From indices to impacts using environmental and socio-economic clustering in Kenya

Study Region: Arid and Semi-arid counties in Kenya. Study Focus: The Horn of Africa faces recurrent droughts, impacting its population. Global warming worsens these events, making early warning systems crucial. However, the link between hazard indices and actual impacts is unclear. This study identifies hazard indices responsible for drought impacts: malnutrition, milk production, and trekking distances to water. Using Spearman correlation and Random Forest regression, relationships between drought indices and impact datasets are investigated. Counties are clustered by vulnerability factors (poverty, water access, etc.) to identify key drought impact drivers. The performance of cluster-specific models is compared to a global model covering all counties. New Hydrological Insights for the Region: Trekking distances to water correlate with meteorological drought indices (2–6 months), indicating sensitivity to short-term droughts. Milk production associates with various indices (5–24 months), showing complex drought effects on livestock. Malnutrition links to vegetation- and streamflow-based indices (5–24 months), highlighting nutritional deficits from prolonged droughts. Clustering regions by vulnerability factors enhances the hazard-impact relationship, confirming regional characteristics' influence on impact severity. Effective models for trekking distances were created in clusters indicative of water access, sanitation, poverty, and aridity. Accurate malnutrition models were built in clusters based on aridity, food consumption, water access, sanitation, and poverty. These findings highlight the need for region-specific drought monitoring and management to enhance resilience in Kenya's arid and semi-arid counties.