The US National Greenhouse Gas Inventory uses the component ratio method (CRM), a volume conversion approach that incorporates models for tree biomass components, for forest carbon assessments. However, the performance of the CRM relative to other methods, as well as influences on its accuracy and precision, must be evaluated. We constructed a data-driven CRM (n-CRM), used it to predict total tree and component biomass for six US tree species, and compared this approach to a reference allometric model. We also assessed the influence of size, crown dynamics, and stem growth on the performance of both methods. Results show that the n-CRM was more accurate for four species, resulting from the inclusion of more predictor variables. Both methods had high uncertainty, but the precision of n-CRM predictions was two to eight times higher for small diameter trees (10 cm) across all species. Accuracy and precision of the crown component models (i.e. branches and foliage) was low, though better for pines than for hardwoods. Species-level analysis suggests that poor precision is influenced by crown traits and the size distribution of fitting datasets. Our results highlight needed improvements to the n-CRM, and motivate further development of data that facilitate predictive evaluation of biomass models.
Bibliographical noteFunding Information:
Data compilation for the legacy data and the independent validation datasets, as well as B. Clough’s time, were funded by the USDA Forest Service Forest Inventory and Analysis Program, Northern Region. Additional funding and support was available from the Minnesota Agricultural Experiment Station. Part of D.W. MacFarlane’s time was supported with funds from Michigan AgBioResearch through the USDA National Institute of Food and Agriculture.