The comparative analysis of performance evaluation of recalibrated reference evapotranspiration models for different regional climatic conditions in Nigeria

Abstract

The study evaluated some radiation based models in four different climatic regions (Coastal, Savannah, Midland and Sahel) in Nigeria. The models (Abtews, Hargreaves, Makkink and Priestly Taylor) were  further calibrated for each of the region. This was done with a view to improving the accuracy of evapotranspiration (ET) estimates and determining suitable model(s) for the estimation of reference evapotranspiration for each region. The models were calibrated for each region by adjusting their respective coefficient across different climatic conditions in Nigeria on monthly timescale locations. The new regional adjusted coefficient include: Abtew (1.34,0.841, 1.072 and 10.637), Hargreaves (0.,026, 0.019, 0.024 and 0.022), Makkink (1.06, 1.12, 0.024 and 0.88) and Priestly and Taylor(3.62, 2.67, 2.63 and 2.65) for the Arid, Midland, Guinea savannah and Coastal regions respectively as were arranged in the parenthesis. These new adjusted coefficients were greater in the arid than the other region. The most suitable model determined and recommended for each region based on the highest coefficient of determination (R² ) and least standard error of estimate (SEE) were Abtew (R² = 0.88, SEE = 2 2 1.62), Makkink (R² = 0.80, SEE = 2.06) and Priestly Taylor (R² =0.80, SEE =2.06) for the Arid. In the Midland, the models were Priestly Taylor (R² =0.84, SEE=1.03) and Makkink (R² =0.84, SEE = 1.05). Priestly and Taylor (R² = 0.84, SEE=0.83 was the only suitable model found for Guinea Savannah while Priestly and Taylor (R = 0.84, SEE=0.47) and Hargreaves (R²= 0.84, SEE = 0.47) were the most suitable for the Coastal environment. The study concluded that the adjusted coefficient of the models for each region  performed better compared with the non-adjusted models.