Pure Appl. Geophys. Ó 2015 Springer Basel DOI 10.1007/s00024-015-1152-4

Pure and Applied Geophysics

Uranium Groundwater Monitoring and Seismic Analysis: A Case Study of the Gran Sasso Hydrogeological Basin, Italy MARTA CIARLETTI,1,2 WOLFANGO PLASTINO,1,2 ANTONELLA PERESAN,3,4,5 STEFANO NISI,6 LORENZO COPIA,1,2 GIULIANO F. PANZA,4,5 and PAVEL P. POVINEC7 Abstract—Uranium groundwater anomalies, observed before the L’Aquila earthquake (April 6th, 2009) and before the seismic swarm, which occurred in the second half of 2010, represent a key geochemical signal of a progressive increase of deep fluids fluxes at middle–lower crustal levels associated with the geodynamics of the earthquake. In this paper, temporal variations of uranium groundwater are studied in association with the seismic pattern around Gran Sasso National Laboratory (LNGS-INFN). The normalized seismic energy release and the number of earthquakes are analyzed in detail by means of monthly sliding time windows. They are compared with uranium anomalies to highlight any possible correlation. Key words: Uranium, seismicity, Gran Sasso, L’Aquila earthquake.

1

Department of Mathematics and Physics, Roma Tre University, Via della Vasca Navale, 84, 00146 Rome, Italy. E-mail: [email protected]; [email protected]; copia@fis. uniroma3.it 2 National Institute of Nuclear Physics, Section of Roma Tre, Via della Vasca Navale, 84, 00146 Rome, Italy. 3 Istituto Nazionale di Oceanografia e Geofisica Sperimentale (CRS), Via Treviso 55, 33100 Cussignacco, (UD), Italy. E-mail: [email protected] 4 Department of Mathematics and Geosciences, University of Trieste, Via E. Weiss, 4, 34128 Trieste, Italy. E-mail: [email protected] 5 SAND Group, The Abdus Salam International Centre for Theoretical Physics ICTP, Strada Costiera 11, Trieste, Italy. 6 National Institute of Nuclear Physics, Gran Sasso National Laboratory, S.S. 17/bis km 18?910, 67010 Assergi, (AQ), Italy. E-mail: [email protected] 7 Faculty of Mathematics, Physics and Informatics, Comenius University, 84244 Bratislava, Slovakia. E-mail: [email protected]

1. Introduction Earthquake prediction is one of the most aimed targets in Earth Sciences, mainly due to the social and economical importance of the problem. For valuable earthquake prediction, existence of an observable approved as a precursor is required. Nowadays, there are strong reasons to doubt the efficacy of many observables proposed in the past as precursory signals, because they can be often affected by their own variability, which is not linked to seismic features (GELLER et al. 1997; BENEDETTI et al. 2011). Our aim in searching for possible precursor has been to investigate a correlation between uranium (U) groundwater content and seismicity in the Gran Sasso hydrogeological basin. Such an analysis is one of the first studies, available so far, in this research field, mainly due to the uniqueness of the achievable observations in the study area. In fact, uranium groundwater content is monitored within the framework of the INFN’s scientific program Environmental Radioactivity Monitoring for Earth Sciences (ERMES) since June, 2008 (PLASTINO et al. 2010). The area under investigation was affected by a seismic swarm from October, 2008 to December, 2009, with the main shock occurring in L’Aquila at 01:33 UT on April 6th, 2009 (ML = 5.9; MW = 6.3). Gran Sasso National Laboratories (LNGS-INFN) are located inside the largest aquifer in central Italy, within the limestone formation of the upturned syncline, near the main overthrust fault. This complex hydrogeological system is characterized by separation of water masses belonging to two different creeks: the former, where the main underground laboratories are located, flows in well-drained

M. Ciarletti et al.

cretaceous formation, while the latter runs within not drained and poorly permeable dolomitic formations (PLASTINO et al. 2006). Depending on the path and on the percolation features, water masses are, therefore, characterized by different chemical–physical properties. The statistical cross-check of these chemical parameters at four different sites (E1, E4 and E3, E3dx) showed the presence of two different water groups (PLASTINO et al. 2009). The active normal faults of the central Apennines are commonly divided into two parallel sets, a western set and an eastern one. Some of the strongest historical earthquakes have been attributed to the rupture of the western set of normal faults in this area. On the contrary, the eastern set has not been activated in historical times (FALCUCCI et al. 2011). On April 6th, 2009 the earthquake occurred nearby L’Aquila, with normal faulting on an NW–SE oriented structure about 15 km long, dipping toward SW and located in the western fault set. The earthquake has been preceded by few months of increased seismic activity and it has been followed by many aftershocks, occurred in the same area (PLASTINO et al. 2011). Geological investigation identified the Paganica fault as the causative fault of the seismic event (FALCUCCI et al. 2011). Hydrogeological setting and groundwater flow system have been taken into account since they are essential in water content analysis. In fact, the LNGSINFN is located inside the hydrogeological system formed by the Gran Sasso and the Velino-Sirente mountains. In this paper, the local seismicity recorded in the considered region has been studied and it has been cross checked with U concentrations groundwater anomalies, to find out if there is a possible correlation between the time trend of uranium content and seismic events, i.e., monthly variation of energy release and number of events. We found that the uranium groundwater anomalies, observed before the main shock of the L’Aquila earthquake, were probably associated with geodynamic processes occurring before the earthquake, which triggered diffusion processes through the overthrust fault (PLASTINO et al. 2009).

