Anomaly detection in earth dam and levee passive seismic data using multivariate Gaussian

As earth dams and levees (EDLs) across the United States reach the end of their design lives, effectively monitoring their structural integrity is of critical importance. This paper investigates automatic detection of anomalous events in passive seismic data as a step towards continuous real-time monitoring of EDL health. We use a multivariate Gaussian machine-learning model to identify anomalies in experimental data from two different laboratory earth embankments. Additionally, we explore five wavelet transform methods for signal denoising; removing different signal components. The best performance is achieved with the Haar wavelets (removing the Level 3 component). We achieve up to 97.3% overall accuracy and less than 1.4% false negatives in anomaly detection. These promising approaches could eventually provide a means for identifying internal erosion events in aging EDLs earlier than is currently possible, thereby allowing more time to prevent or mitigate catastrophic failures.