Modelling soil carbon sequestration of intensively monitored forest plots in Europe by three different approaches

Information on soil carbon sequestration and its interaction with nitrogen availability is rather limited, since soil processes account for the most significant unknowns in the C and N cycles. In this paper we compare three completely different approaches to calculate carbon sequestration in forest soils. The first approach is the limit-value concept, in which the soil carbon accumulation is estimated by multiplying the annual litter fall with the recalcitrant fraction of the decomposing plant litter, which depends on the nitrogen and calcium content in the litter. The second approach is the N-balance method, where carbon sequestration is calculated from the nitrogen retention in the soil multiplied with the present soil C/N ratio in organic layer and mineral topsoil. The third approach is the dynamic SMART2 model in combination with an empirical approach to assess litter fall inputs. The comparison is done by first validating the methods at three chronosequences with measured C pools, two in Denmark and one in Sweden, and then application on 192 intensive monitoring plots located in the Northern and Western part of Europe. Considering all three chronosequences, the N-balance method was generally most in accordance with the C pool measurements, although the SMART2 model was also quite consistent with the measurements at two chronosequences. The limit-value approach generally overestimated the soil carbon sequestration. At the intensive monitoring plots, the limit-value concept calculated the highest carbon sequestration, ranging from 160 to 978 kg ha−1 year−1, followed by the N-balance method which ranged from 0 to 535 kg ha−1 year−1. With SMART2 we calculated the lowest carbon sequestration from −30 to 254 kg ha−1 year−1. All the three approaches found lower carbon sequestration at a latitude from 60 to 70° compared to latitudes from 40 to 50 and from 50 to 60. Considering the validation of the three approaches, the range in results from both the N-balance method and SMART2 model seems most appropriate.