Using X-ray radiography to image rapid water infiltration into soil

While the basic processes of water infiltration into soil are well understood, their details are difficult to quantify due to the opaque nature of soil. In this study, we investigated the potential and limitations of X-ray radiography to measure the water front progression in a narrow sample (15 × 15 × 1 cm) filled with dry soil under simulated rainfall of high intensity (53 mm/h). The temporal resolution of the acquired infiltration movies was 133 milliseconds. We evaluated three different kinds of soil samples. i) Bare samples filled with a 1:1 mixture of a sandy loam and peat. ii) The same soil-peat mixture, but cropped with Trifolium incarnatum, Trifolium repens, Lathyrus odoratus and Lupinus regalis, all of them plants that have been reported to induce water repellency; prior to the experiments, the plants were harvested and only the roots remained in place. iii) Sandy loam soil that had been incubated for four weeks in an outside garden plot. Due to time limitations of the project, the incubation period was in early spring, which meant that plant growth in the samples was negligible. All three sample types were replicated five times, resulting in 15 individual samples. We carried out the infiltration experiments in four-fold replications, from which it follows that we collected 60 individual infiltration movies. After each infiltration round, the samples were placed in a drying room to reset them to a similar initial moisture content. The experiments aimed at testing i) whether the infiltration patterns of the four consecutive infiltration runs conducted on each sample remained identical and ii) to document differences in infiltration patterns between bare, cropped and incubated samples. We found that increasing X-ray scattering with increasing soil water contents made it challenging to evaluate the image data quantitatively. Advantages of our setup are that X-ray captures the complete water content at a specific depth and that sample boxes with irregularly shaped walls can be used to prevent preferential flow along the walls. Preliminary analyses of the data showed that the infiltration fronts in the bare and the incubated samples were less uniform than the ones for the cropped samples. In contrast, the likelihood of observing the same infiltration pattern in all four consecutive infiltration runs was larger for the bare and the incubated samples. The latter fact may have been caused by the interaction with root exudates in the cropped samples that may have been redistributed or mineralized during the wetting-drying cycles. We conclude that the here presented setup has large potential to investigate unstable infiltration phenomena into soil after an image correction approach has been developed that removes X-ray scattering artifacts. Alternatively, scattering may be suppressed by using a collimator during image acquisition.