Long-term Estimates of Reservoir Evaporation Using ARIMA Model and Impact on Water Supply: A Case Study of Erinle Dam, Osun State, Nigeria

It is common practice in water resource management to estimate evaporation of water from reservoirs using nearby measurements of pan evaporation. With the emergence of water supply and food security issues as a result of increasing population and climate change pressures, the need for efficient use of available water supplies is paramount. Management of available resources and improved efficiency require accurate knowledge of evaporation, which is a major water loss pathway. This study used Autoregressive Integrated Moving Average (ARIMA) models to forecast pan-evaporation data. The historical data on pan evaporation (1982 – 2012) at Osogbo, southwest Nigeria, was initially subjected to a regression analysis which showed that the data has an increasing trend, while the plot of autocorrelation function indicated that the data is not stationary. Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), as well as diagnostics of residuals confirmed that ARIMA(3,4,3) is a good fit for both short term data forecast and data generation for pan-evaporation. Estimated long term reservoir evaporation series (2013 – 2062) was applied to the reservoir capacity curve and compared to the water demand curve. The results showed that with the increasing evaporation trend the reservoir will not be able to serve the various benefitting towns after year 2038. This implies that new water sources would be needed to meet the increasing water demand due to increasing population.