Spatial-temporal evaluation of different reference evapotranspiration methods based on the climate forecast system reanalysis data

Evapotranspiration is a major component of the interaction between land-surface processes and the atmosphere. Climate Forecast System Reanalysis (CFSR) data offer a promising database for overcoming the limitations in availability and reliability of climatological data and, hence, for understanding the evapotranspiration process. Using these data on grid-by-grid daily, seasonal and yearly scales, the present study attempts to advance the spatio-temporal evaluation of two radiation-based and three temperature-based methods for estimating potential evapotranspiration (PET) against estimates of grass reference evapotranspiration (ETo) by FAO Penman-Monteith method (FAO-PM). The analysis was performed for the period 1979-2013, considering the second largest (79 000 km(2)) river system in Ethiopia, that is, Omo-Gibe basin, which accommodates national parks and vast hydropower, cultivation and afforestation developments and discharges its flow to Lake Turkana in Kenya. Despite the large regional variations in climate and elevation, the results in overall emphasize the outperformance of the simple temperature method, viz. Hargreaves-Samani method, in capturing both the annual and seasonal FAO-PM estimates. Calibration of the Hargreaves-Samani equation is, however, a requisite for spectacular improvement of its performance. Accordingly, new coefficients of the equation are proposed. The annual trends in the basin's ETo increased with rising temperature and decreasing relative humidity, wind speed, and solar radiation, but with decreasing (increasing) rainfall in the upper region (the middle and lower regions). It is deduced that trends in simple methods do not necessarily reflect the true trends in ETo. Annual ETo decreases with increasing elevation and annual rainfall. The present findings are discussed in the context of a worldwide literature, thereby improving the understanding of the best performing PET methods in similar data-scarce national or transboundary rivers basin in Ethiopia, the region or worldwide. The wider implications regarding water loss from reservoirs and the rain-fed food and sugar production in the basin under study are also highlighted.