Retrieval of tropical peatland forest biomass from polarimetric features in central Kalimantan, Indonesia

In this research, the potential of L-band SAR data is evaluated for tropical peatland forest biomass estimation using polarimetric features and field data. For this, ALOS-2 full polarimetric data are acquired over central Kalimantan, Indonesia. Total 54 sampled plots (20 m × 20 m) were established in the study site; diameter at breast height (DBH) and tree species of every tree were collected in each plot. Locally developed allometric equations were used to convert field data to biomass and plot level biomass, and the upscaling factor was applied to upscale plot level biomass to standard tones per hectare scale. Backscattering coefficient (σo) was computed for HH,HV,V H and VV polarization. Similarly, eigen decomposition was performed to extract: entropy (E), alpha (α), and anisotropy (A); also diversity indices were computed. Yamaguchi decomposition was performed to extract scattering behavior of forest in central Kalimantan. All polarimetric parameters were upscaled to one-hectare scale. Field data were divided into training plots (70 percent → 42 plots) and validation plots (30 percent → 12 plots). Nonlinear regression analysis was performed between polarimetric parameters and training plots. Perplexity, Shannon index, entropy, Gini Simpson index, index of qualitative inversion, Reyni entropy (order 2), σHV, alpha, σVV, and volumetric scattering component were found significantly correlated (ranging R2 from 0.67 to 0.49) with the field data. The corresponding nonlinear model was inverted, and biomass maps were computed for the individual model. The resultant biomass maps were validated using a validation set of referenced measurements. Perplexity, Shannon index, entropy, Gini Simpson index, index of qualitative inversion, Reyni entropy (order 2), σHV, alpha, σVV,and volumetric scattering exhibited a significant correlation between field biomass and predicted biomass computed using developed model. R2 for validation ranges from 0.95 to 0.81 with RMSE ranging from 13.59 Mgha−1 to 25.63 Mgha−1. The estimated biomass in study site ranges from 49.31 Mgha−1 to 290.60 Mgha−1.