The morphology and prevalence of macropores < 10 cm in diameter in forested riparian wetlands is largely unknown despite their importance as a mechanism for preferential flow of contaminants to stream channels. Here, we validate field procedures for detecting and mapping the three‐dimensional structure of near‐surface (15–65 cm deep) lateral macropore networks using non‐invasive ground‐penetrating radar (GPR) technology at a Mid‐Atlantic riparian wetland field study site. Soil core samples used to ground truth the procedures showed that the detection predictions were 92% accurate and tracer dye transmission through the site corroborated the morphology predictions. The results demonstrate the feasibility of using GPR to map preferential flow networks in situ without disturbing environmentally sensitive wetland ecosystems.

Preferential flow of water through soil macropores is known to contribute to groundwater and surface water contamination as well as stream bank instability. However, research on the mechanisms and extent of soil macroporosity is limited due to the lack of a practical technique to study macropores in situ without disrupting the site's ecological function. In this paper, we present a ground‐penetrating radar (GPR)‐based methodology for detecting soil macropores smaller than 10 cm in diameter within 1 m of the soil surface and then creating a computerized tomogram of the macropore network. Manual and automated algorithms for macropore detection were tested for scan data collected using a 900‐MHz radar antenna in a field experiment with a silt‐loam soil. Buried polyvinyl chloride (PVC) pipes were used to simulate soil macropores of different diameters and fill contents intersected by GPR scan lines at four different angles. Pipes ≥ 3.00 cm in diameter were clearly detectable regardless of the scan line orientation relative to the target, and pipes with diameters as small as 1.85 cm were detected at perpendicular angles of intersection. In a second field experiment, PVC pipes of varying dimensions were buried at different depths to simulate a macropore network of preferential flow pathways. A branch‐node algorithm was developed that referenced GPR scan line detections to create an accurate computer‐generated three‐dimensional map of the pipe network.