Soil-gas survey of liquefaction and collapsed caves during the Emilia seismic sequence

The epicentral area of the Emilia seismic sequence is located in the Emilia-Romagna Region (northern Italy), 45 km from the city of Modena (Figure 1). This area is sited within thrust-related folds of the Ferrara Arc, which represent the most external part of the northern Apennines. This sector is considered as having been active during late Pliocene to early Pleistocene times [Scrocca et al. 2007] and encompasses also the Mirandola and Ferrara seismogenic sources [e.g., Burrato et al. 2003, Boccaletti et al. 2004, Basili et al. 2008]. The main sedimentary infilling of the Po Plain is represented by Pliocene–Pleistocene alluvial deposits (alternating fluvial sands and clays) that overlie a foredeep clastic sequence, with a total average thickness of 2 km to 4 km [e.g., Carminati et al. 2010]. Soon after the mainshock, several liquefaction phenomena coupled to ground fractures were observed in the epicentral area (e.g., San Carlo, Ferrara). Soil liquefaction is a phenomenon in which the strength and stiffness of a soil is reduced by earthquake shaking or other rapid loading. […] Collapsed caves reported in the literature and/or local press [e.g., Febo 1999, Martelli 2002] in the epicentral area were previously investigated by our research group in 2008, with several soil measurements of CO2 and CH4 fluxes. Immediately after the May 20, 2012, mainshock and during the Emilia seismic sequence, the collapsed caves were sampled again to determine any variations in these CO2 and CH4 fluxes. In this survey, newly formed collapsed caves were also found and measured (especially in the northern part of investigated area). […]


Introduction
The epicentral area of the Emilia seismic sequence is located in the Emilia-Romagna Region (northern Italy), 45 km from the city of Modena (Figure 1).This area is sited within thrust-related folds of the Ferrara Arc, which represent the most external part of the northern Apennines.This sector is considered as having been active during late Pliocene to early Pleistocene times [Scrocca et al. 2007] and encompasses also the Mirandola and Ferrara seismogenic sources [e.g., Burrato et al. 2003, Boccaletti et al. 2004, Basili et al. 2008].
The main sedimentary infilling of the Po Plain is represented by Pliocene-Pleistocene alluvial deposits (alternating fluvial sands and clays) that overlie a foredeep clastic sequence, with a total average thickness of 2 km to 4 km [e.g., Carminati et al. 2010].
Soon after the mainshock, several liquefaction phenomena coupled to ground fractures were observed in the epicentral area (e.g., San Carlo, Ferrara).Soil liquefaction is a phenomenon in which the strength and stiffness of a soil is reduced by earthquake shaking or other rapid loading.Liquefaction generally occurs in saturated unconsolidated sediments (e.g., sand, mud, and artificial fill) that lose their shear strength [Hazen 1920].As a consequence, liquefied soil cannot support differential stress, thus causing ground failure and damage to the built environment.
Several soil measurements of gas fluxes (CO 2 and CH 4 ) and concentrations were performed on liquefactions and ground fractures located in the Finale Emilia (Modena) area (Via Fruttarola and Santa Bianca) and the Ferrara area (Renazzo and San Carlo) (Figures 1, 2) to determine whether these diffuse phenomena can be correlated with deep fluid migration through preferential leakage pathways linked to the earthquake.
To determine the possible leakage induced by the seismic stress during the Emilia sequence, collapsed caves in the epicentral area were also sampled.These collapse phenom-ena are linked to gas escape, and have been known since the 1970's in some tectonically active areas of the southern Po Plain [Bonori et al. 2000].Individual phenomena occur as localized depressions of the soil in the shape of the cavity, or an 'inverted funnel', or as wide slits that are broad and up to few meters deep (Figure 3).Collapsed caves are considered as superficial events that are likely to have been triggered by compaction of organic-matter-rich soils (e.g., peat) [Castellarin et al. 2006].Complex microbial (bacteria) reactions transform the peat, resulting in volume loss and a consequent slight ground subsidence.Collapsed caves generally develop in orchards, mainly due to the loss of cohesion of the soil, its extreme imbibition, or the transit of agricultural vehicles.
Collapsed caves reported in the literature and/or local press [e.g., Febo 1999, Martelli 2002] in the epicentral area were previously investigated by our research group in 2008, with several soil measurements of CO 2 and CH 4 fluxes.Immediately after the May 20, 2012, mainshock and during the Emilia seismic sequence, the collapsed caves were sampled again to determine any variations in these CO 2 and CH 4 fluxes.In this survey, newly formed collapsed caves were also found and measured (especially in the northern part of investigated area).

