Impact of Simultaneous Multiscale Data Assimilation with Scale- and Variable-Dependent Localization on Hurricane Analysis and Prediction in HAFS

The complex evolution of hurricanes involves interactions across multiple scales. For instance, the large-scale synoptic environmental flow typically dominates the movement of hurricanes, while small-scale convective bursts often impact the intensity evolution of hurricanes. Due to computational capacity limitations, localization is usually the most efficient method to constrain these interactions without introducing spurious noise from the sampling insufficiency. However, traditional hurricane data assimilation (DA) systems rely on a single localization length scale (SSL) method, which is insufficient for accurately estimating the atmospheric state across multiple scales and multiple variables needed to initialize hurricane forecasts. Recent successes in simultaneous multiscale data assimilation (MDA) have shown promising results in the analysis and predictions of tornadic events using scale- and variable-dependent localization (SDL/VDL; Wang and Wang 2023). In this study, the MDA with SDL/VDL capabilities will be further implemented and optimized for the Hurricane Analysis and Forecast System (HAFS), focussing on improving hurricane forecasting. Experiments will be conducted and analyzed to demonstrate the benefits of the SDL/VDL MDA method relative to the SSL method.