Flow-Dependent Vertical Localization in GFS Hybrid 4DEnVar to Improve Tropical Cyclone Track Prediction

In ensemble-based data assimilation for numerical weather prediction, limitations in ensemble size due to computational cost restrictions introduce spurious error correlations. Ideal covariance localization dampens spurious correlations while retaining correlations with dynamic-based significance. Therefore, localization based on atmospheric flow or spatial length scales can help to discern between correlations with or without a dynamic connection. Currently, however, most operational data assimilation systems utilize a fixed localization value for all atmospheric phenomena. Several methods have been proposed to improve horizontal covariance localization based on flow- or scale-dependence. However, the literature on improving vertical covariance localization is limited.

This study will describe a vertical flow-dependent localization (vFDL) scheme to be used within the existing GFS hybrid 4DEnVar for global cycling experiments. Areas are identified for increased or reduced vertical localization through the value of the ensemble-based horizontal wind leading mode variance explained at each grid point. Global impacts of vFDL are greatest near jet level and in areas of low surface pressure. These areas typically exhibit greater correlation length scales in comparison with their surroundings. Additionally, results indicate that vFDL has the potential to improve tropical cyclone track prediction for up to 5 days in lead time compared to an experiment using constant localization. Diagnostics suggest that vFDL allows for more accurate correlations to be used for data assimilation throughout the depth of the tropical cyclone and surrounding environment, leading to improvements in the initial position of a tropical cyclone and its steering flow.