Probability Models for Predicting Flood Events Along the Niger-Benue Confluence in Kogi State

This study applies probability models to the prediction and risk assessment of flood events along the Niger-Benue confluence in Kogi State, North Central Nigeria, a region that experiences some of the most severe and recurrent flooding on the African continent. The confluence zone has been identified in recent hydrological surveys as particularly vulnerable to catastrophic inundation, with flood events causing significant loss of life, displacement of communities, and destruction of agricultural produce. Using annual peak discharge data sourced from the Nigerian Hydrological Services Agency for the period 2012 to 2023, the study fits Gumbel extreme value distributions, log-normal distributions, and Pearson Type III distributions to the observed flood records. Goodness-of-fit tests, including the Kolmogorov-Smirnov and chi-squared tests, are applied to identify the distribution that best characterises the historical flood data. Return period estimates are computed for flood events of varying magnitudes, and probability exceedance curves are constructed to communicate risk to non-technical stakeholders. Results indicate that the Gumbel Type I distribution provides the most statistically defensible fit to the Kogi State flood data. The study recommends that local government authorities incorporate probabilistic flood risk assessments into land-use planning and disaster preparedness protocols. The findings offer a quantitative foundation for evidence-based flood management policy in one of Nigeria's most flood-prone geopolitical zones. Keywords: probability models, flood prediction, extreme value distributions, Niger-Benue confluence, Kogi State