Soutenance

Northern peatlands play an important role in the global C cycle as a long-term CO₂ sink and the one of the largest natural methane (CH₄) sources. Meanwhile, these substantial carbon stores will be exposed in the future to large warming and wetter conditions that characterize climate change in the high latitudes and, because of the large amount of C stored in northern peatlands, their fate is of concern. In this thesis, I integrated a representation of peatlands water and carbon cycling into the ORCHIDEE-MICT land surface model (LSM), with the aim to improve the understanding of peatland C and area dynamics since the Holocene, to explore effects of projected climate change to northern peatlands, and to quantify the role of northern peatlands in the global C cycle.

Firstly (Chapter 2), I implemented peatland as an independent sub-grid hydrological soil unit (HSU) which receives runoff from surrounding non-peatland HSUs in each grid cell and has no bottom drainage, following the concept of Largeron et al. (2018). To model vertical water fluxes of peatland and non-peatland soils, I represented peat-specific hydrological parameters for the peatland HSU while in other HSUs the hydrological parameters are determined by the dominant soil texture of the grid cell. I chose a diplotelmic model to simulate peat C decomposition and accumulation. This two-layered model includes an upper layer (acrotelm) that is variably inundated and a lower layer (catotelm) that is permanently inundated. The simulated water table position was used to separate oxic from anoxic decomposition of carbon in the acrotelm zone, while in the catotelm, carbon is decomposed anaerobically all year round. This model showed good performance in simulating peatland hydrology, C and energy fluxes at 30 northern peatland sites on daily to annual time scales. But the over simplification of the C dynamics may limit its capacity to predict northern peatland response to future climate change.

Secondly (Chapter 3), I replaced the diplotelmic peat carbon model with a multi-layered model to account for vertical heterogeneities in temperature and moisture along the peat profile. I then adapted the cost-efficient version of TOPMODEL and peatland establishment criteria from Stocker et al. (2014) to simulate the dynamics of peatland area within a grid cell. Here the flooded area given by TOPMODEL is crossed with suitable peat growing conditions to set the area that is occupied by a peat HSU. This model was tested across a range of northern peatland sites and for gridded simulations over the Northern Hemisphere (>30 °N). Simulated total northern peatlands area and C stock by 2010 is 3.9 million km2 and 463 PgC, fall well within observation-based reported range of northern peatlands area (3.4 – 4.0 million km2) and C stock (270 – 540 PgC). 

Lastly (Chapter 4), with the multi-layered model, I conducted factorial simulations using representative concentration pathway (RCP)-driven bias-corrected past and future climate data from two general circulation models (GCMs) to explore responses of northern peatlands to climate change. The impacts of peatlands on future C balance of the Northern Hemisphere were discussed, including the direct response of the C balance of the (simulated) extant peatland area, and indirect effects of peatlands on the terrestrial C balance when peatlands area change in the future.  

This thesis proposes an advancement of modelling peatland area and carbon dynamics consistently with the water balance of these systems with the ORCHIDEE-MICT LSM, providing a new tool to study the role of peatlands in the global C cycle. Future work will focus on including influences of land use change and fires on peatland into the model, given that substantial losses of C could occur due to these disturbances. To have a complete picture of peatland C balance, CH4 and dissolved organic C (DOC) losses must be considered.