The Effects of Spatial Patterns on the Accuracy of Forest Vegetation Simulator (FVS) Estimates of Forest Canopy Cover

Abstract

Abstract The Forest Vegetation Simulator (FVS) estimates percent canopy cover without spatially explicit information. Estimates of canopy cover in FVS can be corrected for crown overlap, based on the assumption that trees in a stand are randomly distributed. This research assessed the accuracy of FVS estimates of canopy cover in stands with nonrandom spatial patterns. A method for measuring canopy cover within a geographic information system was developed to compare with FVS estimates of cover for 19, stem-mapped plots across Idaho and Montana. The Ripley's K(d) statistic was used to describe natural and simulated spatial patterns, so that the accuracy of canopy cover estimated by FVS could be considered for groups of plots classified as regular, clustered, or random. Results from the analyses of the effects of spatial patterns indicated that the FVS may underestimate canopy cover by 11% for plots with highly regular spatial patterns and overestimate by 2% for plots with clustered patterns. Although the magnitude of this bias likely is insufficient to justify changing the model's algorithm for canopy cover, users of this model should be aware of the potential bias that can occur as a result of assuming that trees in a stand are randomly distributed. Information on the general spatial pattern of the stand (i.e., clumped, random, and even) could be used by managers to anticipate the expected degree and direction of the bias.