Towards the perfect warning

Protection of lives and property from hazardous weather, through the provision of weather warnings, is a core mission of weather services with growing importance as global climate and human changes increase both the exposure and vulnerability of society to weather-related hazards. Exploring how to achieve that most effectively is the aim of the World Weather Research Programme’s High Impact Weather (HIWeather) project. HIWeather brings together physical and social scientists from a wide variety of disciplines and from across the world to study each step of the process from monitoring the weather to making effective protective responses. HIWeather uses a simplified conceptual model of the warning value chain that identifies the roles of key actors and organisations involved in forecasting the weather, the resulting hazard and its socio-economic impacts, in formulating the warning and communicating it to the end-user. HIWeather has drawn on a wide body of research to explore how these different actors can make their expertise contribute more effectively to the desired outcome of reduced death, destruction and disruption. In this talk I shall summarise the results of that research, identifying key principles that should be incorporated in the design of warning systems. In doing so, I shall connect this work with ideas from the design of community-based warning systems, with developments in social media communication, with research on impact-based forecasting, and with progress in convection-permitting and higher resolution NWP models. A key result is that the communication of knowledge is at least as important as its content, and that the creation and nurturing of partnerships between organisations is critical to that. Looking forward, I shall describe a new HIWeather initiative, launched at the end of last year, to gather data on the real-life performance of end-to-end warning systems in specific events for the purpose of analysing how warning outcomes are related to warning system characteristics.