Multi-polarized Radarsat-2 satellite sensor in assessing forest vigor from above ground biomass

Abstract

Accurate estimates of forest biomass are increasingly important in relation to sequestration of carbon by forest trees. Satellite remote sensing is a useful tool for biomass estimation and monitoring of forest ecological processes. Microwave synthetic aperture radar (SAR) can increase the accuracy of estimations of forest biomass in comparison to optical remote sensing, due to the unique capacities of SAR, including high penetrability, volumetric scattering, interaction with surface roughness, and dielectric property. We studied the potential of multi-polarized C-band Radarsat-2, a SAR technology, with HH, HV and VV polarization for estimating biomass of moist tropical Indian forest. Backscatter values are correlated with field-based biomass values and are regressed to generate models for estimating biomass. HH polarization provided maximum information regarding tree biomass. A coefficient of determination of 0.49 was calculated for HH polarized C-band image with in situ measurements. An exponential model was proved to be best suited for estimating forest biomass. Correlation of 0.62 and RMSE of 24.6 t ha−1 were calculated for the relationship between estimated and predicted biomass values for the best fit model. The average absolute accuracy of the model was 61%, while Willmott’s index of agreement was 0.87. Results suggest that most of the biomass of the area ranged within 70 t ha−1 a probably due to the saturation of C-band around 60–70 t ha−1 for tropical forests.