Pure Appl. Geophys.

2. Environmental Setting 2.1. Seismogenic Sources The central part of the Apennines is characterized by extensional tectonics since the Pliocene epoch, with most of the active faults being normal in type and NW–SE trending. At a regional scale, fault systems and individual faults are organized into larger fault sets that run almost parallel to the Apennines chain axis (AKINCI et al. 2009). This complex system is described in the Database of Individual Seismogenic Sources (DISS), a repository of geologic, tectonic, and active fault data for the Italian territory. The Database highlights the results of several decades of research work, with special emphasis on data and conceptual achievements within the past 25 years. Seismogenic sources of all types are characterized based on the available literature or unpublished original work. This information is organized in summaries of published papers and commentaries on critical issues. The Individual and the Composite Seismogenic Sources are described by a full set of geometric, kinematic and seismological parameters. A typical Composite Seismogenic Source spans an unspecified number of Individual Sources (BASILI et al. 2008). The Individual Seismogenic Sources of central Italy, together with the composite Seismogenic Sources and the Debated Sources, are represented in Fig. 1. The area under investigation has been affected by a seismic swarm from October, 2008 to April, 2009, before the main shock of April 6th, 2009, MW = 6.3, ML = 5.9 (PLASTINO et al. 2011). The main shock was followed by a period of relatively intense seismic activity, associated with aftershocks occurrence. A further, moderate size, swarm occurred in the second half of 2010, NW of the main shock epicentral area, at a distance of about 30 km, from the Paganica fault. The main event in this cluster was an earthquake with ML = 3.5 on August 31st, 2010, accompanied by five earthquakes with magnitude ML C 3.0, and by about 100 additional events with ML C 2.0, which occurred within a small area of 5 km radius, from July to December, 2010. Hence, the swarm is characterized by a quite large number of

Uranium Groundwater Monitoring and Seismic Analysis

Figure 1 Map of central Italy [from Google Earth-CNES/Spot Image-Data SIO, NOAA, U.S. Navy, NGA, GEBCO (2013)]. Individual (yellow lines), Composite (orange strips) and Debated Sources (purple bold lines) in the central Italy, based on geological and geophysical data and characterized by geometric and kinematic parameters, identified by BASILI et al. (2008). The red dot represents LNGS-INFN

events with fairly small magnitudes (PLASTINO et al. 2013). 2.2. Hydrogeological Setting and Groundwater Flow System The hydrogeological system formed by the Gran Sasso and the Velino-Sirente mountains occupies a surface of 2164 km2. This complex system can absorb 714 mm of annual effective infiltration and it can release of 49 m2/s as a mean discharge (PLASTINO and BELLA 2001). The Gran Sasso massif is in the northern part of the system (BONI et al. 1986), and it holds a very large and deep aquifer, which consists of a series of minor aquifers separated by the main structural discontinuities with values of permeability up to 10-4 m/s (A.N.A.S.-CO.GE.FAR. 1980). The LNGS-INFN underground laboratories are located inside the Gran Sasso tunnel. The rocks in the measurement site are mainly composed by detrital limestones and, in particular, are crossed by the overthrust fault, which separates the limestones from the dolomite rocks (PLASTINO and BELLA 2001). Hydrogeological investigations showed that the main springs can be classified in different groups. After the

construction of the tunnels, the aquifer reached a new hydrodynamic equilibrium (BARBIERI et al. 2005). From a hydrogeological point of view, the LNGSINFN is located in the saturated zone in the core of the aquifer (AMORUSO et al. 2013). Based on the hydrogeological scheme by BONI et al. (1986), a polygon with 25 vertices can be outlined for delimiting the aquifer, to be used for the following analysis (Fig. 2). The arrows in the small panel of Fig. 2 represent the direction of the groundwater flow paths.

3. Analysis of Seismicity A reliable analysis of the time dependence of seismicity can be made only if the complete part of the seismic catalog is considered: this requires the removal of the smallest events, which are not systematically recorded. To guarantee the homogeneity of the catalog, only seismic events which occurred after April 16th, 2005 have been considered. In fact, on account of the magnitude heterogeneity in the instrumental earthquake catalogs available for the Italian territory over different time intervals, which has been evidenced by ROMASHKOVA and PERESAN (2013) and later confirmed by GASPERINI et al. (2013),

M. Ciarletti et al.

Pure Appl. Geophys.

