A LiDAR biomass index-based approach for tree- and plot-level biomass mapping over forest farms using 3D point clouds

Spatially continuous mapping forest aboveground biomass (AGB) is crucial for better understanding the capacities of carbon sequestration capacities of forest ecosystems at both individual tree and landscape levels. Collecting field data is one of the most labor-intensive and time-consuming tasks in biomass mapping using airborne laser scanning (ALS) data. Building on a LiDAR biomass index (LBI) developed for use with terrestrial laser scanning (TLS) data, we successfully developed an improved and robust LBI-based approach to estimate forest AGB at both individual tree and plot levels while minimizing the effort required for field data collection. This approach was tested for larch, birch, and eucalyptus over three forest farms in Northeast China and one in Southern China. The results showed that LBI was highly correlated with the diameter, height, and AGB of larch trees. AGB estimates derived using LBI-based models for the three tree species were close to ground measurements at the individual tree level. They explained 81% to 95% of the variance of independent test data not used to calibrate those models. Tree level AGB estimates are required by many applications, but they could not be provided by commonly used plot-based biomass mapping approaches like LiDAR metrics-based regression (LMR) or Random Forest (RF). Calibrated with small fractions of the trees needed to calibrate LMR and RF models, LBI-based biomass models produced plot level biomass estimates comparable to or better than those produced using the two plot-based methods. More importantly, the LBI-based models generalized far better than LMR and RF among the three larch forest farms located hundreds of kilometers apart. These promising results warrant more research on the effectiveness of the LBI-based approach for other forest types and tree species not considered in this study. As LiDAR technology and related algorithms are evolving rapidly, further improvements to this approach might be feasible. A robust LBI-based approach applicable to a wide range of tree species and forest types across the globe will greatly facilitate the use of increasingly better and more affordable ALS data to support REDD+ (Reducing Emissions from Deforestation and Forest Degradation) and other forest-based climate mitigation initiatives.