Gaps in available data for modeling tree biomass in the United States
When estimating tree-level biomass and carbon, it is common practice to develop generalized models across numerous species and large spatial extents. However, sampling efforts are generally incomplete and trees are not randomly selected. In this analysis, of the more than 1,000 biomass-related articles that were reviewed, trees were destructively sampled in over 300 studies to estimate biomass in the United States. Studies were summarized and past sampling efforts were explored to illuminate where the largest data gaps occurred in terms of tree components sampled, tree size, tree form, tree species, and location. The most prominent gaps were in large trees, particularly in Douglas-fir trees in the Pacific Northwest. In addition, tree roots were notably undersampled. Lastly, trees of poor or unusual form and low vigor were often not sampled, and this may introduce a systematic bias if not dealt with appropriately. More than 200 species did not have a biomass model or a single data point. The gaps presented here can be viewed as suggestions for future destructive sampling efforts, but the magnitude of a gap for a given model will ultimately depend on the selected modeling framework and the user's objectives.