Application of scale-selective data assimilation to tropical cyclone track simulation

Tropical cyclone track is strongly controlled by the large-scale environmental circulation. In limited-area models (LAMs) driven by global analyses or forecasts through a conventional lateral boundary nesting approach, the global analyses are often distorted by the use of "sponge zone" or interpolation when they are passed into the LAM. In this study, a dynamical downscaling approach based on scale-selective data assimilation (SSDA) is applied to a limited-area numerical weather prediction model with emphasis on tropical cyclone track simulation. The idea of the SSDA approach is to drive the LAM not only from the lateral boundary but also from the model domain interior. The large-scale flow from global analyses or forecasts is assimilated into the regional model using 3-D variational data assimilation. The large-scale features in the LAM are thus constrained to follow the global analyses while allowing the regional model itself to develop the regional and small-scale characteristics. The results from the case study of Hurricane Katrina (2005) show that both large-and small-scale flows in the regional model benefited from the SSDA approach, leading to an improvement in the accuracy of storm track simulation when provided with an accurate large-scale circulation from global analyses. In addition, the SSDA procedure is shown to be an effective method to construct a nested-grid regional modeling system that reduces model sensitivity to model domain geometry and location.