The upgraded MOGREPS-G ensemble forecasts for the storms in the active and high-impact 2017 North Atlantic hurricane season will be presented and compared to other global ensembles, both individually and in a multi-model ensemble combination, for tropical cyclone genesis, track and intensity. These cases also revealed interesting information about the perception of forecast value, and how dependent this is on the users’ specific needs and focus.
The overall skill and value of ensemble forecasts can only be fully assessed via objective probabilistic verification over multiple cases. The Met Office routinely verifies named storm strike probability forecasts from each global ensemble, and the various multi-model combinations, to assess their probabilistic skill, reliability and value in each tropical cyclone basin. The tropical cyclone ensemble verification capability at the Met Office has been extended to include tropical cyclone activity forecasts, allowing an evaluation of the forecast skill at a range of lead times for both named storms and those forecast to form during the forecast. Verification continues to show that the ensemble forecast model with the highest probabilistic skill varies from storm to storm and that overall the greatest value is achieved by using probabilistic multi-model ensemble forecasts.
Following recent resolution increases in several global ensembles, it is important to re-evaluate current levels of probabilistic forecasting skill for tropical cyclone intensity in global ensemble forecasts. Recent work at the Met Office to verify the skill in global ensemble forecasts of intensity trends will also be presented. This work forms part of a new international collaboration, as a sub-project of the WMO’s HIWeather project, on the topic of the ensemble forecasting and verification of tropical cyclones. Ensemble forecasting of tropical cyclones is vital in capturing the situation-dependent uncertainty in the track and intensity forecasts for tropical cyclones. The HIWeather collaboration aims to evaluate and demonstrate the benefits of using ensemble forecasts, and gather current and future user requirements, with a view to developing new and user-orientated ways to display and verify probabilistic tropical cyclone forecasts, and increase the use of ensemble forecasts in tropical cyclone forecasting.