GNGTS 2015 - Atti del 34° Convegno Nazionale

dot not allow a complete evaluation of the embankment and lack in identifying the (probable) discontinuities within this segment. The remaining three soundings examined the earthen levee, which is homogeneously composed of silty-clayey sand, as expected. Furthermore, they show a clayey level between 11 m and 13 m depth. Given the need of dealing with these infiltration events, new interventions modified the outer face of the reconstructed embankment in 2014 (third step). A stability bank with a 0.5 m thick gravel drainage mattress was built, with a seepage berm and a collector ditch to avoid water stagnation. A section of the resulting geometry is outlined in Fig. 2a. The embankment total height from ground level is circa 8.7 m, with a crown 4 m wide, and both inner and upper outer faces have a ratio of width to height of 3:2. The stability bank is 5 m high from ground level, has a ratio of width to height of 2:1 and a 4 m wide crown. Methodologies and acquisitions. The characterization of this site is performed through the application of four different methodologies: electrical resistivity tomography (ERT), multichannel analysis of surface waves (MASW), ground penetrating radar (GPR), and self- potential (SP). Each technique provides different images of the investigated domain (in terms of resistivity, S-wave velocity, dielectric constant, and potential, respectively), whose combination allows a more comprehensive characterization with respect to the single survey. The first methodology, ERT, is applied with different set-ups, in order to vary depth of investigation, resolution, and section of the embankment. The surveys consisted of: 1. lengthwise profiles: 48 stainless-steel electrodes were placed along the river crest, in correspondence to the reconstructed part. We used two different spacings, 1 m and 2 m, so as to vary both total length (47 m and 96 m, respectively) and depth of investigation (about 10 m and 20 m, respectively). All these surveys share the position of the 23 rd electrode, which was fixed to guarantee the overlapping of the profiles. Furthermore, we employed both Wenner-Schlumberger and dipole-dipole skip zero acquisition schemes; 2. cross-embankment profiles: we used two different lines (one upstream and one downstream, labelled T2 and T1 respectively), with 48 stainless-steel electrodes spaced 0.75 m each. In both cases the acquisition scheme chosen was dipole-dipole skip 0. They covered the outline of the embankment (from the seepage berm to the water level, via the embankment crest) in two cross-sections, selected according to the results of the lengthwise surveys; 3. cross-river profiles: we placed 72 stainless-electrodes spaced 1 m along a line covering the outline of the reconstructed embankment, the riverbed bed, and the profile of the left bank. Also here, we employed both Wenner-Schlumberger and dipole-dipole skip zero acquisition schemes. Each ERT sounding is performed with an IRIS Instruments Syscal Pro resistivity meter measuring both direct and reciprocal resistance values (Daily et al. , 2004), necessary to assess the measurement error. A summary of date, set-up, and array used for each ERT survey can be found in Tab.1 while Fig. 1 shows the location of these soundings. Then, we used the MASW technique to compare the reconstructed sector to a natural adjacent one, given the difference in terms of mechanical properties that should arise as a consequence of the different materials involved. More in detail, we took advantage of a seismic streamer, a specific tool composed of several geophones linked together that allows the dragging of the whole equipment along the embankment crest. We carried out 25 measurements, whereof 19 on the reconstructed sector (in correspondence to the ERT lengthwise profiles) and 6 on the earthen levee. In the first case, we energised with a sledge-hammer at both ends of the line (i.e., at 4 m and 6 m downstream, with respect to the first geophone, and at 4 m upstream with respect to the last geophone); in the second case, we energised only at 4 m downstream from the first geophone. For each MASW sounding we used a Geometrics Geode seismograph with sampling rate equal to 250 ns, acquisition time equal to 2 s, and 24 geophones with 4.5 Hz frequency spaced 2 m (i.e., total length of the line equal to 46 m) placed lengthwise on the embankment’s GNGTS 2015 S essione 3.2 45