Computational performance of reference evapotranspiration in semiarid zone of Africa

Evapotranspiration is a major component of hydrologic cycle and its accurate estimation is essential for agricultural water management. The Penman-Monteith (PM) equation is the universal accurate method for estimating reference evapotranspiration (ET_ref). Its drawback is the large climatic data required which are unavailable in many African semiarid regions such as Burkina Faso. The Hargreaves (HRG) conventional method which requires few data is still used despite of its non-universal accuracy often reported due to the model inability to capture the effect of some important climatic parameters. Therefore, this study assessed the performance of an artificial neural network (ANN) for computing ET_ref in Dedougou region, located in the Soudano-Sahelian zone of Burkina Faso. This study employed ANN and HRG models in order to evaluate their performance by comparing with the true PM. From the statistical comparison results, ANN showed a good performance than HRG which overestimated ET_ref for the observed condition. Furthermore, wind speed has been found as an important factor in ANN accuracy improvement. Using ANN under semiarid zone climatic condition of Africa for computing ET_ref is highly superior to the conventional method.