On 24 August 2013 a sudden gas eruption from the ground occurred in the Tiber river delta, nearby Rome's international airport of Fiumicino. We assessed that this gas, analogous to other minor vents in the area, is dominantly composed of deep, partially mantle-derived CO2, as in the geothermal gas of the surrounding Roman Comagmatic Province. Increased amounts of thermogenic CH4 are likely sourced from Meso-Cenozoic petroleum systems, overlying the deep magmatic fluids. We hypothesize that the intersection of NE-SW and N-S fault systems, which at regional scale controls the location of the Roman volcanic edifices, favors gas uprising through the impermeable Pliocene and deltaic Holocene covers. Pressurized gas may temporarily be stored below these covers or within shallower sandy, permeable layers. The eruption, regardless the triggering cause—natural or man-made, reveals the potential hazard of gas-charged sediments in the delta, even at distances far from the volcanic edifices.
 On 24 August 2013, gas of unknown origin began erupting violently from the ground a few meters away from the compound of the Rome international airport “Leonardo da Vinci,” at Fiumicino municipality, 15 km SW of Rome (Figure 1). The eruption of the Fiumicino Gas Vent (hereafter FGV) produced initially a crater ~0.5 m wide, with vigorous gas bubbling in cold water (~20°C) with minor amounts of mud. It progressively increased in size reaching, after a month, a diameter of ~3 m, with a bubbling jet ~1 m wide. A second fluid blowout occurred (9 September) from a preexisting borehole 3 m away from FGV. As of 22 October, gas emission is still ongoing. We will not discuss the causes—natural or man-made (e.g., shallow drilling)—that may have triggered the eruption but will focus on the possible sources of the gas, whose occurrence in this area is not obvious.
 FGV is located in the deltaic alluvial plain of the Tiber River, along the central Tyrrhenian Sea margin, about 30 km from the ultrapotassic volcanic districts of Alban Hills and Mts. Sabatini, which belong to the Roman Comagmatic Province (RCP) [Peccerillo, 2005]. Surface manifestations of CO2-rich gas are known mainly in the periphery of the volcanic edifices of the RCP [Chiodini and Frondini, 2001; Barberi et al., 2007]. The CO2 has a deep origin (mantle or carbonate thermometamorphism) and reaches the surface along faults and fractures in the Quaternary volcanic and sedimentary cover. In 2005, a CO2-rich gas blowout occurred from a shallow borehole in the Fiumicino delta area, ~1.2 km southeast of FGV [Barberi et al., 2007] (F2 in Figure 1). The presence of geothermal gas in the Tiber delta is unusual, based on the site geology. In fact, a variably thick, 40 to 80 m, impermeable cover of clayey to sandy clayey, water saturated, Holocene sediments [Belluomini et al., 1986; Bellotti et al., 2007; Marra et al., 2013] overlies the several hundreds meters thick package of overconsolidated, Plio-Pleistocene marine clay, which, in turn, overlies the Meso-Cenozoic carbonate substrate [Funiciello and Parotto, 1978; Marra et al., 1995; Mariucci et al., 2008]. FGV was initially suspected to be biogas, produced in shallow peat deposits. We have analyzed molecular and CO2-CH4 isotopic composition of FGV gas and in four minor gas vents we found in the Fiumicino area. The origin and occurrence of the gas is discussed in the framework of the tectonostratigraphic setting of the area, reconstructed based on the analysis of the morphostructural lineaments and available borehole data.
