International Science Community Association Research Journal of Recent Sciences 2277 - 2502 5 11 2016 11 2 Artificial Neural Network for Predicting reference Evapotranspiration under Humid Region 25 31 EN Javier C. Ramirez-Perez 1 Departmento de Química, Facultad de Biología, Universidad Técnica Particular de Loja, Ecuador and Department of Chemistry, Kent State University at Stark, North Canton, OH 44720, USA S.B. Mohapatra 1 School of engineering & Technology, Centurion University, Jatni-752050, Bhubaneswar, Odisha, India D. Swain 2 School of engineering & Technology, Centurion University, Jatni-752050, Bhubaneswar, Odisha, India R.C. Mohanty 3 School of engineering & Technology, Centurion University, Jatni-752050, Bhubaneswar, Odisha, India P. Das 4 Founder of Science Foundation for Rural and Tribal Resource Development, Odisha, India Bhange H.N. 1 Dept of SWE, CTAE, Udaipur, Rajasthan, India Singh P.K. 2 Dept of SWE, CTAE, Udaipur, Rajasthan, India Purohit R.C. 3 Dept of SWE, CTAE, Udaipur, Rajasthan, India Yadav K.K. 4 Dept of SWE, CTAE, Udaipur, Rajasthan, India Ingle P.M. 1 Dept. of Soil and Water Engineering, CTAE, MPUAT, Udaipur, Rajasthan, India Purohit R.C. 2 Dept. of Soil and Water Engineering, CTAE, MPUAT, Udaipur, Rajasthan, India Bhakar S.R. 3 Dept. of Soil and Water Engineering, CTAE, MPUAT, Udaipur, Rajasthan, India Mittal H.K. 4 Dept. of Soil and Water Engineering, CTAE, MPUAT, Udaipur, Rajasthan, India Jain H.K. 5 Dept. of Agri. Statistics and computer application RCA, Udaipur, Rajasthan, India Singh P.K. 6 Dept. of Soil and Water Engineering, CTAE, MPUAT, Udaipur, Rajasthan, India > Research Paper 2016 9 16 2016 11 2 Artificial neural network (AN) was used to assess the reference evapotranspiration under missing or limited climatic parameters as input variables. The climatic data from year 1991-2014 i.e. 24 years was used for study. The results indicated that temperature based ANN architecture 2-2-1 and 3-4-1 found suitable for estimation of reference evapotranspiration under humid conditions. For mass based ANN model 4-4-1 and 5-4-1 architectures found appropriate for forecasting of evapotranspiration. The ANN architecture 6-2-1 gives good outcome than other architectures when all climatic variables were considered in the input layers. The study found that the different ANN architectures may be used under limited or missing data conditions. Temperature based, mass based and combination based models can be used for estimation of evapotranspiration by selecting the ideal nodes in hidden layer. Copyright@ International Science Community Association