**4. Results**

#### *4.1. Ground-Penetrating Radar*

Representative radargrams obtained from the three GPR campaigns are shown in Figure 4. Some clear artifacts, due to electromagnetic noise are still visible in the data, even after the described processing sequence, e.g., the horizontal events at 650 ns and 700 ns in Figure 4b,c. The overall data quality is very poor if compared with the typical S/N ratio of GPR data acquired on other glaciers. High scattering is noticed along the profiles, with extremely high attenuation of the EM signals with depth. Not always clear readability in the processed profiles was observed for the 70-MHz surveys. Even increasing the total recording time (Figure 4c compared to Figure 4b), no reflections were observable in the data below 900 ns. Better results were obtained with the 40-MHz antenna (Figure 4d). In these radargrams, reflections appear more clearly and can be spatially followed and correlated between adjacent profiles. An exemplificative marked interpretation of the ice bottom morphology on the radargrams of Figure 4 is reported in Figure 5. Similar interpretations have been performed on the remaining profiles of the different campaigns for which ice bottom picking was clear and reliable.

The ice bottom appears as a discontinuous reflecting horizon in all the surveys. Layering with different orientations is noticed in some parts of the radargrams at the ice bottom location (e.g., distance = 38 m, time = 1000 ns in profile 15 of Figures 4d and 5d). A low-amplitude reflector deeper than the ice bottom is depicted in profile 12 of Figures 4d and 5d. This interface is characterized by a steeper dipping with respect to the ice bottom, in dip direction opposite to the glacier movement, reaching times higher than 1150 ns (depth > 100 m) in the center of the profile, at a distance of approximately 120 m from the glacier front.

In Figure 5c,d, the approximate locations (A' to D' and A" to D") of the single-station passive seismic measurements (A to D) are projected on the GPR sections, for further comparisons between the two geophysical methods.

The results of ice bottom picking along GPR profiles are summarized in Figure 6. All the 40-MHz radargrams provided quite clear imaging of the subsurface conditions. Conversely, only 12 radargrams acquired on October 2016 (70 MHz) supplied information on ice thickness, close to the front of the N lobe, on the S lobe and at the confluence. These areas correspond to generally lower ice thicknesses with respect to the surrounding zones. Analogously, the 70-MHz GPR campaign of March 2018 resulted in only 5 radargrams for which ice bottom picking was clearly interpretable. No information on ice thickness was recovered from the GPR data for the glacier main body, upstream the terminal bifurcation (Figure 6a). Ice thickness retrieved from ice bottom picking is shown in Figure 6b. A traditional U-shaped bottom morphology seems to be reconstructed for the N lobe, with considerable ice depths in the axial part (100–110 m) despite the proximity to the terminus. The S lobe is conversely characterized by significantly shallower ice depths and a probably flatter morphology.

**Figure 4.** GPR results. (**a**) Location of the GPR profiles are shown in the right and bottom sections. The starting point of each profile is highlighted with a dot. (**b**) Profile 9 (70-MHz antenna, October 2016). (**c**) Profile 20 (70-MHz antenna, March 2018). (**d**) Profile 12 to 15 (left to right, 40-MHz antenna, December 2018). The vertical scale of some radargram is cut (no reflections for higher recorded times).

#### *4.2. Single-Station Passive Seismic Measurements*

The results of HVSR processing are shown in Figure 7. No clear single peaks are found in the results, at least two to three HVSR minor peaks can be depicted for each station. HVSR amplification is low (<2) for all the measured points. For the first three stations, the peak with the highest amplitude (f0) is located at similar frequency values, progressively decreasing from A to C. A higher frequency peak is noticed at station D.

Figure 8 shows HVSR directivity, as a function of frequency and azimuth (0◦ = N, 90◦ = E), measured at the four stations. These results clearly highlight that f0 spectral peaks spread over a wide azimuthal range, covering from 100◦ to 180◦. These resonance frequencies seem therefore almost azimuth independent, denoting the absence of 2-D effects. The secondary peaks located at frequencies lower than 3 Hz conversely showed more focused directivities.

**Figure 5.** GPR results (same as Figure 4) with glacier bottom interpretation. The ice bottom morphology is highlighted in the radargrams with the dashed black lines. The possible bedrock top is underlined in (**d**) with the dotted black line. The approximate location of single-station passive measurements projected on the GPR profiles is shown in (**a**) with letters A' to D' (on 70-MHz profile 9) and A" to D" (on 40-MHz profiles 12 to 15), and in (**<sup>c</sup>**,**d**) with the black rectangles, for further comparison between GPR and HVSR results.

**Figure 6.** (**a**) Summary of the GPR profiles for which ice bottom picking was retrieved. (**b**) Map of the resulting ice bottom picks.

**Figure 7.** HVSR results on stations (**a**) A, (**b**) B, (**c**) C and (**d**) D (location in Figure 2). In each panel, the black bold and dashed curves are the average and standard deviation obtained from all the accepted 120-s window curves for each station.

**Figure 8.** Directional HVSR azimuth (in degrees from N direction) on stations (**a**) A, (**b**) B, (**c**) C and (**d**) D. In each panel, the white dashed lines highlight the location of the highest spectral peaks (f0) identified in Figure 7.
