Numerical simulations of tropical cyclones with assimilation of satellite, radar and in-situ observations: lessons learned from recent field programs and real-time experimental forecasts

The impact of data assimilation on the predictability of tropical cyclones is examined with the cases from recent field programs and real-time hurricane forecast experiments. First of all, mesoscale numerical simulations are performed to simulate major typhoons during the T-PARC/TCS08 field campaign with the assimilation of satellite, radar and in-situ observations. Results confirmed that data assimilation has indeed resulted in improved numerical simulations of tropical cyclones. However, positive impacts from the satellite and radar data are strongly depend on the quality of these data. Specifically, it is found that the overall impacts of assimilating AIRS retrieved atmospheric temperature and moisture profiles on numerical simulations of tropical cyclones are very sensitive to the bias corrections of the data. For instance, the dry biases of moisture profiles can cause the decay of tropical cyclones in the numerical simulations. In addition, the quality of airborne Doppler radar data has strong influence on numerical simulations of tropical cyclones in terms of their track, intensity and precipitation structures. Outcomes from assimilating radar data with various quality thresholds suggest that a trade-off between the quality and area coverage of the radar data is necessary in the practice.

Some of above experiences obtained from the field case studies are applied to the real time experimental hurricane forecasts during the 2010 hurricane season as part of the hurricane forecast improving project (HFIP). Results and issues raised from the case studies and real-time experiments will be discussed.