Disaster risk communication

Adaptation is not a substitute for mitigation. Cutting emissions remains indispensable to preventing the worst outcomes of climate change. Even if we succeed in limiting warming to 1.5°C, we will still face more floods, heatwaves, and ecosystem losses.

Climate science and adaptation strategies can only make an impact in the real world, when people can understand and act upon them.

The farmer’s phone had vibrated just an hour earlier – a community alert of heavy rainfall expected within 48 hours. These messages now arrive regularly.

The inland extent of watches and warnings will now be included, and an experimental cone will capture more areas at risk.

This publication explores how generative language models combine social media, news and weather data to detect disasters quickly and accurately, using satellite data to validate results from flood and wildfire case studies.

This review analyzes peer-reviewed studies to examine how risk communication strategies vary across temporal dimensions, hazard groups and intended purpose.

The study examines how visual cues, especially maps and warning symbols, can improve flood early warning messages by helping citizens understand event magnitude, timing, impacts and recommended protective actions.

This study aims to produce structured storylines and knowledge graphs for over 3,000 global disaster events by processing news articles through a RAG pipeline combining the power of LLMs for text generation and the semantic extraction of factual knowledge