Synergistic use of very high-frequency radar and discrete-return lidar for estimating biomass in temperate hardwood and mixed forests

Abstract

Accurate estimation of aboveground biomass is essential to better understand the carbon cycle in forest ecosystems. The objective of this study was to determine whether biomass in temperate hardwood forests is better estimated using very high-frequency radar data (from BioSAR) alone or in combination with small-footprint discrete-return lidar data (both profiling and scanning). The study area was in the Appomattox-Buckingham State Forest, Virginia, USA (78°41′W, 37°25′N). Aboveground biomass for 28 stands was estimated using 131 basal area factor 10 point samples. The resulting stand biomass estimates were used as the dependent variable in a multiple linear regression. Descriptors of the lidar distributions (both profiling and scanning) and averaged normalized radar cross-sections in each of these stands were used as independent variables. Regression results revealed the following: (1) neither BioSAR nor scanning lidar data alone are good predictors of stand biomass (R 2 = 0.57, root mean squared error (RMSE) = 31.0 tonnes/ha and R 2 = 0.64, RMSE = 28.5 tonnes/ha, respectively); (2) BioSAR data combined with small-footprint discrete lidar data (either profiling or scanning) are the best predictors of stand biomass (R 2 = 0.80, RMSE = 21.3 tonnes/ha and R 2 = 0.76, RMSE = 24.2 tonnes/ha, respectively); and (3) when used with BioSAR data for stand biomass estimation, less costly profiling lidar data convey the same information as more costly scanning lidar data. Useful synergy can be realized by considering lidar and radar measurements jointly in estimating aboveground biomass in hardwood and mixed forests.