Comparison of Time Series Methods and Artificial Neural Networks In Reference Evapotranspiration Prediction (Case Study: Urmia)

Document Type : Research Paper

Authors

Abstract

     Evapotranspiration is one of the important factors in water resources consumption in the agriculture part. Therefore, presenting a method which gives suitable and accurate prediction of reference evapotranspiration can help to take optimum decision for water resource programing. In this research, time series and artificial neural networks methods were compared each other in order to predict the monthly reference evapotranspiration in Urmia synoptic station. To achieve this goal, at the first step, the best time series model between AR and ARMA models and the best artificial neural networks model between radial basis function (RBF) and multilayer perceptron (MLP) neural networks were selected. In the second step, the two models chosen were compared each other. In the mentioned artificial neural networks, the deferent monthly lags of reference evapotranspiration were used as network input. In this process, the monthly reference evapotranspirations were computed from 1971 to 2010 using FAO Penman-Monteith method. The mentioned dates from 1971 to 2005 were used to select the best time series model and the best structure of networks and the dates from 2006 to 2010 were utilized to compare the methods used. The results showed that the AR(11) model has the best performance among other time series models and the RBF model has the lower error than the MLP model. The comparison of the best time series model (AR(11) model) with the best artificial neural networks model (RBF model) showed that the RBF model could predict the reference evapotranspiration by the lowest error from 1971 to 2010 period. The root mean square error in AR(11) and RBF models was obtained 1.85 and 0.999 mm/month respectively. 

Keywords

Irrigation Sciences and Engineering
Volume 38, Issue 4 - Serial Number 4
February 2016
Pages 75-85

Files

History

  • Receive Date: 16 July 2013
  • Revise Date: 08 February 2016
  • Accept Date: 29 June 2014
  • Publish Date: 21 January 2016

Share

How to cite

Statistics