Figure 2 Hydrogeological scheme of central Italy, modified from BONI et al. (1986). The white dotted line represents the Gran Sasso aquifer, symbolized by a 25 vertices polygon. The black dot represents LNGS-INFN. Groundwater flow lines are shown by arrow symbols in the small picture. Hydrogeological complex: light blue (1) is used for shallow alluvial aquifer, dark blue (2) for thick multistata alluvial aquifer. A pattern of red squares shows the carbonate infiltration areas (9, 10, 11, 12, 13), where effective infiltration prevails over both surface runoff and evapotranspiration. The mesh is closer where infiltration is higher. Yellow areas refer to volcanic complex (4) and marly–calcarentic complex (8), where the effective infiltration preferably occurs in the most fractured hard rocks. Violet refers to dolomitic complex (14). Effective infiltration seems to be comparable to that of volcanic complex (4). No data are available regarding surface runoff and evapotranspiration. Light gray (5) is used for areas where both surface runoff and evapotranspiration prevail over infiltration. Percolation and aquifer capacity are consequently negligible. Dark gray (7) shows the occurrence of limited effective infiltration and diffuse percolation. Hydrological symbols: blue filled circles identify springs, considered as a localized natural discharge of ground water emerging in a restricted area. Mean discharge is proportional to the diameter. Blue filled triangles identify linear springs, considered as a natural discharge of ground water emerging into a portion of a stream, which varies in length from a few hundred meters to some kilometers. Mean discharge is proportional to the size

we limited our analysis to the BSI bulletins, compiled since April, 2005 and available on line via the Istituto Nazionale di Geofisica e Vulcanologia (INGV) website (http://iside.rm.ingv.it). Earthquakes which occurred from April 16th, 2005 to October 1st, 2014 have been selected from the Italian Seismological Instrumental and Parametric Data-Base. The completeness of the catalog can be

visually determined from the frequency–magnitude distribution, that is considering the linear distribution of the logarithm of the number of earthquakes within each magnitude grouping interval DM (e.g., DIMRI 2005). The analysis has been performed on events that occurred in the circular region centered at LNGS-INFN with a radius of 100 km. Figure 3a shows the number of events versus magnitude for the

Uranium Groundwater Monitoring and Seismic Analysis

Figure 3 Frequency–magnitude distribution of the seismic events from 16th April, 2005 to 1st October 2014 (a). To test the stability of the threshold, the analysis is repeated considering two distinct periods: before (b) and after (c) the L’Aquila earthquake. The red line represents the completeness threshold Mmin = 1.5

whole period of seismic activity. In this case, the catalog can be considered fairly complete for events with magnitude M C Mmin = 1.5, below which a bending in the distribution is observed. To check whether the completeness of the catalog changed in connection with the occurrence of the L’Aquila main event, we tested the completeness of the catalog considering two different time intervals, before the L’Aquila earthquake, and after the L’Aquila earthquake: Fig. 3b, c guarantees the stability of the selected threshold. The spatial distribution of the events, from April, 2005 to the end of September, 2014, considering a circular region centered at LNGS-INFN with radius of 100 km, is shown in Fig. 4. The seismic activity, expressed as number of earthquakes over a regular grid of 0.05° 9 0.05°, is represented by a color scale from blue to red, similarly to PLASTINO et al. (2013).

At the coordinates 42°150 , 13°150 , the red cluster clearly represents most of the seismic activity which occurred near L’Aquila from February to December, 2009. The energy release, Ei, for each earthquake i with magnitude Mi has been calculated, and then normalized to the energy Emin of the minimum magnitude Mmin using the formula: Ei ¼

Ei 10cþdMi ¼ cþdM ¼ 10dðMi Mmin Þ min Emin 10

ð1Þ

The considered relationship between the energy and the magnitude of each seismic event has the form proposed by GUTENBERG (1956) and referred to as the standard energy–magnitude relation by KANAMORI (1977). The use of the normalized energy Ei allows to highlight the features of seismic activity, and to reduce the number of empirical parameters of the energy–magnitude relation, since only the coefficient

M. Ciarletti et al.

Pure Appl. Geophys.