2 Methodology and Results
 Multiple analytical systems have been used for determining molecular and isotopic composition of the gas released by FGV, in the surrounding soil, and four minor vents (Figure 1; Table 1): V1, gas bubbles in a stream channel close to the Traiano Lake archeological site; V2, V3, and V4 small gas exhalations from the soil in the Fiumicino urban area, all within 3 km of FGV. All these vents were initially analyzed in the field for CO2, CH4, H2S, and H2 by portable nondispersive infrared sensor (Dräger X-am 7000; accuracy < 5%). Gas was then collected by inverted funnel and stored in 150 and 500 ml glass containers with two vacuum stopcocks. Molecular composition of the gas was analyzed by Fourier transform infrared spectroscopy (FTIR Gasmet DX-4030, Finland), with standard spectra library for simultaneous detection of 14 gases (CH4, CO2, CO, N2O, C2H6, C2H4, C3H8, n-C4H10, i-C4H10, n-C5H12, i-C5H12, C6H6, COS, and SO2; typical detection limit of 1 ppmv and accuracy of ±10%) and gas chromatography in two different laboratories (University of Rome, with a Fisons 8000 GC for C1-C3 alkanes, C2H4; accuracy ±3% at 2 ppm; N2, O2 + Ar, and CO2, accuracy ±2%; Isotech Laboratories Inc., Illinois, with a Shimadzu 2010 TCD-FID at for C1-C6 hydrocarbons, He, H2, Ar, O2, CO2, N2; accuracy ±2%). CO2 (δ13C) and CH4 (δ13C and δ2H) isotopic composition was analyzed by isotope ratio mass spectrometry (Finnigan Delta Plus XL at Isotech Laboratories Inc., precision ±0.1 ‰ for 13C and ± 2 ‰ for 2H; ThermoFinnigan Delta Plus XP at Indiana University, ±4 ‰ using a methane preconcentrator based on the design of Miller et al. ). δ13CCH4 was also analyzed by cavity ring-down spectroscopy with a Picarro G2112-I CH4 isotope analyzer (Picarro Inc., California; precision <0.4‰ at 20 ppmv CH4, 5 min, 1σ).
Table 1. Molecular and Isotopic Composition of FGV and Surrounding Ventsd
aH2, He, CO, and C3+ hydrocarbons were below GC detection limit (5 ppmv).
bMean of two samples (0.5 m deep) collected 3 m from FGV.
dMolecular composition in vol.%; FGV, V1 and V2 data are corrected for air contamination; isotopic composition in ‰ VPDB for C, ‰ VSMOW for H. nd: not determined. D: IR Dräger sensor; F: Fourier transform infrared spectrometry; C: cavity ring-down spectrometry; G: Gas chromatography; I: isotopic ratio mass spectrometry (I-s: Isotech Labs; I-n: Indiana University). In the FTIR analyses CO, N2O, C2H4, C3H8, n-C4H10, i-C4H10, n-C5H12, i-C5H12, C6H6, COS, and SO2 were below detection limit (1 ppmv).
 The gas released by FGV is mainly composed of CO2 (~ 98 vol.%), with minor amounts (<2 vol.%) of N2 and CH4 (Table 1). Ethane and H2S are in trace amounts (11 and 1 ppmv, respectively) and all other hydrocarbons, N2O, COS and SO2, are below detection limits (<1 ppmv). Carbon isotopic composition of CO2 and CH4 collected on 2 different days are uniform, δ13CCO2, is −1.0 to −1.1 ‰, and δ13CCH4 −48 to −49 ‰ (with δ2HCH4 ranging from −199 to −179 ‰). The minor vents (Figure 1) are CO2 rich (δ13CCO2: −0.7 to −1.2 ‰), and CH4 reaches 2.5 vol.%, with δ13CCH4 up to −38 ‰.
 We have reconstructed the stratigraphic setting at the eruption site by means of borehole literature data [Belluomini et al., 1986; Bellotti et al., 2007; Marra et al., 2013]. Within the main paleochannel, stretching E-W 2 km north of Fiumicino (Figure 1), the ~ 80 m thick sedimentary succession consists of a coarse gravel layer at the base, ~ 8 m thick, followed by silty clays and sandy silts of transitional environment, including marine-dominated to fluvial-dominated bay muds. Intercalations of discontinuous lenses of lagoon-marsh organic clay, up to several meters thick, occur in the lower and upper portions of the succession. Finally, a variably thick (from few to 25 m) upper layer of beach, dune, and fluvial mouth bar sand association is present at the top of the succession. This delta sedimentary succession unconformably overlies the Plio-Pleistocene marine, overconsolidated clay substrate. In the zone of the gas crater, the substrate is at higher elevation with respect to the paleochannel, and an ~ 50 m thick succession of unconsolidated sediments occurs above it (Figure 1).
