摘要：Core Ideas For preferential flow modeling, different macropore types must be considered. A method was developed for separating biopores and cracks in 3D images from XRCT. The method enabled a more objective determination of structuring element sizes. The voxel‐based approach was found useful for quantification of macropore types. In structured soils, earthworm burrows, root channels, shrinkage cracks, and interaggregate spaces form complex macropore networks. Depending on the type and morphological properties, each macropore surface type is coated with specific organo‐mineral compounds, differently affecting sorption and mass exchange during preferential flow and turnover processes. For a quantitative, macropore type–specific analysis using X‐ray computed tomography (XRCT) with subsequent three‐dimensional (3D) image analysis, a discrimination of biopores from cracks and interaggregate spaces is necessary. We developed a method that allows separating biopores from other larger macropores in 3D images from XRCT of intact soil cores. An image‐processing workflow using the MAVI (Modular Algorithms for Volume Images) software framework ToolIP (Tool for Image Processing) was created to handle XRCT 3D images. Masking steps enabled to retain the surface roughness in the resulting two images of separated biopores and cracks. As a key point, the sizes of the structuring elements used in the spherical opening and dilation were objectively determined. For this purpose, maximum differences in the pore shapes between the 3D images of cylindrical biopores vs. more flat cracks and unregularly interaggregate spaces were focused. At the given resolution of 231‐μm voxel edge length, an optimum size of 2.5 voxels was found for both processing steps. The voxel‐based approach is applicable to XRCT 3D images of different spatial resolution and appears useful for the quantification of physicochemical surface properties of different macropore types for soil volumes, enabling a more precise description of preferential flow and transport.