Figure 4 Spatial distribution of the events with M C Mmin = 1.5, from April, 2005, to the end of September, 2014, considering a circular region centered at LNGS-INFN with radius of 100 km. The seismic activity, expressed as number of earthquakes over a regular grid of 0.05° 9 0.05°, is represented by a color scale from blue to red

d is required in the analysis. In this study, to be conservative in outlining the trend of the energy release rate, we use the value d = 1.5, given by GUTENBERG (1956). P The quantity E = i E*i , together with the numP ber of events, N = i Ni, has been calculated for monthly sliding time windows, as proposed by DE NATALE et al. (2004). Different possible selections of the investigation regions have been considered for a stable, as much as possible, characterization of the

seismic activity in space and time. Specifically, we considered a set of circular areas centered at the LNGS-INFN, which is the observation site for uranium groundwater chemical–physical properties. This rather simplistic choice is motivated by the need to relate the analysis of seismicity to the site where possible groundwater anomalies can be observed, rather than to the (a posteriori) knowledge of past seismicity.

Uranium Groundwater Monitoring and Seismic Analysis

Figure 5 Map of central Italy [from Google Earth-CNES/Spot Image-Data SIO, NOAA, U.S. Navy, NGA, GEBCO (2013)]. Circular regions centered at LNGS-INFN, R = 10, 20, 30, 40, 50, 60 km. The red dot and the blue dot represent LNGS-INFN and L’Aquila, respectively

3.1. Seismic Analysis on Circular Regions The main goals of the study of seismicity around the LNGS-INFN are the quantitative characterization of seismic energy release in space and time, as well as the investigation of possible correlation between the seismicity and uranium groundwater behavior. For that purpose, temporal variation of the number of events and normalized energy release within sliding time windows of 1 month duration have been calculated [NM(t) and EM(t), respectively], considering circular regions centered at LNGS-INFN (42.457°, 13.545°), with values for the radius ranging from 10 up to 60 km, i.e., R = 10, 20, 30, 40, 50, 60 km (see Fig. 5). The non-cumulative area charts allow a clear visualization of the number of events and the seismic energy release, as represented by NM(t) and EM(t) in Fig. 6 in each distance range. Considering the analysis of the entire period, we can observe that most of the seismic activity took

place within a distance from 10 to 30 km from the LNGS-INFN. Within a radius of 10 km, seismic energy release has been negligible, both before and after the L’Aquila earthquake, except for the first few months immediately following the main event. Up to 2 months before the L’Aquila earthquake, seismic activity was very limited within 20 km radius from the LNGS-INFN. Seismic activity around the LNGSINFN raised significantly a few months before the main event, i.e., starting on November, 2008 within R = 30 km, and starting on January, 2009 within R = 20 km. Accordingly, seismicity seemed to progressively concentrate around the epicentral area of the impending earthquake, i.e., closer to the LNGSINFN. After the earthquake, the region was characterized by a generalized increase of seismic activity, particularly at distances between 10 and 30 km from the LNGS-INFN. Furthermore, seismicity at distances larger than 50 km seemed not significantly affected by the L’Aquila earthquake: no evident increase has

M. Ciarletti et al.

Pure Appl. Geophys.

Figure 6 Number of events per month NM(t) and normalized energy release per month EM(t) are shown for the whole period by non-cumulative area charts, in each distance range, for circular regions centered at LNGS-INFN

been observed in the number of events for regions with R = 50 km and R = 60 km prior and after the event. As a rule, for distance ranges up to 40 km, seismicity kept still higher after the L’Aquila earthquake than before the event. 3.2. Identification of Time Intervals of Increased Seismicity Although concentric circular areas centered at the LNGS-INFN can be considered a good choice to outline, in an anticipatory perspective, the basic features of seismicity, a more refined analysis, considering specific properties of the study area, can be performed to get some new insight into the related geophysical processes. The region occupied

by the aquifer and the Seismogenic Sources is, therefore, taken into account to formally define the properties of seismic energy release and to identify periods of increased activity. For a better definition of the time periods of activity, i.e., to distinguish the seismic background from increased seismicity, the choice of a formal threshold is required (DE NATALE et al. 2004). The red outlined ten vertices polygon in Fig. 7 well approximates the 10 km boundary of the aquifer. We chose the region identified by the intersection between the red outline and the Composite Seismogenic Sources (orange strips in Fig. 7). We considered the threshold LE = EM = 356 for the normalized monthly energy release, corresponding to 30 % of the considered period of time, as in

Uranium Groundwater Monitoring and Seismic Analysis

Figure 7 Map of central Italy [from Google Earth-CNES/Spot Image-Data SIO, NOAA, U.S. Navy, NGA, GEBCO (2013)]. Composite Seismogenic Sources (orange strips), boundary of the aquifer (25 vertices black polygon). The ten vertices red polygon well approximates the 10 km boundary of the aquifer. The red dot represents LNGS-INFN. The area under consideration is the intersection between the red polygon and the orange strips

