APPLICATION OF SATELLITE-BASED INDICES FOR PRECISE AGRICULTURAL LAND MANAGEMENT AND ESTIMATION OF EVAPOTRANSPIRATION

Abstract

Remote sensing has become the greatest data source for large-scale studies due to the availability of remotely sensed data from a variety of sensors on various platforms with a wide range of spatial and temporal resolutions. Landsat-8, Sentinel-2A, and Planet imaging data were used in this study to generate LULC maps, crop maps, and crop evapotranspiration (ETc) maps using machine learning methods. The outcomes of the first objective proved that using Landsat-8 and Sentinel-2A-based indices can reduce the need for ground truth data for LULC mapping. Without relying on satellite data, the results of this study also revealed that random forest (RF) is a superior classifier for LULC mapping than k-nearest neighbour (K-NN) and k-dimensional tree (KD-Tree). The results of the second objective showed that when multitemporal NDVI data was merged with multitemporal Planet imagery, the overall accuracy of crop maps using support vector machine (SVM) and decision tree (DT) increased. According to the findings of the third research, created ETc maps with a resolution of 3 m can aid farmers in precisely estimating water crop requirements