• Russell, Matthew B.
  University of Minnesota
  russellm@umn.edu
  
• Weiskittel, Aaron R.
  University of Maine
  aaron.weiskittel@maine.edu
  
• Kershaw Jr., John A.
  University of New Brunswick
  kershaw@unb.ca
  

This study used permanent sample plot data from the USDA Forest Service’s Forest Inventory and Analysis database to benchmark and calibrate three key submodels of the Forest Vegetation Simulator-Northeast variant (FVS-NE). Model predictions for total tree height (ht) and 5-year diameter (Δbh5) and height increment (ht5) for the 20 most abundant species did not indicate any serious spatial trends. FVS-NE predictions for total ht performed moderately well, as mean bias averaged -0.9 ± 5.2 ft (mean ± SD) across all species. FVS-NE dbh5 predictions fell within 15% of observed values between 8.4 and 17.3% of the time and performed best for shade-tolerant species and worst for intermediate shade intolerants. For Δht5, the number of predictions that fell within 15% of observed values averaged 7.7%. Submodel performance generally improved after calibrating FVS-NE predictions using tree size, site, and climate variables. After employing a calibrated Δbh5, 5-year basal area growth continued to be underpredicted across all ecoregions and forest types. Results indicate that (1) an assessment of overall model performance should be conducted if calibrated submodels are used and (2) alternative modeling strategies be explored to better represent the allometry and growth of the important trees species across the northeastern United States.