Measurements and models – a comparison of quantification methods for SOC changes in forest soils

The Swedish UNFCCC1-reporting of the LULUCF2-sector is based on methods in compliance with the “Good practice” as described by the Intergovernmental Panel on Climate Change (IPCC). Biomass and soil inventory data from the Swedish Inventory of Forests is the major source of information used to quantify changes in the various carbon pools on forest land. Even if the reported uncertainties in soil carbon changes are small from a statistical perspective, they are large in relation to the total Swedish emissions of green house gases. This is due to the fact that the soil carbon pool is so large, that even small and statistically non-significant changes may have an impact on the Swedish CO2 balance. Sampling based methods may also result in considerable inter annual variations that may look conspicuous in the reporting. Because of the uncertainty and inter annual variations there has been a discussion on the methods used and if there are possibilities to lower the uncertainty and to get more stable estimates of soil carbon changes by combining measurements and models. In this study results from the two soil carbon models, Yasso07 and Q, were compared with repeated measurements of the soil inventory during the years 1994 to 2000. Soil carbon fluxes were simulated with the two models from 1926 to 2000 with Monte Carlo methodology to estimate uncertainty ranges. The results from the models agreed well with measured data. The simulations of Yasso07 and Q resulted in a soil organic carbon stock in year 2000 of 1600 Mton C and 1580 Mton C, respectively while the measured carbon pool was ¨1670 Mton C. The annual change in soil organic carbon varies substantially between the three methods mainly due to different assumptions regarding annual climate variation. However, the five year averaged mean of annual soil organic carbon change for the two periods 1994-1998 and 1996-2000 indicate the size and direction of the estimated annual changes agree reasonable well. The mean annual cha