Estimating variation in field crew estimates of site index

Abstract

Listen

Translate article

Variances of red maple (Acerrubrum L.) and yellow birch (Betulaalleghaniensis Britt.) site index estimates were determined using nonlinear mixed effects modeling procedures and simulations. Site index curves for each species were constructed using repeated height–age observations for individual trees and plot site index estimates. Plot site index was calculated by linearly interpolating between the height observations corresponding to age observations bracketing the index age for each tree and then averaging over trees. Nonlinear mixed effects modeling procedures were used to estimate the parameters of the site index curve model because of the longitudinal, repeated-measures nature of the height–age data. Plot site index variances were determined by simulating the procedure used by Forest Inventory and Analysis field crews of the USDA Forest Service North Central Forest Experiment Station. At least three site index trees per plot were necessary to consistently obtain 95% confidence intervals whose widths were within ±20% of means. Variation among site index trees, residual variation around estimated height–age curves, and errors in height measurement were the greatest contributors to variation in plot site index estimates. Because the effects of these sources of variation cannot be readily reduced, it is unlikely that the variances of site index estimates can be reduced.