Stochastic Modelling of Rainfall Patterns for Agricultural Planning in Kebbi State

Irregular and unpredictable rainfall patterns in Nigeria's Northwest zone pose severe challenges to rain-fed agriculture, which remains the primary livelihood of rural households in Kebbi State. This study applies stochastic modelling techniques, specifically Markov chain and gamma distribution fitting, to characterize rainfall occurrence and intensity patterns at three meteorological stations in Kebbi State over a 20-year period. Daily rainfall data obtained from the Nigerian Meteorological Agency are used to estimate first-order Markov chain transition probabilities for wet and dry day sequences, as well as gamma distribution parameters for daily rainfall amounts on wet days. The fitted models are validated using the chi-square goodness-of-fit test and compared against observed historical rainfall statistics. Synthetic rainfall sequences are generated for 50-year simulation horizons to assess long-run drought risk and flood probability under two climate scenarios. Results indicate a statistically significant increase in the frequency of prolonged dry spells during the growing season over the past decade, consistent with regional climate change trends. The probability of a crop failure season, defined as fewer than 80 millimetres of rainfall in any 30-day growing period, is estimated at 23 percent. Recommendations include adopting drought-tolerant crop varieties, investing in micro-irrigation infrastructure, and integrating stochastic rainfall forecasts into State agricultural extension planning. This research provides a quantitative basis for climate-adaptive agricultural planning in Kebbi State. Keywords: stochastic modelling, rainfall patterns, agricultural planning, Kebbi State, Markov chain