A dynamic model of wetland extent and peat accumulation: results for the Holocene

Abstract

Abstract. Substantial deposits of peat have accumulated since the last glacial. Since peat accumulation rates are rather low, this process was previously neglected in carbon cycle models. For assessments of the global carbon cycle on millennial or even longer timescales, though, the carbon storage in peat cannot be neglected any more. We have therefore developed a dynamic model of wetland extent and peat accumulation in order to assess the influence of peat accumulation on the global carbon cycle. The model is based on the dynamic global vegetation model LPJ and consists of a wetland module and routines describing the accumulation and decay of peat. The wetland module, based on the TOPMODEL approach, dynamically determines inundated area and water table, which change depending on climate. Not all temporarily inundated areas accumulate peat, though, but peat accumulates in permanently inundated areas with rather stable water table position. Peatland area therefore is highly uncertain, and we perform sensitivity experiments to cover the uncertainty range for peatland extent. The peat module describes oxic and anoxic decomposition of organic matter in the acrotelm, i.e., the part of the peat column above the permanent water table, as well as anoxic decomposition in the catotelm, the peat below the summer minimum water table. We apply the model to the period of the last 8000 years, during which the model accumulates 330 PgC as catotelm peat in the peatland areas north of 40° N, with an uncertainty range from 240 PgC to 490 PgC. This falls well within the range of published estimates for the total peat storage in high northern latitudes, considering the fact that these usually cover the total carbon accumulated, not just the last 8000 years we considered in our model experiments. In the model, peat primarily accumulates in Scandinavia and eastern Canada, though eastern Europe and north-western Russia also show substantial accumulation. Modelled wetland distribution is biased towards Eurasia, where inundated area is overestimated, while it is underestimated in North America. Latitudinal sums compare favourably to measurements, though, implying that total areas, as well as climatic conditions in these areas, are captured reasonably, though the exact positions of peatlands are not modelled well. Since modelling the initiation of peatland growth requires a knowledge of topography below peat deposits, the temporal development of peatlands is not modelled explicitly, therefore overestimating peatland extent during the earlier part of our experiments. Overall our results highlight the substantial amounts of carbon taken up by peatlands during the last 8000 years. This uptake would have substantial impacts on the global carbon cycle and therefore cannot be neglected.