Using LiDAR Data to Estimate Biomass in Afforested Bottomland Oak Sites in the Southern United States

Abstract

Abstract The extent of bottomland hardwood forests in the Lower Mississippi Alluvial Valley (LMAV) has diminished, and federal programs like the Conservation Reserve Program provide incentives to afforest marginal agricultural areas with oaks to provide ecosystem services. Remote sensing technologies, like light detection and ranging (LiDAR), can be used to estimate biomass of these stands to potentially allow landowners to take advantage of carbon markets, but data are expensive to collect. Therefore, we determined whether freely available low-density LiDAR data could capture variability in tree- and stand-level characteristics in the LMAV, including aboveground biomass. We found that multiple regression LiDAR models captured more variability in tree-level than stand-level parameters and including soil type generally improved models. Model r2 values predicting tree and stand parameters including tree height, height to the live crown, quadratic mean diameter, crown area, trees per hectare, stand basal area, and stand biomass ranged from 0.34 to 0.82 and root mean square percent error (RMSPE) ranged from 7% to 36%. Specifically, models for stand biomass had an RMSE of about 19 Mg/ha or about 19% of mean values across sites. Therefore, freely available LiDAR data was useful in evaluating afforested bottomland oak sites for tree- and stand-level structural components in the LMAV.