Methods
CO 2 and CH 4 fluxes were measured by the speedportable 'closed dynamic' accumulation chamber 'time zero' method [e.g., Cardellini et al. 2003] using a West System instrument equipped with CO 2 and CH 4 infrared detectors.The recorded concentrations measured over time, combined with other parameters such as volume and surface of the accumulation chamber, allowed the calculation of the exhalation flux from the soil [e.g., Hutchinson et al. 2000].
Soil-gas samples were collected using a steel probe that was driven into the ground to a depth of 0.6 m to 0.8 m, to avoid the major influence of meteorological variables [e.g., Hinkle 1994].The soil-gas concentrations (CO 2 , CH 4 , He)
were analyzed in the laboratory using a MicroGC Varian 4009 CP, equipped with thermal conductivity detectors.Radon was analyzed immediately in the field, due to its halflife (3.8 days), using a RAD7 Durridge alpha spectrometry instrument, at a depth of 70 cm.

Soil liquefaction and ground fractures
In the epicentral area (e.g., San Carlo, Ferrara) soil liquefaction and sand blows were coupled to ground fractures that also showed noticeable horizontal and vertical displacements, and were observed at sites with young alluvium.A mixture of gray-colored fine particle materials and water bubbled up into streets, parks and fields, and even through the concrete floors of buildings.
The soil liquefaction and ground fractures followed two preferential alignments (N60W and N140W) which can be related both to the main directions of the buried fold axes and to paleo-river bed structures in the area.
Measurements of fluxes (zCO 2 and zCH 4 ) and soil-gas concentrations (CO 2 , CH 4 , He, 222 Rn), as well as the main statistical parameters, are reported in Tables 1 and 2, respectively.These data were compared both to previous soil-gas measurements performed by the authors in 2006 in the area between Rivara and Massa Finalese (Modena) (unpublished data), and to two case studies in central Italy [Annunziatellis et al. 2008] and in foredeep basins [Ciotoli et al. 2007].
The CO 2 concentrations after the May 20, 2012, earth-quake decreased with respect to the Rivara 2006 ones, aligning with those reported in Annunziatellis et al. [2008].
The He and 222 Rn contents did not show any remarkable variations when compared to the 2006 data, and they were lower than the concentrations measured in other Italian sites [Ciotoli et al. 2007, Annunziatellis et al. 2008].Negative He values (i.e., values lower than the atmospheric reference) constituted the bulk of our dataset.In spite of what was claimed by Reimer [1990] and Duddridge et al. [1991], negative anomalies did not appear to be linked to tectonic or morphological features.Several studies have reported He values below the air concentrations [e.g., Reimer 1980, Lombardi andVoltattorni 2010], which suggests a shallow origin of this gas.Therefore, negative He values can result from a disequilibrium between the soil gases and the atmosphere, as a consequence of differential mobility of the gaseous species involved [Ciotoli et al. 1999].
Radon is generally used as a tracer to provide a qualitative idea of gas transfer (velocity and flux), and its characteristics allow it to be used as a tool for mapping active faults in seismotectonic environments.In our samples, radon showed low values and was very similar to the Rivara data, indicating an absence of any deep fluid leakage.
CH 4 showed mean and median values clearly higher than the Rivara 2006 data (224.61,6.01 and 14.65 ppmv/v, respectively).The highest CH 4 concentrations were measured for the ground fractures at San Carlo (890 ppm), and for the soil liquefaction in Via Fruttarola (434 ppm) and Renazzo (338 ppm).
San Carlo showed the highest CH 4 values, which were  not correlated with other pathfinder elements (e.g., 222 Rn and He) [Lombardi and Voltattorni 2010].This might suggest a local anomaly, which would be likely to be due to surficial layer compression during the earthquake.
For Via Fruttarola and Renazzo liquefactions, the high CH 4 , CO 2 and 222 Rn concentrations were correlated with each other (Table 1).Moreover, high values of zCO 2 and zCH 4 were well correlated with CH 4 at Renazzo.These positive correlations among various gaseous species support the theory that CO 2 acts as a carrier for trace gases like radon [Durrance and Gregory 1990, Hermansson et al. 1991, Etiope and Lombardi 1995, Quattrocchi et al. 1999, Beaubien et al. 2003, Ciotoli et al. 2005, Lombardi and Voltattorni 2010].
The Table 2. Flux and soil-gas statistics for liquefactions and ground fractures in the epicentral area, compared both to previous soil-gas statistics obtained by the authors in 2006 in the area between Rivara and Massa Finalese (Modena) (unpublished data), and to the statistics obtained for two case studies in central Italy and in foredeep basins.*, soil-gas statistics taken from Annunziatellis et al. [2008]; # , radon data measured in foredeep basins from Ciotoli et al. [2007].standard).The concentrations of lighter hydrocarbons were below the detection limit (2 ppm) in all of the samples, which suggested a low-temperature origin of CH 4 (i.e., shallow and biogenic production).

