Accuracy and Precision of 10 Year Predictions for Forest Vegetation Simulator—Lake States

Abstract

Abstract This study is a followup to the 5 yr validation of the Lake States TWIGS (The Woodsman's Ideal Growth Projection System) projection system by Guertin and Ramm (1996). Accuracy and precision of 10 yr diameter growth, basal area growth and mortality predicted by the Lake States variant of the Forest Vegetation Simulator (FVS) were evaluated for seven upland hardwood species in Michigan's northern Lower Peninsula. The robustness of FVS predictions was examined by varying projection cycle length and the level of detail of stand and tree-information included in growth projections.The data used in the analysis consisted of individual tree measurements from 44 stands across 10 ecological land type phases in the Manistee National Forest. FVS-Lake States was found to consistently overpredict 10 yr diameter growth across all seven species. Ten year diameter growth was predicted within ±0.5 in. across all projections for nearly all species and size-class combinations for the seven species examined. Basal area and mortality errors were less consistent. Mean errors for trees per acre ranged from -24 for red maple to +14 for white oak. These errors led to a consistent overprediction of basal area per acre for all species combined, while prediction errors for individual species were less than ±8 ft²/ac. Precision was variable, especially for mortality predictions. The most accurate predictions were obtained with longer cycle lengths and with projections using tree diameter, tree height, and crown ratio along with site index and individual tree past diameter growth. North. J. Appl. For. 17(2):62-70.