GIS-based spatiotemporal flood hazard assessment using statistical and fuzzy logic in the Omo Gibe River Basin, Ethiopia

The Omo Gibe River Basin is the largest basin in southwestern Ethiopia. It’s frequently exposed to flooding driven by natural and human-induced factors, threatening ecosystems, human life, and property. This study develops an integrated flood hazard assessment by combining statistical and fuzzy logic approaches within a GIS framework. Despite growing risks, assessments integrating ensemble modeling, multi-factor interactions, spatio-temporal analysis, and comparative evaluation of fuzzy overlay methods remain limited. To address this gap, frequency ratio and fuzzy logic were applied to assess flood susceptibility across three decades. Prior to modeling, multicollinearity among flood conditioning factors was assessed using the Variance Inflation Factor (VIF) and tolerance values. Flood inventory maps were generated from the Modified Normalized Difference Water Index, ground truth and an existing datasets; 70% of flood locations were used for training and 30% reserved for validation. Five fuzzy overlay operators were evaluated using Receiver Operating Characteristic curves, Fuzzy Gamma achieving the highest performance (success rate 89.52%, prediction accuracy 91.42%). Very high flood hazard zones expanded from 5.68% in 1986 to 9.05% in 2006 due to intense rainfall and inadequate land-use management, then declined to 5.53% in 2024 following the development of hydrological infrastructure and improved basin management. This study advances hazard modeling by addressing uncertainty and multicollinearity, providing a validated, data-efficient approach for flood mapping in data-scarce regions. The findings support Sustainable Development Goals 11.5 and provide evidence-based insights for floodplain zoning, infrastructure planning, and resilience building. Future research should incorporate climate change scenarios and socio-economic vulnerability to strengthen long-term flood risk management.