Phenology Model of Oil Palm Plantation Based on Biophysical Parameter on Sentinel-1A Using Multiple Linear Regression (MLR)

Estimating the biophysical parameters during the phenology cycle are very important and the key parameter for indicating the productivity of oil palm plantations. In many countries, the oil palm plantation has a very large area, therefore remote sensing technology is needed to estimate biophysical parameters in large areas. The special characteristics and potential of Synthetic Aperture Radar (SAR) data in acquiring geometric and dielectric properties of biophysical parameters have led to their identification in the context of vegetation monitoring. This study, we have investigated and developed models for estimating the oil palm phenology by applying multiple linear regression (MLR). The methodology includes the biophysical parameters estimated using Sentinel-1A for extracting the canopy height model (CHM), radar vegetation index (RVI), backscattering on VV and VH, aboveground biomass, texture entropy, and texture energy. Then applied multiple linear regression (MLR) analysis for developing model and assess  its ability. The result found the best model for estimating oil palm phenology using 4 parameters. The parameters are CHM, RVI, Backscatter on VV, Backscatter on VH and the best model for estimating oil palm phenology is \(MLR=38.839+0.984*{CHM}_{i}+(-97.214)*{RVI}_{i}+2.476*{VV}_{i}\) $MLR=38.839+0.984\ast {\mathit{CHM}}_i+(-97.214)\ast {\mathit{RVI}}_i+2.476\ast {\mathit{VV}}_i$ + (-0.893) \(*{VH}_{i}\) $\ast {\mathit{VH}}_i$ with R² is 0.977 and RMSE is 1.290.