Quantifying soil structure and preferential flow in intact soil using x-ray computed tomography

Computed tomography (CT) provides a nondestructive means of observing soil structure and monitoring solute breakthrough in real time. We investigated an intact soil column 10 cm in diameter and 30 cm in length using industrial CT with a resolution of 105.5 by 105.5 by 125.25 p m. The satiated soil column was scanned to obtain overall soil structure. Then 60 g L-1 HI solution was injected at 6.6 mL min(-1) for about 23 h and solute transport was monitored in real time by scanning two critical positions in the column and taking digital radiographs. At the end of the experiment, the whole column was scanned again to obtain the overall solute mass distribution. The voxel-based soil porosity and solute concentration were quantified. The three-dimensional visualization of the pore network and solute distribution with time showed that both the pore network and the flow pattern varied considerably with soil depth, in part due to the soil horizonation and different macropores involved. Although the macroporosity below the Ap1 horizon was much lower, macropores were more continuous and less tortuous as a result of limited agricultural disturbance and more earthworm activities. Only part of the macropores at the subsurface were effective, however, in transporting the solute. The results revealed a sequential initialization of the transport process from the macropore domain to the matrix domain and a decreased degree of interaction between the two domains with soil depth. Point-specific breakthrough curves were obtained from real-time point-specific solute concentration and porosity, from which point-specific pore velocity was determined. Preferential flow pathways in this intact structured soil consist of a complex network of earthworm burrows, root channels, interaggregate macropores, and mesopores or even micropores in the soil matrix. Modeling of this flow network and its dynamics requires a new approach different from the classical continuous-domain approach.