GNGTS 2015 - Atti del 34° Convegno Nazionale

crest. Also for the MASW methodology, a summary of date and set-up for each survey can be found in Tab.1 and the profile locations are represented in Fig. 1. Ground penetrating radar is the third methodology used. Its application took place in correspondenceof the cross-embankment ERTlines (T1andT2), thus resulting in two radargrams, G1 (downstream) and G2 (upstream). In order to gain information regarding the embankment’s inner part, we chose a zero offset profile (ZOP) approach, where both the transmitter and the receiver are lowered along the inner and outer embankment faces, respectively. We used a PulsEKKO borehole system with two 100 MHz antennas. All information regarding date and profiling mode can be found in Tab.1 and Fig. 1. The last technique, SP, was used to monitor the variations in terms of potential along the seepage berm and the lower part of the embankment’s outer face.Also in this case all information is summarized in Tab.1 and Fig. 1. Data processing, results and discussion. Electrical resistivity tomography. We performed the inversion of the ERT data at our disposal thanks to two different software, both provided by Lancaster University (UK): Profiler (for lengthwise and cross-embankment profiles) and R2 (for the cross-river profiles). The former creates a quadrilateral finite element mesh, while the latter needs to be fed with a triangular finite element mesh, more suitable for profiles with a more complex topography. The error threshold for the resistance measurements and the error for the inversions are both fixed at 5%. The lengthwise resistivity cross-sections at our disposal well identify the presence of the reconstructed part within the river embankment, as depicted, for example, in Fig. 3a. This domain is highlighted by relatively higher resistivity values (150-600 Ohm·m, between 24 and 88 m, till a depth equal to circa 8 m from the embankment crest), with respect to the natural levee, which, on the contrary, is characterised by relatively lower resistivity values (50-100 Ohm·m). These values well agree with the materials forming these domains (i.e., reconstructions materials and clayey sand, respectively). The lower part of this cross-section shows rather low resistivity values (50 Ohm·m on average), which may be related to either natural sediments or grouting, according to the site description in section 2. Moreover, the cross-sections with 1 m electrode spacing helped us distinguishing two subdomains within the reconstructed part, thus showing some heterogeneity in this structure. The first, downstream, has an average resistivity of 500 Ohm·m; the second, upstream, has an average resistivity of 300 Ohm·m. These differences led us to the two cross-embankment soundings, T1 and T2 (downstream and upstream, respectively), aimed at identifying the nature of this variation (Fig. 1). First of all, T1 has an average higher resistivity, if compared to T2. More in detail, each cross-section can be divided into two subdomains (Fig. 2b): - Lateral parts: both inner and outer parts have higher resistivity values because of the ballasts covering the embankment faces. Moreover, these cross-sections well identify also the drainage mattress within the stability bank; - Central part: in both cases, this subdomain has lower average resistivity values (<150 Ohm·m), likely due to the cement septum, in agreement with other examples in literature (e.g., Sjödahl et al ., 2009; Karhunen et al ., 2010). Furthermore, T1 shows a horizontal zone with resistivity of 150 Ohm·m that does not appear in T2, thus underlying the absence of homogeneity within the reconstructed embankment. These differences may also be the reason behind different seepage phenomena, as described in section 2. Finally, we decided to combine both dipole-dipole and Wenner-Schlumberger resistance values to obtain the cross-river profile, in order to take advantage of both surveys into a unique resistivity cross-section. The result, shown in Fig. 3b, well represents the difference between the reconstructed embankment (on the left) and the natural one (on the right). The left side shares the same features described for the cross-embankment profiles, but has a lower resolution, given the higher electrodes spacing (0.75 m vs. 1 m); the right side shows the natural levee, marked by a more conductive and homogeneous core (likely clayey sand), covered with more resistive GNGTS 2015 S essione 3.2 47