Geophysical surveys for non-invasive characterization of sinkhole phenomena: A case study of Murisengo

The present research investigates the morphology and genetical mechanism of a sinkhole which occurred in 2019 in Murisengo (NW Italy). This landform is representative of several subsidence phenomena that often concern the Monferrato area (NW Italy). In concomitance with the appearance of the sinkhole at the surface, a cone of detrital material was found in the drifts of a nearby underground quarry. A geological survey was performed in the underground quarry in order to understand the interaction between the geological and geostructural features of the rock body and the generation of the sinkhole. Moreover, the underground sinkhole morphology was investigated through electrical resistivity tomography (ERT) surveys performed at the surface. The ERT outputs were combined to obtain a 3D image of the phenomenon and the 3D reconstruction was then compared with the geomorphological and structural setting of the area. Results suggest that a viscoplastic flow of clay-rich sediments within a conduit in the gypsum bedrock (suffusion process) generated the sinkhole.


| INTRODUCTION
Geological areas consisting of highly soluble rocks (e.g., evaporites or carbonates) often develop karst-related morphologies (e.g., caves or long conduits in the underground and sinkholes or dolines at the surface) that may pose multiple hazards. These phenomena need to be addressed through specific investigation methods for risk mitigation, taking into account the peculiarities of karst settings (De Waele et al., 2011;Gutiérrez et al., 2014;Parise, 2010). Karst-related phenomena are often directly or indirectly induced by anthropogenic activities. The increase in human occupation of karst terrains and the frequent increase of hazards caused by anthropogenic alterations result in scenarios of continuously rising impacts and risk (De Waele et al., 2011). National and local authorities are becoming increasingly aware of the sinkhole hazard, and therefore promote risk-assessment campaigns based on multiple techniques (Argentieri et al., 2015).
Risk scenarios may be exacerbated in cases of interference with underground tunnels for civil or mining purposes (Coli et al., 2020).
The intersection of underground drifts with karst morphologies may result in unexpected incidents, damages and environmental consequences such as depletion of water resources, permanent lowering of the water table, extinction of local springs and subsidence phenomena at the surface (Golian et al., 2021;Milanovic, 2002). Case histories from around the world have been detailed and described by many authors (e.g., Bonetto et al., 2008;Clay & Takacs, 1997;Coli et al., 2020;Day, 2004;Golian et al., 2021;Hou et al., 2016).
The most common type of surface subsidence morphology in karst areas is the sinkhole. Subsidence sinkholes have a typical subcircular shape, diameters up to hundreds of metres and depths ranging from a few metres to tens of metres. They commonly occur in evaporites, due to their high solubility and low mechanical strength (Gökkaya et al., 2021;Gutiérrez, Cooper, Johnson 2008), and less often in carbonates (De Waele et al., 2011). The weakening effect of water on the strength and rheology of evaporite rocksusually more significant than on carbonates (Caselle et al., 2022)additionally increases the hazard of these phenomena in evaporite contexts.
Subsidence sinkholes in evaporite subsoils can be classified on the basis of the material involved (cover, bedrock or caprock) and the dominant process (collapse, suffusion or sagging) (Gutiérrez, Guerrero, Lucha 2008. In particular, the processes can be defined as follows: • Sagging. The gradual settlement of the overlying cover by passive sagging or bending. • Suffusion. Depending on the features of the cover sediments (granular or cohesive), this may consist in a progressive downwashing transport of the cohesionless cover or in a downward migration of clay-rich sediments as a viscoplastic flow. In general, cover suffusion sinkholes do not form catastrophically, but gradually, and have funnel-or bowl-shaped geometries, with typical diameters of a few metres.
• Collapse. This develops when the cover consists of cohesive deposits with brittle rheology. It describes a progressive upwards migration of an initial arched cavity over the karst conduit. When the upwards collapse of the cavity roof intercepts the ground surface, the sinkhole is abruptly created. In this case, sinkholes may develop over dissolution pipes of relatively small diameter located at very high depths (e.g., karst conduits 60 m deep and <1 m wide).
The characteristics of sinkhole genesis and evolution (e.g., velocity of phenomenon propagation) are strictly related to geological and hydrogeological settings and to anthropic interactions (if any). A deep knowledge of these aspects therefore provides useful information for risk assessment and mitigation in a specific area.
In recent years, many methods for sinkhole risk mitigation have been developed, including: (i) methods that use subsurface sensors (e.g., conventional seismic stations, nano-seismic monitoring, borehole extensometers and reflectometry techniques; Dahm et al., 2011;Gutiérrez et al., 2019;Land, 2013); (ii) remote sensing methodologies (e.g., radar interferometry for the evaluation of subsidence rates, airborne laser scanning and photogrammetry; Galve et al., 2011;Gutiérrez et al., 2019); (iii) methods that use apparatus in direct contact with the ground surface (e.g., trenching for the precise delimitation of specific sinkholes, ground-based interferometric measures, high-precision topographic profiling, Differential Global Navigation Geophysical approaches have also been used as alternative methods of investigation, with the advantage of reducing uncertainties related to superficial anthropogenic or natural modifications (Argentieri et al., 2015;Youssef et al., 2020). The material contrast between the sinkhole sediments and host rock (e.g., density, electrical resistivity, electrical permittivity) makes it possible to identify limits and underground geometries of pre-existing karst, even in the absence of surface evidence (e.g., Kaufmann  However, geophysical results are strongly dependent on the particular features associated with karst processes, such as sinkhole fillings, decompaction of underground materials, water table changes, denser vegetation growth and/or structural and geometrical changes of the underground units affected by cavity propagation (e.g., Pueyo Anchuela et al., 2015).
Among the geophysical techniques available, electrical resistivity tomographies (ERTs) have often provided reliable results with the advantages of reduced costs and processing time. In particular, the boundary between host rock and sinkhole material can often be identified as high-or low-resistive sectors, as a consequence of the presence of water, clay-rich sediments and/or fractured and loose rock (with high-resistive anomalies if fissures are air filled or low resistive if they are water/clay filled). Nevertheless, the usefulness of the ERT technique must be evaluated case by case, also in relation to the genetical sinkhole process, the expected features of sinkhole materials and the features of the host rock.
Due to the presence of huge volumes of Messinian gypsum and evaporite sediments, the area of Monferrato (Piedmont Region, NW Italy) is susceptible to subsidence, collapses and surface karst events (Banzato et al., 2017;Bonetto et al., 2008;Vigna et al., 2010Vigna et al., , 2017. Moreover, many gypsum quarries are located in the area and some of these phenomena might be associated with gypsum exploitation. In     Profiles 1a and 2a (Figures 6a and b) Figures 6a-c). These sectors represent soils with high hydraulic conductivity and (probably) high saturation degree that might be associated with potential flow paths, in agreement with the superficial topographical evidence and the presence of material in the underground drifts of the quarry.

