Modeling the impact of drainage on peatland CO 2 and CH 4 fluxes and its underlying drivers

Abstract. Peatland drying is an important process affecting greenhouse gas (GHG) emissions. Ditching of peat for drainage to plant forest has been a widespread management practice in the Nordic countries, and drying increasingly occurs also from climate change induced drought. Previously published meta-analyses from literature suggest that drainage increases soil CO2 emissions by enhancing oxic decomposition in aerated upper layers while suppressing CH4 emissions. However, these data do not elucidate short-term variations of GHG fluxes during drainage and usually only regress GHG emissions as a function of the annual mean water table. Here we developed a new parameterization of peat drainage in a land surface model that represents peat processes and fluxes of CO2 and CH4, by adding a machine-learning module to predict the daily water table depths from simulated soil moisture in the upper soil layers and a ditch that receives drainage water. Because peatland pre-drainage GHG emissions vary between sites and influence subsequent changes following drainage, idealized simulations were performed for virtual drainage applied to a collection of 10 pristine sites, where the model parameters are calibrated against observed GHG fluxes. Different drainage intensities are simulated by prescribing lower water table depths from setting the ditch depth from 5–80 cm below the initial water surface. The resulting GHG flux changes across sites are compared with meta-analysis data from northern sites and show realistic results with a reduced CO2 sink and reduced CH4 emissions. Additional comparison with continuous flux data collected in the UK for different sites associated with increasing drainage levels also shows good model performance. Overall, using GWP100 to compare the effect of CH4 vs. CO2 flux changes, our model simulations suggest only small net GHG emission changes in CO2-equivalent GHG emissions under drainage scenarios over multi-decadal timescales, yet with differences between sites. Over time, simulated emission factors of CO2 flux decrease because of exhaustion of labile soil organic substrate for decomposition, while reductions in CH4 emissions are amplified due to decreased availability of material for anoxic decomposition. The sensitivities of CO2 flux changes to increased water table depth changes are primarily controlled by initial CO2 and CH4 fluxes, initial soil carbon content, peat vegetation community, air temperature and initial water table depth. The influence of peat vegetation on the GHG flux sensitivities in the model occurs via differing lability of soil organic carbon pools, with moss-dominated sites having a lower sensitivity due to their longer peat turnover time. Nonetheless, our calibrated global model remains limited in its ability to mechanistically represent drained peatland systems, particularly regarding extrapolation and representation of dynamic soil and hydrological processes. Our model-simulated sensitivities of GHG fluxes to drainage can be approximated by linear regressions using site-level variables, which, despite the limitations, may offer a simplified, exploratory tool for estimating drainage effects.