Evapotranspiration Prediction Using M5T Data Mining Method

Abstract

Evapotranspiration (ET) estimation takes an important role in hydraulic designs and irrigation management. Even these imperative importance ET estimation methods are not clear and easily employable enough. This study focused on M5T data mining method to estimate ET due this method is in use for nonlinear physical cases. 1543 daily Solar Radiation (SR), Air Temperature (AT), Relative Humidity (RH) and Wind Speed (U) meteorological parameters are used to create a M5T model. 1153 daily data is used for training the model and 385 left data is used for testing model results. Data set is taken from St. Johns, Florida, USA weather station.The correlation coefficient (R) is calculated as 0.983 for the M5T. Model results are compared with Turc empirical formula and it is found that M5T data mining method has better performance than Turc empirical formula.