| DISCUSSION
Following the geomorphological sinkhole classification developed by Gutiérrez, Guerrero, Lucha (2008, we propose an interpretation of the Murisengo sinkhole based on the cover suffusion genetic model ( Figure 7). As explained and discussed in the previous paragraphs, the initial geological context is represented by an irregular gypsum body, consisting of big blocks differently oriented and separated by a fine-grained gypsum-rich matrix (i.e., the Valle Versa Chaotic Complex), overlaid by a clay-rich cover up to the ground surface ( Figure 7a). This initial geological context is further perturbated by the presence of an underground infrastructure (i.e., quarry tunnels). Due to the heterogeneity of the rock mass in terms of material strength and competence, the disturbance created by the quarry is not uniform throughout the tunnels' extension. In addition, as observed by the ERTs, the contact between the gypsum rock mass and the cover sediments is strongly irregular in consequence of the block-in-matrix organization of the orebody. The irregular geometry of the gypsum-cover contact implies that the quarry tunnels are separated by the clay-rich cover material by a random and unpredictable thickness of rock. A potential interference between the underground tunnels and the surfaces will be more probable in the portions with the thinnest rock mass septum between drifts and cover material and/or in the portions with a prevalently matrix-supported rock mass.
Following the proposed interpretation, the process is triggered by water infiltration in the covering soil. The presence of a topographical depression in the area where the sinkhole occurred concurs with the increase of water accumulation at the surface and infiltration in the sediments (Figure 7a), inducing an internal erosion that generates some preferential paths within the sediments, until the evaporitic deposit ( Figure 7b). Due to the irregularity of the gypsum-cover contact, the water tends to concentrate in correspondence with a natural depression of the roof of the gypsum body. This brings it to the opening of a karst conduit (and/or to the infiltration of water along permeability contrasts between gypsum blocks and matrix). Here, the thickness of the gypsum septum at the top of the quarry drifts has its minimum thickness. Hence, the conduit reaches the drifts (Figure 7c).
The further contribution of meteorological water eventually saturates the clay-rich cover sediments that slowly flow through the karst conduit ( Figure 7d).
The features of the detrital cone found in the quarry drifts (not completely disarticulated and with evidence of strike lines; see In Figure 8a, the yellow surface represents the tridimensional reconstruction of the contact between gypsum bedrock and clay-rich sedimentary cover (i.e., the electrical resistivity isosurface that corresponds to the red dashed line in Figure 6). The upper limit of the gypsum body has a main dip direction towards the NE (red square in Figure 8a). This NE-dipping orientation that brings the gypsum to outcrop in the SW portion of the survey area (as represented by the hole in the yellow isosurface) is consistent with the mean orientation of stratigraphical and structural surfaces measured in the site ( Figure 4) and with the mean orientation of the main regional tectonic structures (Villadeati structure, Figure 1b).