Linking distributed hydrological processes with ecosystem vegetation dynamics and carbon cycling: Modelling studies in a subarctic catchment of northern Sweden

Sammanfattning: The Arctic and Subarctic regions are of particular importance to the global climate change and are now experiencing a climate warming that is higher than the global average. Around 50% of the global soil carbon is stored in high latitude soils, especially in permafrost and peatland soils. Permafrost thawing, speeding up the decomposition of previously frozen soil carbon, is expected to result in strongly positive feedbacks to global warming. Meanwhile, increased air temperature may strongly impact vegetation growth and distributions in this region. Dynamic ecosystem models are powerful tools to study climate change influences on ecosystem processes and also to quantify ecosystem feedbacks to the atmosphere. However, these models often focus on the vertical transfer of carbon and water between the atmosphere, the land surface vegetation and soils. Therefore, they generally do not consider the horizontal water and soluble carbon flows between the modelled spatial units (grid cells), which could result in an incomplete estimation of water and carbon budgets, especially for climatically sensitive high latitude regions. In this thesis, we aim to overcome this limitation by implementing spatial topographical indices into a state-of-the-art dynamic ecosystem model, LPJ-GUESS, and to incorporate water and carbon (mainly dissolved organic carbon, DOC) interactions between the grid cells. Modelling approaches and algorithms developed in this thesis were applied to study the subarctic Stordalen catchment, located in northern Sweden, and to explore the potential influence on the model’s hydrological and ecological estimations. Extensive sets of observation data were used for model evaluation throughout. We proposed a distributed hydrological (DH) approach to dynamically simulate water flow from cell to cell within the catchment and compared the hydrological and ecological impacts resulting from different flow routing algorithms. The results indicate an improved accuracy of runoff estimation when using the proposed DH scheme in the Stordalen catchment. They also show that the choice of flow algorithm can have strong impacts on water and carbon flux estimations in this region. Furthermore, a complete estimation of the catchment carbon budget was assessed using our developed model. We found that the catchment is a carbon sink at present and could become a stronger sink in the near future, a result which is, however, very dependent on future atmospheric CO2 concentrations and methane (CH4) emissions from the peatlands. Additionally, the model was further extended to dynamically model soil water DOC and the lateral transport of DOC across the landscape. The modelled outputs suggest that DOC production and mineralization largely contribute to DOC fluxes and that wet fen peatland is and will be a hotspot for DOC export. In conclusion, this thesis demonstrates the feasibility of implementing topographical indices into LPJ-GUESS to describe water flows, and the importance of considering spatial heterogeneity in hydrological conditions when modelling carbon dynamics at high latitudes. Furthermore, the integration of vertical and horizontal carbon fluxes at high spatial resolutions can be used to provide more accurate estimations of a complete carbon budget and can dynamically simulate the fate of different carbon components in response to climate change.