Nonlinear Approaches to Predicting Diameter of the Largest Limb at Breast Height in Young, Douglas-fir (Pseudotsuga menziesii (Mirbel) Franco) Plantations Growing in the Pacific Northwest

Abstract

This study examined two nonlinear modeling methodologies to predict the diameter of the largest limb at breast height (DLLBH) in young, Douglas-fir (Pseudotsuga menziesii (Mirbel) Franco) plantations in the Pacific Northwest using a combined power exponential function. The first method examined admitting predictor variables as additional factors in the power function portion and linear terms in the exponential portion. The second method used DBH as the base factor and estimated the parameters with simple linear functions of tree- and stand- variables. The second method proved superior as it simplified the model forms and preserved the allometric relationship between DLLBH and DBH. DLLBH was predicted with a residual standard error of .1627 and AIC of -13,398. Using DLLBH as a surrogate for product recovery, the final model was used in the context of process capability analysis, where thresholds were set on DLLBH and predicted cumulative distributions were generated in order to show how suppliers or purchasers might use the model in order check conformance of lumber to particular standards.