Spatial-Temporal Variation and Prediction of Rainfall in Northeastern Nigeria

In Northeastern Nigeria seasonal rainfall is critical for the availability of water for domestic use through surface and sub-surface recharge and agricultural production, which is mostly rain fed. Variability in rainfall over the last 60 years is the main cause for crop failure and water scarcity in the region, particularly, due to late onset of rainfall, short dry spells and multi-annual droughts. In this study, we analyze 27 years (1980–2006) of gridded daily rainfall data obtained from a merged dataset by the National Centre for Environmental Prediction and Climate Research Unit reanalysis data (NCEP-CRU) for spatial-temporal variability of monthly amounts and frequency in rainfall and rainfall trends. Temporal variability was assessed using the percentage coefficient of variation and temporal trends in rainfall were assessed using maps of linear regression slopes for the months of May through October. These six months cover the period of the onset and cessation of the wet season throughout the region. Monthly rainfall amount and frequency were then predicted over a 24-month period using the Auto Regressive Integrated Moving Average (ARIMA) Model. The predictions were evaluated using NCEP-CRU data for the same period. Kolmogorov Smirnov test results suggest that despite there are some months during the wet season (May–October) when there is no significant agreement ( p < 0.05) between the monthly distribution of the values of the model and the corresponding 24-month NCEP-CRU data, the model did better than simply replicating the long term mean of the data used for the prediction. Overall, the model does well in areas and months with lower temporal rainfall variability. Maps of the coefficient of variation and regression slopes are presented to indicate areas of high rainfall variability and water deficit over the period under study. The implications of these results for future policies on Agriculture and Water Management in the region are highlighted.