Abstract:This study proposed a real-time seepage health monitoring approach for water-retaining earthen infrastructures, such as earth dams, dikes, and levees. The approach utilized the principle of heat–seepage interaction and the local Moran's index, a spatial autocorrelation measure, to detect clusters of seepage-induced temperature anomalies within raw temperature data measured through fiber optic Distributed Temperature Sensing (DTS). As new temperature readings are acquired during monitoring, the index is calculated for each reading and is compared against a threshold to classify them as normal or outliers (anomalous). This process is compiled into an algorithm that generates an Anomaly Detection Parameter (ADP), which can be traced as a function of DTS sensing position and time. Verification based on analytical modeling (using MATLAB) and laboratory experiments on small-scale earth dams involving embedded DTSs and artificially induced seepage anomalies demonstrated that ADP could provide interpretations, such as early detections, spatial coverage, and severity of seepage issues from temperature data. The proposed approach can be implemented through computer programming to reduce costs related to manual inspections,while offering a spatially extensive, reliable, long-term, and real-time monitoring solution for water-retaining earthen infrastructures.

Real-time seepage health monitoring using spatial autocorrelation of distributed temperature data from fiber optic sensor