Application of SAR data for oil palm tree discrimination

Abstract

Oil palm tree discrimination is an important step for tree counting. In order for proper planning and management of the plantation, it is important to identify and classify oil palm tree species distinctively from other tree species and weeds. As oil palm trees are green leafy plants, it is difficult to differentiate between oil palm trees and weeds through color classification alone. Tree height can be determined through proper photogrammetric method. However, it is time consuming. SAR data processing is becoming a promising technology in the field of geospatial. Backscatter coefficient value is influenced by the roughness of surface, type of target and moisture content of target. The aim of this research is to utilize L-band ALOS PALSAR-2 dataset and open source C-band Sentinel-1 SAR datasets to discriminate oil palm trees from weeds as there is significant height difference between them. This research determines the backscatter coefficient value range of oil palm trees and weeds, it investigates the suitability of utilizing C-band and L-band SAR data on oil palm tree discrimination. Several existing oil palm field parameters are tested based on the backscatter value of SAR image. The results and discussion of backscatter value ranges between oil palm tree species and weed species will be further discussed in the paper.