Based on the results above, a conceptual model of flow dynamics is shown in Figure 11 and justified in the following discussion section.
Bonini (2008) presents a geological model where pressurized fluids move up through discontinuities in the Ligurian Units, and accumulate in shallower reservoirs controlled by the lithological boundary between the impermeable claystone FAA and the underlying, more permeable, Epi-Ligurian Units and Colombacci Fm deposits. Bonini (2007) stated that anticlines in combination with the brittle structures associated with folding provide an efficient system for trapping and transferring fluids to the surface, where overpressured fluids are periodically expelled from the main reservoir (Marnosa-Arenacea Fm) through the creation or the reactivation of pre-existing fractures/faults (Figure 11d). Permeability contrasts, tectonic loading and gas generation likely represent the main factors triggering fluid overpressures (Bonini, 2007). Vannoli et al. (2021) suggest that a seal-bypass system, such as extrados fracture corridors developed along the fold crest, is needed to allow fluids to reach the surface. Furthermore, they claim that mud volcanoes can persist if they are connected by a network of deep long-lived structures possibly associated with background seismicity.
Our data suggest the presence of small shallow aquifers between 4-20 m depth where rising gas, from deep structures in the Epi-Ligurian units (Capozzi and Picotti, 2010), is trapped and temporarily stored. The impermeable FAA (in which the shallow aquifers are encased) allows pore fluid pressure to build up till this overcomes the tensional strength of the rock, so that gas can escape along fractures or fractures-aligned pipe conduits, reaching the surface and dragging and fluidifying solid material along the way up (Figure 11d). This flow dynamics allows both clay particles, gas, and eventually some silt (from the shallow aquifers) to rise to the surface with formation of gryphons and mud pools. As most of the mud pools are located around the gryphons, it is suggested that the overburden of the tall gryphons causes collapse and fracturing through which the fluids migrate, mixing with shallow meteoric waters (Mazzini, 2009; Mazzini and Etiope, 2017). Most of the gas is rising from deep reservoirs (gas dominated fluid) and, perhaps, some deep connate water is carried along with the gas. The fluids reaching the surface may also mix with brackish water, which is present in the shallow aquifers and this justifies the EC values measures in the field.
The presence of local shallow aquifers is confirmed by particle size distribution of mud samples collected in the field (Figure 10) showing coarser granulometry than clay sediments. The presence of coarse sediments (sand and silt) within FFA is also confirmed by the AGIP S.p.A. core logs near the study area (Maranello 001, Levizzano 001 and 002, refer to Figure 1a for location) and the explanatory notes of the Geological Maps of Italy (Scale 1:50000, Sassuolo sheet, RER, 1999). Here, the FAA is described as consisting of silty and slightly marly clays, with a thin to medium stratification marked by intercalations of fine sands in flat, isolated, or connected lenses. On the right of the Secchia River, where Nirano is located, the sandy levels become thicker, laterally continuous and the stratification more powerful. Core logs show clays with frequent sandy and silty intercalations within the first 100 m of the FAA sequence.
Sciarra et al. (2019), who carried out extensive geochemical soil gas and exhalation fluxes (CO2 and CH4) surveys, indicate the presence of high permeability areas that act as preferential leakage pathways for gas migration. These areas positively correlate with the dome-shaped conductive anomalies mapped by Lupi et al. (2016) at 20 m depth. We suppose that these permeable areas are none other than shallow aquifers with variable size and thickness possibly leaking to the surface along circular faults formed during the collapse of the area and the formation of the caldera-like morphology (bowl). The salinity of the muds is low (around 7 g/l) and well-correlates with the salinity of formation waters recovered during DST tests in the Levizzano 1 and 2 wells, as well as Maranello 1 from sandy layers within the FAA. The mud reaching the surface also does not contain any microfossil older that the age of the FAA, suggesting no direct mass transport from below them (Papazzoni, 2017).
The connection between shallow aquifers is variable in time and depends on the gas flow activity. The measurement of conduits depths indicates that they range varies from 0.5 to 5 m with a mode around 1.5 m; at N2, however, we measured a depth of 8 m and 15 m in two distinct occasions. Our observations suggest that conduits opening is variable through time, some conduits may close and then reactivate; the whole conduit network appear to be in a state of continuous change and individual conduits are temporary features. Furthermore, the observations of Kopf (2002) that mud pools conduits are larger than those of gryphons is also confirmed by our observations. The N2 mud pool, in fact, is the only one where we could lower our sounding line in several points without any impediment and it is the one where we have reached the largest depth (15 m).
Correlation coefficients of mud levels variations between volcanoes depend on the conduit geometry, type, as well as their connection (i.e., direct connection and degree of tortuosity). High correlation coefficients between ML vents (Table 2 and Figure 4) could be explained by the connection to the same source whereas low correlation coefficients may represent separation of the conduits feeding the different pools and mud vents.
