PARENT SESSION
Poster Session # 2: Statistical Ecology and Ecological Modeling.

Monday, August 4 Presentation from 5:00 PM to 6:30 PM. SITCC Exhibit Hall B.

Uncertainty in scaling up carbon estimates in peat soils of Finland: Comparison among four methods.

Li, Harbin^*,, Minkkinen, Kari , Wang, Zhengquan , Trettin, Carl,

ABSTRACT- Policy or decision making often requires extrapolating information across scales (i.e., scaling up) because most management problems are occurring at large scales, but much of our knowledge and data is accumulated at small scales at which research activities of most disciplines take place. In the process of scaling up, errors from data and models may inevitably get propagated into results. Uncertainty is a fundamental characteristic of modeling (or scaling) because errors, caused by incomplete data, limitations of models, and lack of understanding of underlying processes, exist in every aspect of modeling. Thus, uncertainty of large-scale estimates must be studied and quantified as an integral part of scaling to guarantee the adequacy and reliability of the scaling results. In this study, we calculate carbon estimates in peat soils of Finland based on data from intensive sampling at the plot level, quantify uncertainty of the estimates, and examine effectiveness of different uncertainty analysis techniques. The scaling algorithm used is the mathematical expectation with aggregated area data of vegetation types. The four methods of uncertainty analysis compared are: probability theory, Taylor series approximation, Monte Carlo simulation, and sequential partitioning. The key error sources considered are: peat depth, peat bulk density, peat carbon content, and total areas of vegetation types. The results suggest that uncertainty of the carbon estimates is high and must be reduced to be useful to policy-making and that peat depth and bulk density show the highest relative contributions to the uncertainty of the total carbon estimates. All four methods of uncertainty analysis work well in this relatively simple scaling exercise, indicating that, for complex situations, sequential partitioning is a promising method given that models can be divided into independent compartments to deal effectively with uncertainty assessment.

Key words: scaling, peatland , uncertainty, carbon