DE NATALE et al. (2004). This threshold allows to identify a limit for the monthly number of events: LN = NM = 35. The periods of activity are then defined as the time spans within which both the normalized energy release EM and the number of events NM exceed the threshold of 30 %, as shown in Fig. 8. Noticeably, the identification of the periods of activity is not overly dependent on the thresholds selection; for instance, if the threshold of 25 % is considered, the identified periods remain practically the same. We can thus identify four periods of increased seismic activity: Period A, from February, 2009 to December, 2009, which consists of the preparation phase, the main phase and the several aftershocks of the L’Aquila earthquake; from July to December, 2010, another period of relatively intense seismic activity is identified (Period B), with much less normalized energy release and number of events than the previous one; two further periods of small increase of seismicity have been identified from August to November, 2011 and from January to April, 2013 (Periods C and D). Nevertheless, the

selection criteria based on the Composite Seismogenic Sources intersection with the edge of the aquifer, as considered for the identification of periods of increased seismic activity, seem not so sensitive if compared with selection of the aquifer region only. In Fig. 9, we can see the geographic coordinates and the depth of the seismic clusters in Periods A, B, C and D, to show the scale of magnitudes and number of events, for a better comparison between a large seismic event and lighter seismic clusters. The sequence A in Fig. 9a affected the area under investigation from February to December, 2009. It consisted of almost 2800 recorded events with ML ranging from 2.0 up to 5.9 (magnitude of the main shock). The maximum magnitude associated with the earthquakes represented in Fig. 10b–d reaches values of ML = 3.4–3.6. Considering clusters B, C and D relative to both, the LNGS-INFN and the aquifer (Fig. 10), and comparing them with Fig. 7, some additional comments are relevant: the cluster B is located in the northwestern area of the LNGS-INFN (in a normal fault area), while the clusters C and D are positioned

M. Ciarletti et al.

Pure Appl. Geophys.

Figure 8 Time periods of activity defined by the thresholds LE = EM = 356 and LN = NM = 35, corresponding to 30 % of the considered period of time. The periods of activity are defined as the time intervals within which both the normalized energy release EM and the number of events NM exceed the threshold of 30 % (red lines). The purple dashed boxes represent Period A (February–December, 2009), Period B (July– December, 2010), Period C (August–November, 2011) and Period D (January–April, 2013)

north with respect to the LNGS-INFN, in a very close thrust fault region. As it could be observed from Fig. 9, the number of events associated with B, C and D is very different from that of A. The identified periods of the seismic activity well correlate with the time spans when the number of earthquakes exceeds that predicted by the Omori’s Law. The number of aftershock events in unit time interval after the main shock of a significant earthquake, N(t), decays as the inverse power law in elapsed time after the main shock: N ðtÞ ¼ k2 ðt þ k1 Þp

ð2Þ

where k1 and k2 are constant and the p value ranges from 0.5 up to 2.5, but normally comes closer to 1 (DIMRI 2005). We studied the aftershocks of the L’Aquila earthquake, considering seismic events which occurred in the circular region with radius of 30 km (R30), centered at the epicentral coordinates of the L’Aquila main shock. The residuals, i.e., the

difference between the observed value and the theoretical value for each month, are plotted together with the monthly number of events, evaluated after the L’Aquila earthquake (see Fig. 11). Positive residuals correspond to an increased number of events (as we can see in the case of June–December, 2010, February, 2011, August–November, 2011, November– April, 2013).

4. Comparison Between Seismic Activity and Uranium Groundwater Anomalies The uranium groundwater monitoring within the framework of the INFN’s scientific program Environmental Radioactivity Monitoring for Earth Sciences (ERMES) started on June, 2008 with the aim of better defining the 222Rn groundwater transport processes through the cataclastic rocks, as well as to check its contribution to the neutron background

Uranium Groundwater Monitoring and Seismic Analysis

Figure 9 L’Aquila earthquake and clusters B, C and D are represented in the three spatial coordinates (Longitude, Latitude and Depth), to better visualize the 3D position of each single event, as well as to show the different entities of the swarms. a, b, c and d are related to these periods: a L’Aquila earthquake, February–December, 2009; b cluster B (July–December, 2010); c cluster C (August–November, 2011); d cluster D (January–April, 2013). The size of the markers shows the magnitude of the events of each seismic swarm (ML ranges from 2.0 up to 5.9 for swarm A and from 2.0 up to 3.4–3.6 for clusters B, C and D)

at the LNGS-INFN (PLASTINO et al. 2010). In fact, neutrons from cosmic radiation are properly shielded by dolomite and flint rocks [1400 m of overlying

rocks are equivalent of 3100 meter of equivalent water (mwe)]. A contribution of neutrons from the spontaneous fission of uranium in rocks and in water

M. Ciarletti et al.

Pure Appl. Geophys.

