Using Geostatistical Techniques to Map The Distribution of Tree Species From Ground Inventory Data

Abstract

Some kinds of information on biodiversity can only be collected from on-the-ground surveys, providing informtaion complementary to that gained by remote sensing. But a ground inventory, particularly over large aras, is expensive and labor intensive. Using geostatistical techniques, we can use known values at sample points to estimate the values of unsurveyed areas in between. By explorign characteristics such as the spatial pattern and variation in the sample data collected, additional information about the unsurveyed aras can be spatially modeled and estimated. From these estimates, a modeled map and dataset of the resource can be created, complete with understandable parameters of accuracy and uncertainty. The importance of identifying the objectives, understanding the phenomena, choosing the appropriate interpolation method and assessing the results is discussed, and some of the tools available to address each of these areas are presented. Ecological subregions are suggested as appropriate units for identifying species subpopulations for modeling and interpolation. Tree-species composition in Pennsyulvania is used as the example, using data from the U.S. Department of Agriculture national forest inventory. The resulting modeled dataset provides a picture of individual tree-species distribution and relative abundance across the entire region.