B43D-0285:
Uncertainty in accounting for carbon accumulation following forest harvesting
Thursday, 18 December 2014
Ruth D. Yanai, SUNY College of Environmental Science and Forestry, Syracuse, NY, United States, Paul Lilly, Spatial Informatics Group, LLC, Oakland, CA, United States, Mary A Arthur, University of Kentucky, Lexington, KY, United States, Kikang Bae, KFRI Korea Forest Research Institute of the Korea Forest Service, Seoul, South Korea, Steven Hamburg, Environmental Defense Fund Boston, Boston, MA, United States, Carrie R Levine, University of California Berkeley, Berkeley, CA, United States and Matthew A Vadeboncoeur, University of New Hampshire, Durham, NH, United States
Abstract:
Tree biomass and forest soils are both difficult to quantify with confidence, for different reasons. Forest biomass is estimated non-destructively using allometric equations, often from other sites; these equations are difficult to validate. Forest soils are destructively sampled, resulting in little measurement error at a point, but with large sampling error in heterogeneous soil environments, such as in soils developed on glacial till. In this study, we report C contents of biomass and soil pools in northern hardwood stands in replicate plots within replicate stands in 3 age classes following clearcut harvesting (14-19 yr, 26-29 yr, and > 100 yr) at the Bartlett Experimental Forest, USA. The rate of C accumulation in aboveground biomass was ~3 Mg/ha/yr between the young and mid-aged stands and <1 Mg/ha/yr between the mid-aged and mature stands. We propagated model uncertainty through allometric equations, and found errors ranging from 3-7%, depending on the stand. The variation in biomass among plots within stands (6-19%) was always larger than the allometric uncertainties. Soils were described by quantitative soil pits in three plots per stand, excavated by depth increment to the C horizon. Variation in soil mass among pits within stands averaged 28% (coefficient of variation); variation among stands within an age class ranged from 9-25%. Variation in carbon concentrations averaged 27%, mainly because the depth increments contained varying proportions of genetic horizons, in the upper part of the soil profile. Differences across age classes in soil C were not significant, because of the high variability. Uncertainty analysis can help direct the design of monitoring schemes to achieve the greatest confidence in C stores per unit of sampling effort. In the system we studied, more extensive sampling would be the best approach to reducing uncertainty, as natural spatial variation was higher than model or measurement uncertainties.
