Khaled Ahmadaali - Assistant Professor, Department of Arid and Mountainous Regions Reclamation, University of Tehran, Karaj
AbdolMajid Liaghat - Professor, Department of Irrigation and Reclamation Engineering, University of Tehran, Karaj
Omid Bozorg Haddad - Professor, Department of Irrigation and Reclamation Engineering, University of Tehran, Karaj
Nader Heydari - Associate Professor, Iranian Agricultural Engineering Research Institute (AERI), Karaj

Iran is a mostly arid or semi-arid country and agriculture sector which plays an important role in its economy is the main user of its water resources. Lack of water is one of the significant problems in agricultural developement. Moreover, it is the most climate-dependent production sector therefore, accurate assessment of the climete change impact on agriculture for achieving its sustainability is a crucial task. Modelling virtual water can help in rasing the awareness of water scarcity and water resource planning and management. In this research, virtual water content (VWC) of Iran’s main crops (wheat, potato, corn, suger beet, barley, and alfalfa) were calculated. Then, efficiency of ANNs, SVM, ANFIS, K-NN as well as Geostatistical techiques in estimation of VWC using measured data of the yield and crop water requirement (CWR) and climatic factor was analysed. According to the R2 and RMSE statistics, it was proved that the K-Nearest Neighbour (K-NN) is the most accurate model in calculating the virtual water for all agricultural products. The obtained R2 and RMSE statistics by application of K-Nearest Neighbour (K-NN) were 0.96 and 96.96 for wheat, 0.90 and 157.58 for barley, 0.8 and 123.27 for corn, 0.95 and 113.98 for alfalfa, 0.80 and 132.27 for potato, 0.92 and 58.89 for sugar beet. Finally, based on the results of the best model, the average VWC of Iran’s major agricultural products scattering map were created and analyzed.

Virtual water, Artificial intelligent, Geostatistics, Climatic factors, modelling