Sky Journal of Agricultural Research Vol. 4(6), pp. 114 - 122, September, 2015 Available online http://www.skyjournals.org/SJAR ISSN 2315-8751 ©2015 Sky Journals Full Length Research Paper Modeling of daily reference evapotranspiration using climatic factors for arid regions of Algeria Laaboudi A. 1 , Mouhouche B. 2 and Slama A. 3 1 National Institute of research in agronomy of Algeria. Experimental station of Adrar, Algeria. 2 National high school of agronomy. El Harrach, Algers, Algeria. 3 Sustainable development and informatics laboratory, University of Adrar, Algeria. Accepted 5 August, 2015 There are different methods to predict reference evapotranspiration (ETo). In recent decades, Artificial Neural Network (ANN) has shown very high ability to deal with non linear process. In this study, ANN models were developed to estimate daily reference evapotranspiration in arid regions of Algeria based on explanatory climatic factors. These factors have been used as inputs, and ETo values computed by the Penman-Monteith formula have been used as outputs. Different combinations have been performed and different models have been developed. The performance of the different models was evaluated by comparing the corresponding values of determination coefficient (R²), root mean squared error (RMSE), meanabsolute error (MAE) and mean absolute relative error (MARE). The drawn conclusions have shown that model with temperature, relative humidity and wind speed as inputs showed high prediction accuracy confirming that the ANN models could provide ETo estimation very close to the reality. Thus, with a network of two hidden layers and twenty (20) neurons per layer, we obtained, during the test phase, values of 0.998, 0.482 (mm/day)², 0.096 (mm/day) and 0.034% respectively for R², RMSE, MAE, and MARE. Key words: Reference evapotranspiration, arid regions, Penman-Monteith formula, artificial neural network models. INTRODUCTION Reference evapotranspiration (ETo) is an important parameter for computing the irrigation demands of various crops (Chowdhary and Shrivastava 2010; Dinpashoh 2006). Current irrigation scheduling is based on a well-established crop coefficient and on ETo procedures to estimate daily crop evapotranspiration (Hunsaker et al., 2007). Thus, improper ET calculation appears to be associated with poor estimation of ETo (Shujiang et al., 2009). Therefore, in order to estimate ETo, much research has been carried out across the world, and a significant number of equations have been developed. But comparison of their results reveals a wide divergence between them (Smadhi, 2000; Lu et al., 2005). Food and Agriculture Organization (FAO) of the United Nations has recommended the Penman-Monteith method known for its prediction accuracy for varied *Corresponding author. E-mail: Laaboudiark@yahoo.fr. environments (Allen et al., 1998).The fundamental obstacle to widely applying the P-M method is the numerous input parameters that are not always available at many locations. To overcome this obstacle, Artificial Neural Network (ANN) model was employed to estimate daily reference evaporation based on three explanatory climatic factors. The Artificial Neural Network (ANN) modeling is a nonlinear statistical technique that can be used to solve problems that are not amendable to conventional statistical or mathematical methods (Eslamian et al., 2008). The application of up-to-date ANN technology allows modeling by black-box NN tools (Aytek et al., 2009).Usually, ANNs are trained so that a particular set of input produces, as closely as possible, a specific set of target outputs (Dechemi et al., 2003). The interest of neurons (Dreyfus et al., 2004) lies in the properties resulting from their association in networks. The combination of nonlinear functions performed by each neuron and their ability to glean internal information from