Machine Learning Prediction of Evaporation in South-Western Part of Nigeria

Evaporation is a fundamental process in the hydrological cycle. Accurate prediction of evaporation is essential for effective water resource management, flood forecasting, and the development of adaptive strategies to mitigate the impacts of climate change. This work analyzes the capabilities of Seasonal Autoregressive Integrated Moving Average (SARIMA), Random Forest, and Support Vector Machine machine learning models in predicting evaporation in the south-western part of Nigeria. The results show that the SARIMA performs best in Ikeja, Akure, Ado Ekiti, Ibadan, and Osogbo. Its highest score (R2) was observed in Ibadan (0.9859) followed by Akure (0.9159), Ado Ekiti (0.8711), and Osogbo (0.8509). This suggests that 85% to 98% of the variability in evaporation in these locations can be explained by the meteorological inputs, which is good given the complexity and non-linearity of the atmospheric process. Moreover, we revealed a decreasing regional trend from 2001 to 2010, which suggests a decline in solar radiation and/or air temperature.