Classifying the smallholders’ oil palm tree conditions on peatland area using ALOS-2 PALSAR-2 data

- A case study in Riau Province, Indonesia -

Abstract

The oil palm cultivation on peatland area requires particular peat management techniques, such as compaction and drainage, to maintain water table conditions and minimize the leaning oil palms. However, these managements are hardly ever applied by smallholders, which cause frequent leaning oil palms or even toppling trees. This study attempted to explore the ALOS-2 PALSAR-2 full polarimetry data to classify the normal, replanted, and leaning oil palm conditions in three representative areas of smallholders’ plantations on peatland in Riau, Indonesia. Texture analysis was applied to an individual and combination of backscatter, radar vegetation index, and polarimetric parameters derived from decomposition. The results indicated that mean feature produced the optimum classification among the other texture features, while the polarimetric parameters are particularly useful for characterizing tree stand conditions of leaning oil palms. The most effective combination is the mean texture analysis of 15-bands Synthetic Aperture Radar (SAR) parameter in a 7 × 7 window size with an overall accuracy of 68.69% (Kappa statistic: 0.60), 70.18% (0.55), and 72.75% (0.60) in areas 1, 2, and 3, respectively. This study demonstrated a step forward for the methodology of classifying oil palm conditions using single data of PALSAR-2.