**6. Conclusions**

Geophysical surveys, including 40-MHz and 70-MHz GPR profiling and single-station passive seismic measurements, were acquired at the Belvedere Glacier with the aim of ice thickness reconstruction. The surveys were intended to fill a knowledge gap on the glacier bottom morphology, since the existing bottom information was only retrieved from sparse and local past geophysical measurements. Despite a dense GPR survey distribution, the ice bottom surface was detected only in radargrams acquired on the frontal lobes, while noisy and uninterpretable results were obtained upstream. The globally observed low data quality was attributed to the presence of a widespread debris cover, the absence of direct coupling between antenna and glacier surface and the probable widespread presence of debris and/or small-scale water bodies within the ice column. The concurrence of all these features probably caused the observed significant scattering and attenuation of the transmitted EM pulses. The lower frequency 40-MHz survey generally showed clearer results. Ice thickness estimations based on these results were interpolated to reconstruct a detailed ice thickness map that was compared with the previous study of De Visintini [35]. Some discrepancies between the two estimations arose, mainly due to different data coverage and acceleration in the glacier retreat over the last decades. A better fitting with the results of the radar measurements of VAW-ETH [36] was obtained. Also in this case, continuous GPR profiling provided denser data coverage with respect to previous local measurements, enabling for a better definition of the ice bottom morphology. Significant ice thickness variations were detected in the upper part of the N lobe, with a transition from convex to concave glacier topography. Below the ice bottom, low-amplitude reflectors having steeper dip were identified only in the frontal portion of the N lobe. These elements may indicate the bedrock presence at a depth of around 80 m from the glacier surface close to the northern terminus and rapidly deepening upstream. Consequently, a thick layer (more than 40 m) of subglacial deposits may be present between ice and bedrock.

Single-station passive seismic measurements were processed following the HVSR method. The results showed the absence of a clear contrast in acoustic impedance (i.e., ice on bedrock) at depth, providing at least a general confirmation of the hypothesized subsurface conditions. However, without the reference GPR profiles, no information on the ice bottom could have been retrieved from HVSR curves. These tests highlighted the limitations of passive seismic measurements for glacier characterization, i.e., 1D approximation of the investigated subsurface, need to simultaneously have a dense grid of measurements but avoiding 2D effects, time-consuming acquisitions with respect to continuous GPR profiling, poor results in absence of a bottom ice-bedrock interface. Differently from GPR investigations, for which the denser data coverage of continuous profiling revealed the key point to improve the past knowledge about bottom morphology, due to the peculiar investigated conditions passive seismic methods did not succeed in improving past active seismic results, despite the lighter instrumentation, logistically easier acquisition and no need of active sources.

Future perspectives of the method may be addressed to glacier monitoring. Multi-station long-term passive seismic measurements can potentially be used for the investigation of the ongoing glacial processes (hydrogeological modifications, meltwater flow, seepage and accumulation) and of the glacier movements and stability conditions (e.g., icequakes, opening of crevasses, basal movements, serac falls and stability of the frontal compartments).

Further glaciological analyses are planned to understand the influence of the glacier subsurface conditions on the measured geophysical data. Future geophysical campaigns on site should be addressed to reach satisfactory imaging of the glacier bottom in the upstream sectors, and to monitor ice thickness variations over the investigated frontal areas. Alternative survey configurations, including the test of parallel-polarized antennas for data collection, will be tested to map the ice bottom, reducing the effect of clutter on GPR data quality.

**Author Contributions:** Conceptualization, C.C. (Chiara Colombero), C.C. (Cesare Comina), E.D.T., D.F. and A.G.; Data processing, C.C. (Chiara Colombero) and E.D.T.; Funding acquisition, A.G.; Investigation, D.F.; Methodology, C.C. (Chiara Colombero), C.C. (Cesare Comina) and A.G.; Supervision, C.C. (Cesare Comina) and A.G.; Validation, C.C. (Chiara Colombero); Visualization, C.C. (Chiara Colombero); Writing—original draft, C.C. (Chiara Colombero); Writing—review and editing, C.C. (Chiara Colombero), C.C. (Cesare Comina), E.D.T., D.F. and A.G.

**Funding:** This research was partially supported by educational funds from Alta Scuola Politecnica Milano-Torino (ASP project DREAM, DRone tEchnology for wAter resources Monitoring).

**Acknowledgments:** The geophysical campaigns described in this study were carried out in the framework of the Climate Change Glacier Lab of the Department of Environment, Land and Infrastructure Engineering of Politecnico di Torino. The authors are grateful to the Alpine guides that supervised and helped the field operations.

**Conflicts of Interest:** The authors declare no conflict of interest.
