Peatland dynamics in response to past and potential future climate change : A regional modelling approach

Sammanfattning: The majority of the northern peatlands developed during the Holocene as a result of a positive mass balance between net primary productivity (NPP) and heterotrophic decomposition rates. Over that time they have sequestered a huge amount of carbon in terrestrial ecosystems. A significant proportion of these areas also coincides with areas underlain with permafrost and shows a diverse range of peat accumulation patterns. Thus, for predicting and understanding the long-term evolution of peatland carbon stocks across the pan-Arctic, mechanistic representations of both peatland and permafrost dynamics are needed in the modelling framework. In this thesis, a novel implementation of dynamic multi-layer peatland and permafrost dynamics in the individual- and patch- based dynamic vegetation and ecosystem model (LPJ-GUESS) is described. The major emphasis of this work goes into enhancing the current understanding of the processes involved in the long-term peat accumulation and its internal dynamics, including how these systems are influenced by small-scale heterogeneity, vegetation dynamics and interactions with underlying permafrost. A simple two-dimensional microtopographical (2-DMT) model was also developed to address the established hypotheses concerning stability, behaviour and transformation of these microstructures and the effects of this small-scale heterogeneity on the coupled dynamics of vegetation, hydrology and peat accumulation. LPJ-GUESS was calibrated and validated using data from a mire in Stordalen, northern Sweden, and evaluated using data from multiple sites in Scandinavia and from Mer Bleue, Canada. It was subsequently applied across the pan-Arctic to advance the existing knowledge on carbon accumulation rates at different spatial and temporal scales, and also to demonstrate the potential implications of current warming on these climate sensitive ecosystems. Both of the models developed in this thesis performed satisfactorily when confronted with experimental data.LPJ-GUESS is quite robust in capturing peat accumulation and permafrost dynamics including reasonable vegetation and hydrological conditions at temporal and spatial scales across various climate gradients. The simulations improved our knowledge of peatland functioning in the past, present and future. It was found that Stordalen mire will continue to accumulate carbon in the coming decades but later will turn into a carbon source. It was also found that permafrost-free regions that are predicted to experience reduced rates of precipitation may lose significant amount of carbon in the future due to reductions in soil moisture. Conversely, peatlands currently underlain with permafrost could gain carbon due to an initial increase in soil moisture as a result of permafrost thawing. My modelling results also suggest that peatlands can show diverse range of behaviour with alternative compositional and structural dynamics depending on the initial topographical, climatic conditions, and plant characteristics, therefore, it will be challenging to represent such dynamics in current Earth System Models (ESMs). With the inclusion of aforementioned processes, LPJ-GUESS has now become quite robust. The resultant model can now be coupled with ESM where it can address issues related to peatland-mediated biogeochemical and biophysical feedbacks to climate change in the Arctic and globally.