In the Nirano system, ML and Patm are not correlated (Table 2, Figure 4) and this is further confirmed by the barometric efficiency analysis in Figure 7. The influence of barometric pressure on a groundwater surface can follow several scenarios. In confined aquifers, the change in water level is caused by a change in the force applied to the Earth's surface by the atmospheric pressure—higher atmospheric pressure causes a greater load, which is transferred to water at depth, causing the water pressure within the aquifer to rise (Rasmussen, 2005). The unconfined aquifers show delayed responses due to the delay in transmitting the atmospheric pressure signal through the vadose zone to the water table surface (Spane, 2002). If the system were affected by barometric-pressure change, the slopes of the fitted linear curves in Figure 7 would be positive. As it can be seen, the coefficients of determination (R-squared values) are all small, indicating no significant barometric efficiency (Figure 7). Gas storage and overpressure in small shallow aquifers and flow from those to the surface seems more important than the barometric pressure change.
The mud level variance (Figure 6) is an indicator of gas activity in the monitored mud pools. Peaks appear following periods of quiescence during which gas accumulates and overpressure increases in the small aquifers system. The trend of these values indicates a chaotic system characterized by non-constant gas flow rates, with an alternation of quiescence and extrusion activity periods (Figure 4 and Figure 6). This is further confirmed by some drawdown tests carried out on site (by emptying the N3 mud pool), during which no linear recovery of mud level was observed, testifying the presence of a non-continuous and constant flow in the conduits.
The different mud levels measured at the vents seem to be controlled by the variable gas-liquid ratio in the mud conduit (Figure 8 and Figure 9). Low mud levels correspond to the vents that do not allow gas accumulation (fast continuous degassing) in their conducts and their pressure head is dominated by mud density (low gas-liquid ratio) whereas volcanoes whose conduits are occupied by many gas bubbles that move slowly (high gas-liquid ratio) have a higher mud level dominated by the gas volume. The latter are more likely to erupt mud suddenly and unexpectedly than the former type. In Figure 9, mud level and gas flow data highlight two main groups (N1-N14 on the Eastern side, and N15-N18 on the Western side of the Nirano Salse), which could indicate two different shallow aquifers at different depths (Figure 11d). Neighbouring volcanoes have comparable but not equal mud level values. The mud viscosity is variable and heterogeneous, depending on climatic conditions, depth of measurements along the conduits, and, more importantly, gas flow. Also mud density is heterogenous and as gas flow increases, density will decrease along with viscosity. However, as shown in Table 2, gas flow is not the only cause. Dilution with rainwater and evaporation during hot and dry periods could contribute to density variability.
One other issue important for discussion is the source of fluids (gas and water) at the Nirano Salse. According to Kopf (2002) the possible fluid sources for overpressuring and mud extrusion can be summarized as: (1) pore fluids from compaction; (2) biogenic methane from degradation of organic matter; (3) fluid migration along deep-seated thrusts; (4) thermogenic methane; (5) fluids from mineral dehydration; (6) hydrothermal fluids, and alteration of crustal rock; (7) fluid expulsion from internal deformation within the diapiric intrusion. By reviewing the mechanisms above on the basis of what we know for the Nirano Salse and our observations, we can argue that: (1) fluid expulsion during compaction has already occurred at Nirano, because the FAA already went through primary compaction; (2) biogenic methane from degradation of organic matter is not supported by the geochemical isotopic characteristics of the gas analyzed (Martinelli et al., 2012); (3) fluid migration along deep-seated thrusts is possible for the gas (deep source of Martinelli et al., 2012); furthermore, there might have been mixing between deep connate and shallower aquifer waters; 4) the methane is of thermogenic origin as suggested by Oppo et al. (2017), Martinelli et al. (2012), and Tassi et al. (2012); (5) there are only traces of smectite in the Nirano mud suggesting no dehydration reactions involving opal-smectite reactions to cause overpressures (Vezzalini et al., 2017); (6) fluids in Nirano do not have hydrothermal characteristics (Martinelli et al., 2012); (7) fluid expulsion from internal deformation within the diapiric intrusion does not fit well Nirano Salse dynamics and formation given that the gas source is below the FAA. Furthermore, mud diapirs are interpreted today as areas of seismic attenuation caused by intrusion of mud dike swarms formed by hydrofracturing due to overpressures (Figure 11d) and not by the density driven phenomena of viscous flow that are typical of salt diapirs (Tingay et al., 2009).
By considering the possible fluid sources that we discussed above and our observations, we suggest that the reason for fluid release at the Nirano Salse is leakage of a hydrocarbon seal (Figure 11d). Abnormal overpressure in a deep reservoir (Marnoso-Arenacea Fm) would be generated by gas accumulation following secondary migration. Fluids could escape from a seal broken by a fault, or gas may leak from the spill-point of a faulted reservoir layer. The seal could also have a valve behavior and fail when the overpressure in the gas reservoir increases due to the continuous gas migration from below. Gas following upward migration routes, such as faults and fractures (Figure 11d), would accumulate in shallow aquifers confined within the FAA and then be released when the fluids overpressure would exceed the tensional strength of the seal (Gibson, 1994). In this way, the conduits forming the mud volcanoes would start at the depth of this shallow aquifers (5 to 30 m from the surface) as also measured by our soundings. Conduits may have different shapes (cylindrical to fracture-like) and may be interconnected where mud volcanoes are in proximity as also suggested by our mud level correlations. The system of fractures and conduits feeding from the deep source into the shallow aquifers cannot be assessed by our work and is better addressed by geophysical methodologies.