Abstract/Summary

Ensemble forecasts play an essential role in providing early warnings to mitigate the impact of hazardous weather events. However, there are still many areas where ensemble information is not fully exploited. One key issue limiting the uptake of ensemble forecasts is the jumpiness that can sometimes occur between successive forecasts. Ensemble forecasts show the range of future weather scenarios that can occur, allowing users to make appropriate risk-based decisions. Occasionally a new forecast seems to contradict the previous forecast by introducing a new weather scenario that was not represented in the previous ensemble. Such inconsistencies can cause users to lose confidence in the forecasting system. This thesis aims to improve the diagnosis and understanding of these ensemble forecast inconsistencies. First, a methodology is developed to quantify the consistency between a sequence of ensemble forecasts valid for a given time, taking account of the full ensemble distribution. This enables a quantitative evaluation of the consistency or jumpiness between successive forecasts, providing insights into the relationship between jumpiness, skill and spread. The thesis also provides practical guidance to address user concerns about ensemble jumpiness. It provides specific guidance that will enable users to make better use of the available operational ensemble tropical cyclone track and genesis forecasts. The thesis shows that evaluation of forecast consistency is complementary to the current focus on skill and ensemble spread, and that an integrated approach using both skill and consistency measures can be beneficial in evaluation of ensemble forecast performance. Implementation of this approach at NWP centres will ensure that users have the necessary information and guidance to mitigate the impact of run-to-run jumpiness and will provide feedback to model developers on model weaknesses, complementing existing evaluation tools. The research in this thesis will help improve the utilisation of ensemble forecasts to provide early warnings of significant weather hazards, contributing to the UN Early Warnings for All initiative.