Comparative Study of M5 Model Tree and Artificial Neural Network in Estimating Reference Evapotranspiration Using MODIS Products

Abstract

Reference evapotranspiration (ETO) is one of the major parameters affecting hydrological cycle. Use of satellite images can be very helpful to compensate for lack of reliable weather data. This study aimed to determine ETO using land surface temperature (LST) data acquired from MODIS sensor. LST data were considered as inputs of two data-driven models including artificial neural network (ANN) and M5 model tree to estimate ETO values and their results were compared with calculated ETO by FAO-Penman-Monteith (FAO-PM) equation. Climatic data of five weather stations in Khuzestan province, which is located in the southeastern Iran, were employed in order to calculate ETO. LST data extracted from corresponding points of MODIS images were used in training of ANN and M5 model tree. Among study stations, three stations (Amirkabir, Farabi, and Gazali) were selected for creating the models and two stations (Khazaei and Shoeybie) for testing. In Khazaei station, the coefficient of determination (R2) values for comparison between calculated ETO by FAO-PM and estimated ETO by ANN and M5 tree model were 0.79 and 0.80, respectively. In a similar manner, R2 values for Shoeybie station were 0.86 and 0.85. In general, the results showed that both models can properly estimate ETO by means of LST data derived from MODIS sensor.