Large ensemble tropical cyclone intensity forecasting

With the capability to generate hundreds of TC intensity forecasts for individual storms in the 0-15 day time scale it is necessary to develop efficient methods to communicate the results in a manner that is easy to understand and to use in decision-making. This entails communicating uncertainty (i.e., probabilistic information) as well as communicating quantitative skill metrics of these probability forecasts. We adapted the approach of Emanuel et al. (2008, BAMS, 347-367) for 0-15 day ensemble forecasts. For each storm the result is an ensemble of (100 or more) wind speed maps at each forecast time. This corresponds to a probability distribution of forecast wind speed at each location for each forecast time. We will compare several approaches to visualize and to compare to reality such large-ensemble forecasts outputs for interesting historical cases.