Characterizing soil piping networks in loess‐derived soils using ground‐penetrating radar

Soil piping remains a relatively unexplored phenomenon despite its substantial impacts on water and sediment transfer at the watershed scale at numerous locations around the world. One of the main limitations regarding the study of this singular process is the characterization of the pipe networks (in terms of number, position, dimensions, and connectivity of pipes). In this context, noninvasive subsurface imaging using ground‐penetrating radar (GPR) has proven to be a promising technique. This study used two three‐dimensional (3D), high‐resolution GPR surveys performed in loess‐derived soils to characterize small pipe networks with little prior information about their location. The adopted methodology relies on high spatial resolution scanning, 3D subsurface imaging, and automated detection of reflection hyperbolas using a 200‐MHz center‐frequency antenna. Two small watersheds affected by piping were investigated at Sippenaeken and Kluisbergen (Belgium). Over the two scanned zones, results revealed significant subsurface continuous patterns. Even though the most obvious patterns corresponded to recent or past anthropic activities (e.g., artificial drainage pipes), validation tests confirmed that the chosen methodology can be used for pipe network characterization because some important continuity patterns were related to soil piping: for instance, a fairly small pipe (approximately 10–15 cm in diameter) was detected and validated for more than 100 m. Nevertheless, the high variability in size, depth, and orientation of the pipes imply that GPR may only be truly efficient when using very high spatial resolution scanning, which limits its application to specific conditions not always met in piped areas.