Uncertainty of inventory-based estimates of the carbon dynamics of Canada's managed forest (1990-2014)

摘要

Canada's National Forest Carbon Monitoring Accounting and Reporting System (NFCMARS) quantifies the carbon (C) dynamics and greenhouse gas (GHG) emissions and removals of Canada's managed forest to fulfill reporting obligations under international climate conventions. Countries are also requested to assess the uncertainty associated with these estimates, which we report here. We used Monte Carlo simulation to quantify uncertainty of carbon stock and flux estimates from the Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3), the core ecosystem model of the NFCMARS. We evaluated the impacts of model algorithms, parameters, and the input data used to describe forest characteristics and disturbance rates. Under our assumptions, 95% confidence interval widths averaged 16.2 Pg C (+8.3 and -7.9 Pg C, or +/- 15%) for total ecosystem C stock and 32.2 Tg C.year(-1) (+16.6 and -15.6 Tg C.year(-1)) for net biome production relative to an overall simulation median of -0.8 Tg C.year(-1) from 1990 to 2014. The largest sources of uncertainty were related to factors determining biomass increment and the parameters used to model soil and dead organic matter C dynamics. Opportunities to reduce uncertainty and associated research challenges were identified.