Improving Tropical Cyclone Track Forecasts through Ensemble-Based Sensitivity Analysis Methods

Ensemble-based sensitivity analysis enables one to identify regions that can have a significant impact on subsequent forecasts of tropical cyclone tracks or other metrics. As such, this method can be used to suggest where assimilating additional observations might improve the forecast, or even what group of ensemble members might be more likely to verify. This research investigates whether these ensemble based diagnostics combined with observations can be used to identify which ensemble members are likely to produce a more accurate forecast at longer lead times with relatively minimal computational expense. Using the ECMWF ensemble TC track forecasts for 2014-2017 we will present a methodology to reduce ensemble track error and spread for 48-120 hr forecasts