Figure 9 continued

inside the underground facility is under investigation (PLASTINO et al. 2009). The uranium content detected in four different sampling sites showed different features (PLASTINO et al. 2009). Two distinct groups of water could be identified: the former (E3 and E3dx) has shown relatively high and strongly varying content of uranium,

and the latter (E1 and E4) has shown relatively low and weakly varying content of uranium (PLASTINO et al. 2011). The uranium groundwater content followed a seasonal trend, with the exception of some anomalous values observed until March 2009. From the end of the upper Miocene until the Pleistocene, a sequence of tectonic events produced a

Uranium Groundwater Monitoring and Seismic Analysis

Figure 10 Spatial distribution of the events with M C Mmin = 1.5. During the four identified time intervals A, B, C and D. The intensity of seismic activity, expressed as number of earthquakes over a regular grid of 0.05° 9 0.05°, is represented by a color scale from blue to red. The black dot represents LNGS-INFN. The aquifer is also shown

complex geological structure, crossed by three main faults (PLASTINO et al. 2010). The LNGS-INFN is located inside the largest aquifer of central Italy

within the limestone formation of the upturned syncline, near the main overthrust fault (PLASTINO et al. 2010). This fault actually divides water masses

M. Ciarletti et al.

Pure Appl. Geophys.

Figure 11 Number of monthly aftershocks after the L’Aquila earthquake for R30 versus time, plotted together with the residuals of the analysis. Markedly positive residuals correspond to an increased number of events

belonging to two distinct flow paths: the former (sites E1 and E4) flows in a well-drained cretaceous formation (where the main laboratories are excavated), while the latter (sites E3 and E3dx) is within not drained and poorly permeable dolomitic formations (PLASTINO 2006). Simultaneous measurements of the following quantities were performed for each water sample at the monitoring sites inside the LNGS-INFN since June 2008: contents of uranium, sodium, magnesium, potassium, calcium, electrical conductivity (EC), oxidation–reduction potential (ORP), pH and hardness of water. Collected groundwater is stored in 1 L cleaned polyethylene bottles, after 5 min of water flushing at maximum flow. The elemental analysis of water samples is performed by ICP-MS technique. Concerning site E3, two different regions are investigated (Fig. 12): E3 is parallel to the main overthrust fault, northward; E3dx is orthogonal to the fault,

Figure 12 Schematic view of the LNGS-INFN. The overthrust fault (gray line) and monitoring sites E1, E3 and E4 are also shown. At site E3, there are two sampling points: E3 which is parallel to the overthrust fault in the North direction; and E3dx which is orthogonal to the fault in the E4 direction into the cataclastic rocks

toward E4. Site E3dx is very significant, thanks to its position [it is the nearest site to cataclastic rocks (PLASTINO et al. 2013)]. It is also characteristic for a

Uranium Groundwater Monitoring and Seismic Analysis

Figure 13 U time series for sites E1 (blue), E3 (green), E3dx (red) and E4 (cyan), together with Period A (February–December, 2009), Period B (July– December, 2010), Period C (August–November, 2011) and D (January–April, 2013) (dashed purple rectangles). Uranium groundwater anomalies are identified before the L’Aquila earthquake in all four sites, differently from the other three seismic activations. The relative accuracy for U is always less than 5 %

better description of water–rock interactions through the main overthrust fault. In Fig. 13, U groundwater content in sites E1, E3, E3dx and E4 is shown. Periods A, B, C and D are plotted as dashed purple rectangles. In all four sites, we can observe that, before the L’Aquila earthquake, uranium content in groundwater sampled shows some spike-like anomalies estimated by outliers ±3r to the mean value of the time series. Such anomalies, already studied in detail by statistical analysis in PLASTINO et al. 2011, are also found before the beginning of Period B, only in site E3dx (end of June, 2010), while they do not precede Period C and Period D. A possible physical explanation of uranium anomalies observed before the L’Aquila earthquake might be in a progressive increase of deep CO2 fluxes at middle–lower crustal levels. Petrological and geophysical modeling explains how carbon is efficiently cycled in the upper mantle beneath Italy and the Western Mediterranean region at the Ma scale, via low-fractions of carbonate melts generated by melting of carbonate-rich lithologies of the subducted

Adriatic lithosphere, induced by the progressive rise of mantle temperatures behind the eastward-retreating subducting plate (FREZZOTTI et al. 2009). During the upwelling, part of the CO2 could not reach the surface and could remain trapped beneath the Moho and the lower crust, especially where continental crust is thick. The entrapment of this CO2 could origin overpressurized reservoirs, rich of noble gases, U, Th, K and other incompatible elements. These facilitate seismogenesis, in normal fault areas (PLASTINO et al. 2011). Fault mechanisms are highly influenced by fluids, by their pore pressure, their chemistry and the variation of friction along the fault plane. Fluids play a major role, triggering the rupture, and they are passively squeezed out during seismic events. A low amount of CO2 in zones of active faults is sufficient to give the aquifer efficiency on dissolving carbonate wall rock. Throughout the seismic process, the lower part of the crust is constantly subject to shear, while the upper crust is locked. During this time interval, the volume between the upper crust and the lower

