Pore network modeling as a new tool for determining gas diffusivity in peat

Abstract

Peatlands are globally significant carbon stocks and may become major sources of the greenhouse gases (GHGs) carbon dioxide and methane in a changing climate and under anthropogenic management pressure. Diffusion is the dominant gas transport mechanism in peat; therefore, a proper knowledge of the soil gas diffusion coefficient is important for the estimation of GHG emissions from peatlands. Pore network modeling (PNM) is a potential tool for the determination of gas diffusivity in peat, as it explicitly connects the peat microstructure and the characteristics of the peat pore network to macroscopic gas transport properties. In the present work, we extracted macropore networks from three-dimensional X-ray micro-computed tomography (µCT) images of peat samples and simulated gas diffusion in these networks using PNM. These results were compared to the soil gas diffusion coefficients determined from the same samples in the laboratory using the diffusion chamber method. The measurements and simulations were conducted for peat samples from three depths. The soil gas diffusion coefficients were determined under varying water contents adjusted in a pressure plate apparatus. We also assessed the applicability of commonly used gas diffusivity models to peat. The laboratory measurements showed a decrease in gas diffusivity with depth due to a decrease in air-filled porosity and pore space connectivity. However, gas diffusivity was not extremely low close to saturation, which may indicate that the structure of the macropore network is such that it enables the presence of connected diffusion pathways through the peat matrix, even in wet conditions. The traditional gas diffusivity models were not very successful in predicting the soil gas diffusion coefficient. This may indicate that the microstructure of peat differs considerably from the structure of mineral soils and other kinds of porous materials for which these models have been constructed and calibrated. By contrast, the pore network simulations reproduced the laboratory-determined soil gas diffusion coefficients rather well. Thus, the combination of the µCT and PNM methods may offer a promising alternative to the traditional estimation of soil gas diffusivity through laboratory measurements.