Modeling aspen cover type diameter distributions in Minnesota
This article is part of a larger document. View the larger document here.Abstract
An attempt was made to model diameter distributions of aspen (Populus spp) stands. The aspen (1,020,150 acres) cover type has the greatest acreage on DNR lands and it contains a variety of hardwood and softwood species. Aspen is a valuable pulpwood species, annually comprising around 50 percent of total timber harvest on DNR lands.
Modeling distributions of this cover type has been minimal; stands often contain a variety of tree sizes and species, making modeling difficult, particularly given a disproportionate amount of smaller trees. For this analysis, a three-parameter Weibull-based modeling approach was used, where the parameters are predicted using the 0, 25th, 50th, and 95th estimated percentiles of the distribution. The percentiles are predicted as functions of quadratic mean diameter and stand/plot age. Currently, species compositions were ignored.
Data used in model fitting were obtained from the Forest Inventory and Analysis (FIA) database and from all regions of the state.
Prediction errors are rather large for smaller diameter classes and onset of merchantability (e.g. 4 to 6 inches). When looking at the average error across all diameter classes at a particular site, each diameter class within a site had an average error of 20 trees per acre (smaller diameters generally had larger absolute tree per acre errors).
For common merchantable rotations (e.g. 40 to 50 years), total errors (total across all diameter classes) are less relative to younger and older ages. However, for diameter classes of most interest (6 to 12 inches), prediction errors are generally larger for these rotation ages relative to younger stands.