Using a regional climate model and a machine learning algorithm for farmer - herder conflict prediction in the Sahel region

Abstract

Climate variability and environmental stress are increasingly recognized as key factors influencing conflict dynamics in vulnerable regions. This study explores the relationship between climate variability and conflict risk in northern Nigeria's Sudano–Sahelian region by combining regional climate modeling with machine learning techniques. High-resolution climate projections were produced using the Weather Research and Forecasting (WRF) regional climate model to simulate essential climate variables, including precipitation, temperature, and soil moisture. These data were used to develop environmental stress indicators such as rainfall anomalies, drought frequency, and the Standardized Precipitation Index (SPI). Past conflict event data from the Armed Conflict Location and Event Data (ACLED) project and the Uppsala Conflict Data Program (UCDP) were integrated with climate variables and socioeconomic factors to train machine learning models for prediction. The findings show that drought severity, rainfall variability, and soil moisture deficits are among the most significant predictors of conflict risk in this area. Of the tested algorithms, the Random Forest model provided the highest accuracy, with an Area Under the Curve (AUC) of about 0.86, surpassing traditional regression methods. Spatial maps highlight several conflict hotspots in northern parts of the Sudano–Sahelian region where climate stress and population pressure are especially intense. The results suggest that climate-induced environmental stress can escalate competition over land and water resources, raising the likelihood of resource-based conflicts, especially between farming and pastoral groups. By combining regional climate modeling with machine learning-based prediction techniques, this study offers a new framework for understanding and foreseeing climate-related conflict risks in climate-sensitive areas. The findings underscore the potential for integrating climate science and data-driven conflict prediction to support early warning systems and guide proactive policies that enhance resilience and prevent conflicts in the Sahel and similar vulnerable regions.