Estimation of Reference Evapotranspiration Using Support Vector Machines: a Case Study of Adana, Turkey.

Abstract

Evapotranspiration is an important parameter in hydrological and meteorological studies. Evapotranspiration forecasting plays an important role in irrigation management and hydraulic designs, especially during dry periods. In this study, average temperature (T), relative humidity (RH), wind speed (U), solar radiation (SR) parameters were used to estimate the daily evapotranspiration amount. Daily evapotranspiration estimation (ET0) was made according to the Penman-Monteith method recommended by FAO (Food and Agriculture Organization) as a standard method. The Penman-Monteith method was considered as the reference equation. Support Vector Machines (SVM) methods with four different input combinations were used to estimate the daily evapotranspiration of Adana province. SVM models were compared with each other and the reference equations’ results. According to the results obtained from SVM models, SVM3 model giave slightly better results according to the higher determination coefficient and lowest error data.

Keywords

• tropospheric no2
• sentinel-5p
• air pollution sources
• elevation
• romania