An Improvement in the Relation Between PALSAR-2 Backscatter and Forest Stand Volume

• Main Authors: Hong, Min-Gee, Kim, Choen
• Format: Conference Proceeding
• Language: English
• Subjects: Backscatter; Biomass; DEM-adjusted Backscattering Coefficient; Dynamic range; Forest Stand Volume; Forestry; Gamma Nought; PALSAR-2; Remote sensing; Spatial resolution; Synthetic aperture radar
• ISSN: 2153-7003
• DOI: 10.1109/IGARSS.2018.8518914

This study was conducted to improve normalized backscattering coefficients of synthetic aperture radar (SAR) data for topographic effects reduction. The SAR data in the study is ALOS-2 PALSAR-2 with 6-m resolution for Kwangneung experimental forest (KEF) that was acquired on April 7, 2015. We analyzed the relationship between backscattering coefficients and forest stand volume based on the 6th national forest inventory (NFI) data. The following topographic correction method was used for accurate forest stand volume estimation. The potential for improving the accuracy of the stand volume estimation at the test site located at the KEF on the basis of PALSAR-2 data should be evaluated using the digital elevation model (DEM)-adjusted topographic normalized backscattering coefficients. In this context, two backscattering coefficients are of particular importance, namely the backscatters on flat terrain \gamma^{\circ} and on the tilted terrain \gamma_{\mathrm{s}}^{\ \circ} . In this study, the relation between \gamma^{\circ} and the forest stand volume that is derived using a 5-m DEM is compared to that between \gamma^{\circ} and the stand volume derived using a 30-m ASTER GDEM. Similarly, based on DEM resolution, the comparison of forest stand volume was conducted on basis of \gamma_{\mathrm{s}}^{\ \circ} . The coefficient of determination R2 for PALSAR-2 \gamma^{\circ} using the 30-m ASTER GDEM is observed to be smaller than that for PALSAR-2 \gamma^{\circ} based on the 5-m DEM by 0.038. Additionally, the coefficient of determination R 2 for PALSAR-2 \gamma^{\circ} is always observed to be lower than PALSAR-2 \gamma_{\mathrm{s}}^{\ \circ} . Therefore, it is inferred through comparison based on the observed standard deviation to the change in forest stand volume that \gamma_{\mathrm{s}}^{\ \circ} depicts higher sensitivity. Owing to an improvement in the spatial resolution, the accurate estimation of forest biomass that is derived from such a relation requires a 6-m DEM which is equivalent to the resolution of PALSAR-2.