Soil-atmosphere CH4 and CO2 fluxes of multiple peatland forests simulated with JSBACH-HIMMELI model

Around half of the European peatlands are drained, and in Finland, most of them are drained for forestry. Drainage degrades the soil organic matter (SOM) and lowers the soil water-table level (WTL), increasing oxygen levels in the soil. This suppresses the production and increases in-soil oxidation of methane (CH₄) but enhances the decomposition of SOM, accelerating aerobic soil respiration. Consequently, emissions of CH₄ from soil to the atmosphere may decrease while those of CO₂ may increase. Process-based models are useful in estimating greenhouse gas emissions and sinks from large areas.  In previous modelling studies, the focus has mainly been on pristine peatlands with high CH₄ emissions. However, simulations of soil CH₄ and CO₂ fluxes of multiple forestry-drained peatland sites over several years that are compared with measurement data remain still scarce.

We simulated the soil-atmosphere CH₄ and CO₂ fluxes from six Finnish forestry-drained peatlands with a process-based model, JSBACH (Jena Scheme for Biosphere–Atmosphere Coupling in Hamburg) coupled with a peatland CH₄ model, HIMMELI (HelsinkI Model of MEthane buiLd-up and emission). Our aim was to better understand and evaluate the accuracy of the predicted soil CO₂ and CH₄ fluxes from multiple peatland forest sites, and to identify the sources of uncertainty in the modelled fluxes. To do this, we used WTL and chamber flux data measured over 2-5 years from each site.

The average modelled soil CO₂ fluxes varied between 0.7 and 1.42 µmol m^-2 s^-1 among the sites. The model overestimated emissions in two sites and underestimated them in three sites. The mean differences between model and measurement varied from 0.05 to 2.02 µmol m^-2 s^-1 among all sites. There was a clear interannual variation on this. The average modelled CH₄ fluxes varied between -1.09 and 3.77 nmol m^-2 s^-1 among the sites. The model underestimated sink or predicted occasional CH₄ emission peaks in four sites. In turn, the CH₄ sink was overestimated by the model in two sites. The measurements indicated all the sites being, on average, small sinks of CH₄. The mean differences between modelled and measured CH₄ fluxes were between 0.44 and 5.44 nmol m^-2 s^-1 among the sites. Generally high WTL of a site was associated with larger discrepancies between modelled and measured CH₄ fluxes. The WTL was considered high for three sites (modelled WTL on average -30  – (-32) cm), and low for three sites (modelled WTL on average -42 – (-64) cm). We found that by tuning the CH₄ production and oxidation parameters in the model, we can improve the prediction accuracy of the modelled CH₄ fluxes.

The results of this work will be useful for further model development and when aiming to estimate soil CH₄ and CO₂ sinks and emissions of forestry-drained peatlands.