Statistical Modelling of Rainfall Patterns in Plateau State for Agricultural Planning

Accurate statistical characterisation of rainfall patterns is fundamental to agricultural planning in Plateau State, North Central Nigeria, where smallholder farmers remain predominantly rain-fed and highly vulnerable to seasonal precipitation variability. This study develops statistical models of rainfall patterns in Plateau State using data from the Nigerian Meteorological Agency Jos Station spanning 2000 to 2022. Monthly rainfall totals were analysed for distributional properties, stationarity, seasonality, and trend. Kolmogorov-Smirnov tests assessed distributional fit, and Mann-Kendall tests examined monotonic trend significance. Fourier analysis decomposed seasonal periodicity, and ARIMA modelling provided short-term forecast capability. Mean annual rainfall was 1,312 mm with a coefficient of variation of 19.4 percent, indicating moderate inter-annual variability. Mann-Kendall analysis revealed a statistically significant declining trend in June rainfall (tau = -0.31, p = 0.018), implying delayed onset of the wet season. April and May rainfall showed significant increasing trends. Fourier decomposition confirmed a dominant annual cycle with a secondary semi-annual harmonic. The ARIMA(2,1,1)(1,1,0)12 model provided the best seasonal forecast fit with a root mean square error of 31.4 mm. The study provides agricultural planners and the Plateau State Agricultural Development Programme with a quantified rainfall uncertainty profile and recommends integration of seasonal forecast outputs into crop calendar advisory services. Keywords: rainfall patterns, ARIMA, agricultural planning, Plateau State, Mann-Kendall trend