Analyzing The Uncertainty Of Diameter Growth Model Predictions

Abstract

The North Central Research Station of the USDA Forest Service is developing a new set of individual tree, diameter growth models to be used as a component of an annual forest inventory system. The criterion for selection of predictor variables for these models is the uncertainty in 5-, 10-, and 20-year diameter growth predictions estimated using Monte Carlo simulations. Particular attention is focused on the uncertainty in the model predictions that can be attributed to five sources of uncertainty: (1) residual variation after model calibration; (2) covariances for parameter estimates; (3) measurement error for diameter at breast height; (4) measurement error for crown ratio; and (5) measurement error for crown class. The cumulative effects uncertainty in model predictions for individual trees on the uncertainty of basal area estimates at the plot and regional levels are also investigated. For four Lakes States species. the results suggest that uncertainty in predictor variables has a non-negligible effect on the uncertainty of diameter predictions, but that these effects are attenuated when results are aggregated at the regional level.

Citation

McRoberts, Ronald E.; Lessard, Veronica C.; Holdaway, Margaret R. 1999. Analyzing The Uncertainty Of Diameter Growth Model Predictions. In: Proceedings, Society of American Foresters 1998 national convention; 1998 September 19-23; Traverse City, MI. SAF Publ. SAF 99-01. Bethesda, MD: Society of American Foresters: 319-324.