Data
The flux measurements of CO 2 and CH 4 after the mainshock showed the same trends as the soil-gas concentrations.The zCO 2 values fit those measured in 2006, while the zCH 4 mean and median were higher.
The CO 2 values were within the typical range of vegetative exhalation of the cultivated soil [Baldocchi and Meyers 1991], minimizing its provenance from depth.The increased methane fluxes can be linked to the methane concentrations, and can be explained by the presence of peat layers in the most shallow strata.
EMILIA SEISMIC SEQUENCE: SOIL-GAS SURVEY

Collapsed caves
Collapsed caves (Figure 3) in the epicentral area were sampled both in June 2008 and June 2012, with stable and dry weather conditions.The 2012 measurements were repeated in the same spot as those of 2008 when this still existed after four years, as well as in newly discovered collapsed caves.
All of the data were processed with a statistical approach using normal probability plots, to define the statistical populations for each parameter, and to compute the contour maps using experimental kriging (Figure 4).
The spatial distributions of the soil-gas concentrations and fluxes measured in 2008 and 2012 are shown in Figure 4.A comparison between the CO 2 fluxes of 2008 (Figure 4A) with those measured in 2012 (Figure 4B) shows a remarkable increase over time.The areal distribution of anomalous values is very similar, but the maximum CO 2 flux changed from 70 g/m 2 day to 220 g/m 2 day.The zCH 4 showed the greater variations, going from 30 g/m 2 day to 2200 g/m 2 day.These higher zCH 4 values in both 2008 and 2012 were in the southern part of the investigated area, close to the Panaro River (Ca' Bianca locality).
In the northern part of the study area (Villa Gardè locality), the anomalous CO 2 and CH 4 concentrations, which were higher than the Italian averages [Annunziatellis et al. 2008], corresponded to the maximum values of zCO 2 .
In the southern part, a positive correlation was highlighted between the anomalous CH 4 concentrations and the maximum values of zCO 2 .The highest CO 2 and CH 4 concentrations were found south of the Panaro River, between the Ca' Nuova and Palata Pepoli localities.The presence of anomalous values in collapsed caves close to the Panaro River suggests a surficial origin of these phenomena, which is likely to be due to redox processes in the alluvial sediments.Conversely, in the northern part of the investigated area, isotopic analyses aimed at determining the origin of the CH 4 were performed on a sample (CH 4 = 522.6 ppmv/v), which highlighted a prevalent shallow biogenic origin (d 13 C = -59.64‰vs. PDB; dD = -153.39‰vs. SMOW).Therefore, the anomalous gas concentrations in collapsed caves is likely to be correlated to decomposition of shallow peat and/or lignite layers, which produce CH 4 through microbial activity [Bonori et al. 2000].

Conclusions and remarks
Soon after the May 20, 2012, mainshock (M L 5.9) and during the Emilia seismic sequence of May-June 2012, geochemical field investigations were carried out into the epicentral area.
The soil-gas concentrations and flux measurements for liquefactions, ground fractures, and collapsed caves suggest a superficial origin of these phenomena.This is probably related to the stratigraphy of the shallowest layers of the Po Plain.The results gathered support the hypothesis that soil liquefactions are surficial phenomena [Bhattacharya et al. 2011] that affect only the shallowest layers of the ground (tens or hundreds of meters).
The results of the collapsed caves measurements show that the CO 2 had remained essentially constant with respect to the 2008 survey, while the CH 4 appeared to be higher after the seismic sequence.However, no hints of deep degassing can be inferred for the study area after the earthquake, as suggested by isotopical analyses carried out both on soil liquefaction and in collapsed caves.
The results obtained in this study constitute the starting point for subsequent geochemical surveys, which will be carried out over time, both on liquefactions and collapsed caves, to assess the temporal variations and to better understand the geochemical processes related to the seismic sequence.

Figure 2 .
Figure 2. A, B. Liquefactions with N60W direction observed in Via della Fruttarola-Finale Emilia (Modena) corn field and Santa Bianca (Modena), respectively.C. Sand blowout from a well in San Carlo (Ferrara).D, E. Ground fractures with a N140W direction observed in the San Carlo area (Ferrara) soon after May 20, 2012 (see Figure 1 for location map).Geographic coordinates UTM WGS 84 32N.
d 13 C analyses were carried out only in the San Carlo sample (over the minimum of detection for the analyses: 450 ppm), which indicated a prevalent biogenic origin (d 13 C = -67.25‰vs. Pee Dee Belemnite [PDB] standard; dD = -164.77‰vs.Standard Mean Ocean Water [SMOW]

Figure 3 .
Figure 3. A, B. Soil-gas and flux measurements in collapsed caves.B. Detail of the steel probe driven into the collapsed caves to collect soil-gas samples.Geographic coordinates UTM WGS 84 32N.

Figure 4 .
Figure 4. Collapsed cave contour maps in the Finale Emilia, Camposanto and Ponte San Pellegrino areas (Modena).A, B. zCO 2 measured in 2008 and 2012, respectively.C, D. zCH 4 measured in 2008 and 2012, respectively.E. CO 2 concentrations measured in 2012.F. CH 4 concentrations measured in 2012.Green dots, 2008 sampling points; red dots, 2012 sampling points.The areal distribution has different extents due to different numbers of sampling points in 2008 and 2012.

Table 1 .
Measurements of fluxes and soil-gas concentrations for liquefactions and ground fractures in the Finale Emilia (Modena) area (Via Fruttarola and Santa Bianca), and in the Ferrara area (Renazzo and San Carlo), during 2012 earthquake sequence.