Evaluation of temperature and radiation based reference evapotranspiration methods over Cross River basin Nigeria

Abstract

The accurate estimation of reference evapotranspiration (ETo) on seasonal and annual scale is vital for determination of water requirement of crops and impact of climate change on irrigated agriculture. This study investigates the influence of climatic variables (temperature, relative-humidity, solar-radiation, and wind speed) on ETo and comparison of the performance evaluation of temperature and radiation-based models over Cross River basin on seasonal and annual basis. Estimation of ETo was done using temperature-based; Schendel, Samani, Trajkovic, Droogers & Allen-2, Dorji, Hadria, Hargreaves-samani, Blaney-Morin-Nigeria, and radiation-based models; Jensen-Haise, Stephens and Stewart, Oudin, Abtew, Irmak-1, Copais, Tabari & Talaee-4, and Hargreaves under humid-tropic condition, with Penman–Monteith (PM-ETo) as a reference. Remotely-sensed meteorological variables from 1987 to 2017 were sourced from the Climatic Research Unit (CRU) database, over 22 stations. These variables were used for estimating ETo. Models were evaluated with coefficient of determination, a root-mean-square-error, Willmott’s index of agreement, Percentage-Bias and Nash–Sutcliffe Efficiency. Sensitivity analysis (± 5%) revealed that PM-ETo was most sensitive to solar-radiation and temperature, compared to relative-humidity and wind speed. Blaney-Morin-Nigeria demonstrated seasonal estimation bias. Further analysis revealed that radiation-based models out-performed the temperature-based models across all categories. Blaney-Morin-Nigeria and Copais recorded the best performance in dry season, while Hargreaves–Samani and Abtew recorded the best performance in rainy season. Abtew and Hargreaves–Samani recorded the best performance on annual basis. An increasing trend (slope = 0.0029 mm/day) of PM-ETo further suggests global warming scenario. This demonstrates the ability of radiation-driven models for crop water requirement estimation in data-scarce regions.