M. Ciarletti et al.

crust should suffer dilatancy: fluids are expected to fill in the fractures formed during the interseismic period. This separation produces a weak volume of rocks that sustain the hanging wall until it suddenly falls down to refill the stretched volume. This process is shown by the main shock of an earthquake. The fluids previously introduced are squeezed out. After the collapse, the hanging wall is expected to gradually adjust to its new position, generating a long sequence of aftershocks, as those observed after the main event of April 6th, 2009 of the L’Aquila sequence. PLASTINO et al. (2011) proposed the progressive increase of deep CO2 fluxes at middle–lower crustal levels as a possible physical explanation of U anomalies observed before the L’Aquila earthquake, where the area is cross-cut by several NW–SE trending active normal faults. To further justify this hypothesis, new uranium observations are required. Differently from the swarms A and B, we did not observe any anomalous value in U quantities before the swarms C and D. This might be due to the different structures which have been involved in these four periods of increased seismicity: as we stated before, upwelling of CO2 is favored in normal fault areas (as in the case of the L’Aquila earthquake and Swarm B), while it is inhibited in thrust fault areas (see Figs. 7, 10).

Pure Appl. Geophys.

30 km. After the earthquake, the region was characterized by a generalized increase of seismic activity, particularly between 10 and 30 km from the LNGSINFN, while seismicity at distances larger than 50 km was not significantly affected by the L’Aquila earthquake: no evident increase in the number of events for regions with R = 50 km and R = 60 km was observed prior and after the event. To identify a threshold for the seismic activity, we choose the region identified by the intersection between a ten vertices polygon, which well approximates the 10 km boundary of the aquifer, and the Composite Seismogenic Sources. Four periods of relative intense seismic activity were identified: Period A, i.e., the preparation phase, the main shock and the aftershocks of the L’Aquila earthquake, Periods B, C and D. Quite naturally, the identified periods of seismic activity well correlate with the time spans when the number of earthquakes exceeds that predicted by the Omori’s law. Therefore, we propose that uranium groundwater anomalies observed only before Period A, i.e., before the seismic swarm and the main shock (which occurred on April 6th, 2009 in L’Aquila) and before the beginning of Period B (July–December, 2010), provide a key geochemical signal of a progressive increase of deep CO2 fluxes at middle–lower crustal levels: repeated sharp U enrichments in groundwater could be directly associated with the geodynamics of the earthquake.

5. Conclusions Acknowledgments The aim of this study has been the identification of possible correlations between uranium groundwater anomalies observed before the L’Aquila earthquake and the seismic features around the Gran Sasso area. Our results show that most of the seismic activity took place within a distance from 10 to 30 km from the LNGS-INFN. Within a radius of 10 km, seismic energy release was negligible before the L’Aquila earthquake, and keeps very low after it, except for the first few months immediately following the main event and for the two activation periods C and D. The seismic activity was quite limited within 20 km radius from the LNGS-INFN, up to couple of months before the L’Aquila earthquake, when the activity got comparatively higher at distances up to

We are grateful to two anonymous reviewers for their positive and constructive comments. The authors greatly acknowledge the support by the National Scientific Committee Technology of the National Institute of Nuclear Physics for the ERMES project, and by the Chemistry and Chemical Plants Service of the LNGS-INFN. REFERENCES AKINCI, A., and GALADINI, F., and PANTOSTI, D., and PETERSEN, M., and MALAGNINI, L., and PERKINS, D. (2009), Effect of Time Dependence on Probabilistic Seismic-Hazard Maps and Deaggregation for the Central Apennines, Italy, Bulletin of the Seismological Society of America, 99, 2A, 585–610.

