Application of Land Surface Temperature Extracted from Satellite Images for Zoning Reference Evapotranspiration

Abstract

Reference evapotranspiration is one the most important climatic parameters to determine crop water requirement, climatic and hydrological studies and water resources management. The aim of this study was spatial estimating and zoning of reference evapotranspiration using the MODIS sensor images performed for Sefidroud basin. For this purpose, among the MODIS sensor products, Land Surface Temperature was selected and based on that, daily values of maximum and minimum air temperatures were modeled for studied area using Multiple Linear Regression and non-linear model of Support Vector Machines-based. Then, map of reference evapotranspiration zoning was prepared for studied area using Hargreaves-Samani method. Afterwards, by developing a linear regression model, the extracted reference evapotranspiration from Hargreaves-Samani method was converted to the FAO-Penman-Monteith method and based on that, map of reference evapotranspiration zoning on the basis of the FAO-Penman-Monteith method was extracted for the study area. Based on performed analyses, Support Vector Machine and Multiple Linear Regression models were selected to model minimum and maximum air temperatures, respectively and by applying cross validation method, the adjusted coefficient of determination for these models obtained was 0.81 and 0.92. The results showed that it is possible to make spatial estimation of reference evapotranspiration with an appropriate accuracy by considering an algorithm on the basis of combining satellite-based Land Surface Temperatures and statistical models. Zoning of reference evapotranspiration for any region can be extracted by only having some meteorological data in few point stations and by utilization from the satellite-extracted Land Surface Temperatures.

Keywords

• fao-penman-monteith
• regression model
• remote sensing
• water requirement