Modelling peatlands as a complex adaptive systems

Abstract

A new conceptual approach to modelling peatlands, DigiBog, involves a Complex Adaptive Systems consideration of raised bogs. A new computer hydrological model is presented, tested, and its capabilities in simulating hydrological behaviour in a real bog demonstrated. The hydrological model, while effective as a stand-alone modelling tool, provides a conceptual and algorithmic structure for ecohydrological models presented later. Using DigiBog architecture to build a cellular model of peatland patterning dynamics, various rulesets were experimented with to assess their effectiveness in predicting patterns. Contrary to findings by previous authors, the ponding model did not predict patterns under steady hydrological conditions. None of the rulesets presented offered an improvement over the existing nutrient-scarcity model. Sixteen shallow peat cores from a Swedish raised bog were analysed to investigate the relationship between cumulative peat decomposition and hydraulic conductivity, a relationship previously neglected in models of peatland patterning and peat accumulation. A multivariate analysis showed depth to be a stronger control on hydraulic conductivity than cumulative decomposition, reflecting the role of compression in closing pore spaces. The data proved to be largely unsuitable for parameterising models of peatland dynamics, due mainly to problems in core selection. However, the work showed that hydraulic conductivity could be expressed quantitatively as a function of other physical variables such as depth and cumulative decomposition. DigiBog architecture was used to build a simple, vertical, ecohydrological model of long-term peat accumulation. As model complexity was increased under a self-organisation approach, model predictions of peat accumulation rates and surface wetness changed dramatically, revealing the importance of feedbacks between peatland hydrological behaviour and peat physical properties. This work may have important implications for palaeoclimatic reconstructions which assume peatland surface wetness to be a reliable climatic indicator. The expansion of the model to include horizontal space altered model behaviour in quantitative and qualitative terms.