Monday, November 13, 2006 - 2:15 PM
100-5

The Imputation of Forest Soil Carbon Content for Reporting Purposes.

Charles Perry and Barry T. Wilson. USDA Forest Service, 1992 Folwell Avenue, St. Paul, MN 55108

While several different forest soil databases exist, only the USDA Forest Service Forest Inventory and Analysis program is designed to collect and remeasure soils data on a short time interval. This is one component of its Phase-3 monitoring network. There is approximately one plot for every 39 000 ha of forest land. The utility of this data would be greatly improved if more of the landscape could be assigned a meaningful value. In this study, we used imputation techniques to estimate soil carbon storage in the top 10 cm of soil in the forested part of Minnesota and Wisconsin.  Nearest neighbor imputation is a commonly used method for dealing with incomplete survey data.  By examining forest structure attributes collected on Phase-2 plots, forest soils information is assigned to all plots from the data collected on Phase-3 plots.  Bootstrapping provides a method for producing population estimates and variances of forest soils attributes of interest from these imputed plots. We expect the precision of our population estimates to be improved by this approach.  Furthermore, the resultant map is of higher density than the original plot map and provides an improved representation of spatial distribution of forest soil carbon.