GP4F - A Gaussian Process Regression Model For Forest Biomass Retrieval Utilizing Simulated NISAR Data

In this study, we propose a Matérn-3/2 kernel-based Gaussian process regression model to estimate the above-ground biomass of an entire forest and its major forest types. We have utilized L-band full polarimetric simulated NISAR data and LiDAR height measurements in our present work. We evaluated the performance of the GPR model to estimate AGB within the range 8 to 100 Mg $\mathrm{ha}^{-1}$ at a spatial resolution of 100 m. Finally, we also demonstrate and discuss the AGB maps generated over the entire study area for two above-ground biomass ranges: (a) 8 to 100 Mg $\mathrm{ha}^{-1}$(b) 8 to 470 Mg $\mathrm{ha}^{-1}$.