algorithm for the DF and the ensemble Kalman Filter (EnKF) for ensemble perturbations. The 3DEnVar uses an augmented covariance consisting of 6 hr forecasts 80-member operational Global Data Assimilation System (GDAS) ensembles as well as 20-member ensembles from the self-cycled 6-km resolution EF members, while the EnKF is based on 20 self-cycled ensemble perturbations. The real-time experiment was initialized on June 28 2023, and implemented the DA system with continuous self-cycling updates.

Compared to the operational HAFS system, which was implemented in operations on June 27, 2023, this real-time experiment is aimed at 1) providing a proof of concept that a self-cycled HAFS DA system with mixed HAFS/GDAS ensemble covariance, and online radiance bias correction can provide improved TC initial conditions; 2) better understanding and tuning HAFS model and observation error characteristics; 3) providing improved track and intensity forecasts over its deterministic system; 4) providing probabilistic forecasts, including probability forecasts of Rapid Intensification (RI) and helping forecasters in the decision-making process; and finally 5) providing uncertainty-based analyses used for aircraft reconnaissance flight mission tracks.