Flood Susceptibility Mapping and Climate Change Impact Prediction Using Probabilistic Machine Learning and Statistical Analysis: A Case Study of Kigali, Rwanda

ABSTRACT Flooding is among the most destructive natural hazards globally, and its frequency and intensity are rising. Climate change and rapid urbanization are increasing flood susceptibility, especially in cities with limited infrastructure. This study assesses current and future flood susceptibility in Kigali, Rwanda, using the Maximum Entropy (MaxEnt) model. Presence‐only flood occurrence records were combined with environmental predictors including slope, soil texture, drainage density, land use/land cover, rainfall, flow accumulation, and topographic wetness index, together with two climate change scenarios (SSP245 and SSP585). Model performance was strong (AUC = 0.84). Slope is the most influential factor (≈84.9% contribution), followed by soil texture, drainage density, and land use/land cover. Currently, about 117.16 km 2 (16.19%) of Kigali is classified as high and very high flood susceptibility zones, mainly located in low‐lying areas. In the SSP585 scenario, these zones expand to 118.44 km 2 (16.36%) by 2040 and remain similar (118.36 km 2 , 16.35%) by 2080. The projections indicate a marginal increase in high and very high flood susceptibility zones as climatic conditions intensify. These findings emphasize the need for integrated flood risk management through improved drainage infrastructure, land use regulation, and proactive urban planning to build resilience. The results provide spatially explicit risk maps to support policymakers and planners in Kigali and other rapidly growing African cities facing climate change pressures.