Application of local porosity to define pore networks and pore geometry in soils: A case study along a carbon dioxide and temperature gradient

Pore network models are useful tools to investigate soil pore geometry. These models provide quantitative information of pore geometry from 3D images. However, there are limitations in image sizes and resolutions to extract networks. This study presents a modified pore network model to characterize large images with local porosity. The objectives of this work were to apply the modified model to characterize pore structure from large images at different scales (aggregate and soil column), image sizes, and resolutions and to characterize changes in pore structure induced by different levels of CO2 and temperature. Soil samples were taken from three sites (urban site with the highest, suburban with intermediate and rural with the lowest CO2 concentration and temperature). Undisturbed columns (5.5 cm in diameter and 12 cm in height) and aggregate samples were taken from each site and scanned with a computer tomographer at resolutions of 22 (column) and 6 microns (aggregates).;Pore networks were extracted by medial-axis transformation from local porosities at a unit cell and were used to measure pore geometry from aggregates and bulk soils. Three image volumes and 12 cell sizes were used to define image and cell size scaling effects. The configuration entropy and universal multifractals were employed to characterize pore spatial distributions, and water retention and hydraulic conductivity were measured on bulk soils.;Pore numbers and pore volumes measured in soil columns and aggregates had a linear relationship in log-log plots across cell sizes, while pore length and tortuosity did not show any specific trend. These results imply that some properties cannot be accurately projected to different scales within aggregate and laboratory scales.;Pore spatial distribution in bulk soils from all three sites and pore geometry information from aggregates and bulk soils of urban and rural site showed that different level of CO2 and temperature affected pore structure formations. Pores from urban site were more widespread and were greater than rural site. Hydraulic properties confirmed that urban soil had more connected and less tortuous pores than rural soil.;The modified pore network model is a powerful tool to characterize pore properties from large size images.