Tree-level imputation techniques to estimate current plot-level attributes in the Pacific Northwest using paneled inventory data

Abstract

The Forest Inventory and Analysis program (FIA) of the US Forest Service conducts a nationwide annual inventory. One panel (20% or 10% of all plots in the eastern and western United States, respectively) is measured each year. The precision of the estimates for any given year from one panel is low, and the moving average (MA), which is considered to be the default estimator, can result in biased estimates of current conditions. An alternative to the MA is sought, and studies comparing different alternatives to the MA approach for estimating current forest attributes in the Pacific Northwest are lacking. Paneled data from national forests in Oregon and Washington were used to explore nearest neighbor (NN) imputation methods to project all panels to a common point in time. When using the most recent ground measurements of the panels measured in prior years as ancillary data, tree-level NN imputation outperformed the MA estimator in estimating basal area/ha, stems/ha, volume/ha, and biomass/ha in terms of bias and root mean square error (RMSE) and plot-level NN imputation in terms of RMSE. When basal area/ha, stems/ha, volume/ha, and biomass/ha were summarized by three species groups, tree-level NN imputation outperformed plot-level NN imputation in terms of both bias and RMSE. Tree-level NN imputation outperformed the MA in terms of bias and RMSE for estimating basal area/ha, stems/ha, volume/ha, and biomass/ha for species group 'pine' and provided comparable results in terms of bias and RMSE for species groups 'Douglas-fir' and 'other.'

Parent Publication

• Forest Inventory and Analysis (FIA) Symposium 2008; October 21-23, 2008; Park City, UT

Citation

Eskelson, Bianca N. I.; Hailemariam, Temesgen; Barrett, Tara M. 2009. Tree-level imputation techniques to estimate current plot-level attributes in the Pacific Northwest using paneled inventory data. In: McWilliams, Will; Moisen, Gretchen; Czaplewski, Ray, comps. Forest Inventory and Analysis (FIA) Symposium 2008; October 21-23, 2008; Park City, UT. Proc. RMRS-P-56CD. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 13 p.