Probabilistic Global Convective Hazard Forecasts and Verification

Severe convection can cause a range of weather hazards which are a danger to both the general public, claiming many lives each year, and various commercial sectors, costing industry many millions of dollars each year. In 2012, 310 natural disasters were recorded in the Emergency Events Database (EM-DAT), claiming 9,930 lives and causing economic damages of US$138billion. In 2011, 553 people in the United States were killed by tornadoes, and in 2012 a single hail storm in Texas caused estimated damages of $1billion. Convective storms can have implications on a global scale for the aviation industry, causing delays and cancellations that cost the industry millions of dollars each year. Accurate prediction of these events can help enable mitigating action to take place potentially saving lives and minimising the losses to businesses.

The UK Met Office have a role in providing advice on the potential for hazardous global weather to the FCO and to NGOs overseas. The Met Office produce forecasts for a range of convective hazards produced by the Met Office Convection Diagnosis Procedure (CDP) system (Hand, 2011). Initially the CDP was set up to use deterministic model data, however since 2012 the system has been running from the Met Office Global and Regional Ensemble Prediction System in the global configuration (MOGREPS-G). This has the potential to provide forecasters with global probabilistic guidance for a range of convective events such as tornadoes, lightning and hail. The use of probabilistic rather than deterministic forecasts further enhances the skill and value of the CDP system.

In this study we look at several case studies to assess the performance of the CDP system at locations around the world. Subjective verification will be used to assess performance throughout the 2013 tornado season in the United States, to identify strengths and potential improvements of the tornado diagnostic. Objective verification of lightning forecasts over Europe is also presented to demonstrate the skill and reliability of the product. The lightning forecasts are verified against a stroke count within a certain radius of airports in Europe over the forecast period. The individual stroke locations are registered by the Met Office ATDnet (Arrival Time Difference network) system that detects the low frequency radio waves (‘Sferics') generated by a lightning stroke. This study will help us to explore the value of ensemble and probabilistic forecast products on a global scale.