ROUGE: A database of disaster impacts in the Global South using Red Cross reports and Large Language Models

Damage from natural hazards exacts a heavy toll on society and is expected to increase under climate change. Yet, existing impact datasets remain limited and often biased toward the Global North and monetary loss metrics. This bias constrains our capacity to robustly assess socio-economic consequences, particularly in regions where impacts are most acute and least documented. To help address these gaps, we present ROUGE (Red cross Operations Unified Global Emergency), a new global socio-economic impact database obtained using textual operational reports from the International Federation of Red Cross and Red Crescent Societies (IFRC). These reports are systematically collected and provide extensive  coverage of regions that are commonly underrepresented in existing impact datasets.

Using large language models, we extract qualitative and quantitative information from 717 reports spanning 2016-2025. The dataset records 11,370 impacts across 20 socio-economic impact subtypes, reported at both national and sub-national scales. The most frequently reported impacts relate to Water, Sanitation and Hygiene, Agriculture and Access to Food, Affected People, Residential Buildings, and Economy and Livelihood. We validate the database against manually-labelled reports and established disaster impact databases, including EM-DAT or IFRC-GO. Results show that extraction performance varies across impact subtypes, with precision ranging from 0.3 to 0.9 and recall ranging from 0.1 to 0.8. Comparisons with external datasets reveal differences in impact figures, reflecting the inherent challenge associated with quantifying natural disaster impacts. However, for overlapping events, our database more frequently provides quantitative impact values than existing datasets. 

Overall, ROUGE opens new avenues for disaster impact research by delivering geographically explicit socio-economic impact data from IFRC reports. The resulting dataset captures the impacts of natural hazards on both populations and the built environment, with spatial resolution extending to the subregional level, capturing impacts that are rarely represented in conventional databases. By doing so, ROUGE enables more precise, inclusive, and globally representative analyses of the socio-economic consequences of natural hazards worldwide.