Initializing the WRF Model with Tropical Cyclone Vital Record for Typhoon Forecasts based on the Ensemble Kalman Filter Algorithm

This study presents an approach to directly assimilate tropical cyclone vitals (TCvital) and the University of Wisconsin-Cooperative Institute for Meteorological Satellite Studies (CIMSS) Atmospheric Motion Vectors (AMV) data into the Weather Research and Forecasting (WRF) model for real-time tropical cyclone (TC) forecast applications. Unlike current approaches in which the TCvitals are used to either generate a bogus vortex or dynamically spin-up model initial condition, the proposed approach in this study ingests the TCvital information through blending a dynamically consistent synthetic vortex structure with the CIMSS-AMV data. The combined data is then assimilated into the WRF initial condition, based on the local ensemble transformed Kalman filter (LETKF) algorithm. This blending approach helps enhance both the storm-scale TC representation and the large-scale environment without complications of removing an initially weak vortex or inserting a bogus vortex into the model initial condition. Examination of the performance of blending the TCvital and the CIMSS-AMV data for several typhoon cases in the north Western Pacific basin in 2013 shows that this approach could effectively increase both the TC circulation and the large-scale environment that TCs are embedded in. While the TC track forecasts are benefited more from the improvement in the large-scale flows, our result suggests that use of the TCvital-based synthetic information helps reduce the initial vortex spin-up time, thus allowing for model storms to quicker adjust to the ambient environment and reducing the intensity errors at longer ranges. Our approach will be most beneficial for future TC models, which are initialized directly from very high resolution global models in which storm initial locations are sufficiently accurate at the initial analysis that there is no need to carry out any artificial vortex removal or insertion process.