This workshop is part of the ICMS Knowledge Exchange Catalyst project “Transforming Scottish avalanche forecasting”, a collaboration between the University of St Andrews and the Scottish Avalanche Information Service.
Forecasting rare but high-impact events such as avalanches, floods, heatwaves, and terrorist attacks poses distinctive scientific and practical challenges. These events are often characterised by sparse data, complex causal chains, and high uncertainty. Yet decisions based on their forecasts can have major impacts on public safety, resource allocation, and risk communication. Understanding how forecasts are produced, verified, and interpreted in these settings is crucial for improving their accuracy, reliability, and usefulness.
This interdisciplinary workshop brings together researchers and practitioners from natural hazard forecasting, risk analysis, psychology, philosophy and statistics to explore how severe event forecasts are made and evaluated. The goal of the workshop is to address the technical and behavioural challenges of severe event forecasting, in particular:
- forecast verification in low-feedback environments,
- the role of statistics, machine learning, and AI in the forecasting and verification process,
- behavioural influences on forecasts, and
- how best to operationalize any insights that various tools provide.
While avalanches provide a focal case study, many of the same challenges arise across domains, and workshop participants are drawn from various applications areas. By comparing approaches across different severe-event contexts, we aim to identify shared problems, transferable solutions, and potential directions for cross-disciplinary collaboration.
