A fast and integrated geophysical imaging system for large scale levee monitoring

During recent floods in Northern and Central Italy as well as in the entire European Region several levee failures occurred and caused severe damages to the agriculture, to the nearby buildings and to the infrastructures. In most of the cases the collapse was due to high permeability anomalies within and below the levee body and frequently to recent development of small to medium size cavities caused by biological activity. The conditions of river embankments, especially when the structures are very old, turn then into a social and vital issue as a failure could cost up to hundreds of millions of Euro of damages and in extreme cases could also result in several casualties. Visual inspection of the surface along with sparse geotechnical testing has been the standard for several years for levee safety assessment but this approach is not fully satisfactory. Cone penetration tests and boreholes provide a detailed stratigraphy of the levee itself and of the underlying deposits but unfortunately it provides local information and no data are available in between tested spots. Geophysical methods could then play a major role in filling these gaps. Resistivity profiling (ERT), among the various geophysical techniques, is probably the most effective but it is fairly slow and expensive to become a practical tool for extensive monitoring of thousands of kilometres of levees. An integrated geophysical imaging system (namely EMAR), based on multifrequency electromagnetic (FDEM) induction and multichannel radar (GPR), has been developed, tested and validated along more than 100 kilometres of river embankments in North-eastern Italy. Using this approach it was possible to investigate up to 20 km of embankment per day showing interesting potentials for regional surveying and large scale monitoring. The EMAR anomalies (an average of two per kilometre) were further investigated with ERT profiles and, when confirmed, a small trench was opened to visually inspect the inner portion of the levee. A percentage of more than 90% of the first phase anomalies resulted in ERT anomalies caused by sand layers in the levee structure. The FDEM techniques are known to be affected by environmental noise and often also by sudden changes in the field background. To avoid the complexity of frequent FDEM site calibration and of the daily magnetic drift of the sensors, we developed a novel approach for EM data processing. Each “homogeneous” levee segment was processed separately and the resistive anomalies (high permeability sand layer or cavity), were identified as large negative deviation from a mean value typical of that levee segment. This procedure was repeated for all the various spacings and frequencies used during the survey. GPR data, after a semi-automatic check of the cart trajectories (15 channels), were also processed automatically to map anomalous reflectors spanned in 12 time planes, calculated every 5 ns in the two-way traveltime interval 0 ns — 60 ns. A GIS multivariate analysis algorithm was finally adapted to compare and weight the different signals (typically 3 EM frequencies and 12 radar reflectivity planes) and assign the levee heterogeneity to different classes of failure risk.