COMPARISON OF PREDICTION MODELS OF REFERENCE CROP EVAPOTRANSPIRATION Slaviša Trajković

Abstract

The prediction of evapotranspiration is necessary for a reliable management ofirrigation systems. This paper is based on several models used for the prediction ofreference crop evapotranspiration in the area of Niš, Yugoslavia. Two simplemathematical models (Yearly Differencing model and Monthly AVerage model), in whichprediction is based on appropriate previous values of reference crop evapotranspiration,are present here. A seasonal ARIMA (AutoRegressive Integrated Moving Average) modelis identified and compared to the previous models. Based on the results of modelcomparisons it is concluded that seasonal ARIMA can provide a reasonably accurateprediction of reference crop evapotranspiration. The obtained results show that theseasonal ARIMA model is a very effective and reliable prediction model.