Uranium Groundwater Monitoring and Seismic Analysis AMORUSO, A., and CRESCENTINI, L., and PETITTA, M., and TALLINI, M. (2013), Parsimonious recharge/discharge modeling in carbonate fractured aquifers: The groundwater flow in the Gran Sasso aquifer (Central Italy). Journal of Hydrology, 476, 0, 136–146. A.N.A.S.-CO.GE.FAR, Gran Sasso, Il traforo autostradale (1980). BARBIERI, M., and BOSCHETTI, T., and PETITTA, M., and TALLINI, M. (2005). Stable isotope (2H, 18O and 87Sr/86Sr) and hydrochemistry monitoring for groundwater hydrodynamics analysis in a karst aquifer (Gran Sasso, Central Italy), Applied Geochemistry, 20, 11, 2063–2081. BASILI, R., and VALENSISE, G., and VANNOLI, P., and BURRATO, P., and FRACASSI, U., and MARIANO, S., and TIBERTI, M.M., and BOSCHI, E. (2008), The Database of Individual Seismogenic Sources (DISS), version 3: Summarizing 20 years of research on Italy’s earthquake geology, Tectonophysics, 453, 20–43. BONI, C.F., and BONO, P., and CAPELLI, G. (1986), Schema idrogeologico dell’Italia Centrale. BENEDETTI, C.C., and BUSCEMA M., and RUGGIERI, M. (2011), System for Predicting Natural Disasters, Advanced Networks, In Algorithms and Modeling for Earthquake Prediction (ed. BUSCEMA M. and RUGGIERI, M.) (River Publishers, Aalborg 2011), pp. 239–254. DE NATALE, G., and KUZNETZOV, I., and KRONROD, T., and PERESAN, A., and SARAO`, A., and TROISE, C., and PANZA, G.F. (2004), Three decades of seismic activity at mt. Vesuvius: 1972–2000, Pure and Applied Geophysics, 161, 1,123–144. DIMRI, V.P., Analysis of seismological data, In Fractal Behaviour of the Earth System (ed. DIMRI V.P.) (Springer, Hyderabab 2005) pp. 11–19. FALCUCCI, E., and GORI, S., and MORO, M., and PISANI, A.R., and MELINI, D., and GALADINI, F., and FREDI, P. (2011), The 2009 L’Aquila earthquake (Italy): What’s next in the region? Hints from stress diffusion analysis and normal fault activity, Earth and Planetary Science Letters, 305, 3–4, 350–358. FREZZOTTI, M.L., and PECCERILLO, A., and PANZA, G. (2009), Carbonate metasomatism and CO2 lithosphere-asthenosphere degassing beneath the western Mediterranean: An integrated model arising from petrological and geophysical data, Chemical Geology, 262, 1–2, 108–120. GASPERINI, P., and LOLLI, BH., and VANNUCCI, G. (2013), Empirical calibration of local magnitude data sets versus moment

magnitude in Italy, Bulletin of the Seismological Society of America, 103, 4, 2227–2246. GELLER R.J., and JACKSON, D.D., and KAGAN, Y.Y., and MULARGIA, F. (1997), Earthquakes Cannot Be Predicted. Science, 275, 1616 GUTENBERG, B. (1956). The energy of earthquakes. Q. J. Geol. Soc. London, 112, 1–14. KANAMORI, H. (1977). The energy release in great earthquakes. Journal of Geophysical Research, 82, 20, 2981–2987. PLASTINO, W., Monitoring of geochemical and geophysical parameters in the Gran Sasso aquifer, In Radionuclides in the Environment, Int. Conf. On Isotopes in Env. Studies (ed. POVINEC, P., and SANCHEZ-CABEZA, J.A.) (Elsevier 2006), pp. 335–341. PLASTINO, W., and BELLA, F. (2001), Radon groundwater monitoring at underground laboratories of Gran Sasso (Italy), Geophysical Research Letters, 28,14, 2675–2677. PLASTINO, W., and NISI, S., and DE LUCA, G., and BALATA, M., and LAUBENSTEIN, M., and BELLA, F. (2009), Environmental radioactivity in the ground water at the Gran Sasso National Laboratory (Italy): a possible contribution to the variation of the neutron flux background, Journal of Radioanalytical and Nuclear Chemistry, 282, 3, 809–813. PLASTINO, W., and POVINEC, P.P., and DE LUCA, G., and DOGLIONI, C., and NISI, S., and IOANNUCCI, L., and BALATA, M., and LAUBENSTEIN, M., and BELLA, F., and COCCIA, E. (2010), Uranium groundwater anomalies and L’Aquila earthquake, 6th April 2009 (Italy), Journal of Environmental Radioactivity, 101, 1, 45–50. PLASTINO, W., and PANZA, G.F., and DOGLIONI, C., and FREZZOTTI, M.L., and PECCERILLO, A., and DE FELICE, P., and BELLA, F., and POVINEC, P.P., and NISI, S., and IOANNUCCI, L., and APRILI, P., and BALATA, M., and COZZELLA, M.L., and LAUBENSTEIN, M. (2011), Uranium groundwater anomalies and active normal faulting, Journal of Radioanalytical and Nuclear Chemistry, 288, 1, 101–107. PLASTINO, W., LAUBENSTEIN, M., NISI, S., PERESAN, A., POVINEC, P.P., BALATA, M., BELLA, F., CARDARELLI, A., CIARLETTI, M., COPIA, L., DE DEO, M., GALLESE, B., IOANNUCCI, L. (2013), Uranium, radium and tritium groundwater monitoring at INFN-Gran Sasso National Laboratory, Italy, Journal of Radioanalytical and Nuclear Chemistry, 295, 1, 585–592. ROMASHKOVA L., and PERESAN, A. (2013), Analysis of Italian earthquake catalogs in the context of intermediate-term prediction problem, Acta Geophysica, 61, 3, 583–610.

(Received March 17, 2015, revised July 13, 2015, accepted July 14, 2015)