Neural Network Approach for Estimating Reference Evapotranspiration from Limited Climatic Data in Burkina Faso

摘要

The well known Penman-Monteith (PM) equation always performs the highest accuracy results of estimating reference evapotranspiration (ETo) among the existing methods is without any discussion. However, the equation requires climatic data that are not always available particularly for a developing country such as Burkina Faso. ETo has been widely used for agricultural water management. Its accurate estimation is vitally important for computerizing crop water balance analysis. Therefore, a previous study has developed a reference model for Burkina Faso (RJVIBF) for estimating the ETo by using only temperature as input in two production sites, Banfora and Ouagadougou. This paper investigates for the first time in the semiarid environment of Burkina Faso, the potential of using an artificial neural network (ANN) for estimating ETo with limited climatic data set. The ANN model employed in the study was the feed forward backpropagation (BP) type using maximum and minimum air temperature collected from 1996 to 2006. The result of BP was compared to the RMBF, Hargreaves (HRG) and Blaney-Criddle (BCR) which have been successfully used for ETo estimation where there is not sufficient data. Based on the results of this study, it revealed that the BP prediction showed a higher accuracy than RMBF, HRG and BCR. The feed forward backpropagation algorithm could be potentially employed successfully to estimate ETo in semiarid zone.