3 Discussion and Conclusions
 Fiumicino gas vents release a CO2-rich gas that is similar to that emitted by geothermal gas manifestations in the volcanic districts of the Latium RCP, Alban Hills, and Mts. Sabatini. FGV has, then, the same composition and δ13CCO2 of the gas that erupted in 2005 from a shallow borehole about 1.2 km SE [Barberi et al., 2007]. The high δ13C of CO2 implies it is certainly of deep origin, either from the mantle or thermometamorphism of limestones. The distinction of these two sources can be made by considering the mantle 3He/4He end member, which is the most appropriate for the investigated area [Sano and Marty, 1995]. The mantle beneath central Italy (RCP) is not MORB (Mid-oceanic ridge basalt)-type, rather it seems to have variable components of asthenospheric mantle similar to HIMU (High U/Pb) ocean island basalts, and enriched (radiogenic) mantle, the latter being generated from subduction of the Ionian/Adriatic plate [Tedesco, 1997; Martelli et al., 2004]. Due to this crustal contamination, RCP mantle 3He/4He ratio (R/Ra) is generally <5. The Latium-RCP magmatic sector has the lowest R/Ra, from 0.4 to 1.7, in olivine-pyroxene fluid inclusions [Martelli et al., 2004]. Fiumicino gas has R/Ra: 0.31 [Barberi et al., 2007]. Accordingly, as suggested by 3He/4He vs δ13C (Figure 2a), Fiumicino gas may have a considerable (up 77%) mantle component. We cannot exclude, however, that RCP mantle end member has R/Ra ratios higher than those of olivine-pyroxene fluid inclusions which would imply the crustal (limestone) component is larger.
 A remarkable difference between Fiumicino gas and the geothermal vents of the Latium-RCP volcanic districts (Alban Hills and Mts. Sabatini) is in the methane origin and concentration. CH4 at the periphery of the volcanoes was interpreted as a mixture of geothermal (abiotic) and thermogenic (biotic) gas [Cinti et al., 2011; Tassi et al., 2012]. CH4 in the Fiumicino gas (including FGV and V1 to V4) appears to have a dominant thermogenic component from moderately mature (oil window) kerogen (Figure 2b), and its concentration is one to three orders of magnitude higher than that generally occurring in Latium-RCP sites (0.004 vol.% at Pomezia, 0.01 vol.% at Palidoro, 0.03 vol.% at Tivoli [Tassi et al., 2012]; 0.25 vol.% at S. Maria delle Mole [Barberi et al., 2007]). Pliocene and Meso-Cenozoic gas reservoirs, with microbial and thermogenic gas, respectively, were reported in explorative boreholes west of Rome. Unfortunately, no geochemical data of this gas were published. However, if the amount of geothermal CH4 in the Fiumicino gas, which may be genetically linked to the deep CO2, is the same with that in the perivolcanic manifestations (orders of 0.001–0.1 vol.%), then most of CH4 in Fiumicino may derive from the thermogenic Meso-Cenozoic petroleum system. No significant microbial origin, for example, from Holocene peat-rich deposits, is however evidenced by isotopic data (Figure 2b): even mixing between 0.1 % geothermal (δ13C of ~ −25 to −10 ‰) and 0.9 % microbial (~ − 60 to −80 ‰) end-members would result in δ13C < −55 ‰, which is not compatible with the measured values. Combining the C1/C2 ratio (~1000) with δ13C in a Bernard diagram [see, for example, Etiope et al., 2009], Fiumicino gas does not fall on microbial-thermogenic mixing lines, but it is similar to thermogenic gas altered by molecular fractionation (lost of C2+ during migration), a process frequently observed in seeps associated to water discharges [Etiope et al., 2009]. FGV is not a biogas as initially thought. The existence of several gas emission points in the investigated area (V1–V4; Figure 1), coupled with the frequently high CO2 concentrations in the soil (even >10 vol.%) observed during a still ongoing survey, suggests that FGV is just the local manifestation of a pervasive occurrence of gas-charged sediments in the whole delta. After the submission of this paper (end of September 2013), a further gas eruption was reported offshore, 600 m west of the coast (2 km from FGV). High CO2 and CH4 concentrations might also occur inside the airport compound; the runways are only 700 m from FGV and 1800 m from the offshore eruption. But why does this geothermal gas occur in the Tiber delta, 30 km from the volcanic edifices?
 Regional NE-SW and NS trending faults are considered to act as preferential pathways for deep fluids on the Tyrrhenian Sea margin of Italy. Unlike the main NW-SE trending extensional faults, which have listric geometry, NE-SW and N-S faults acted as transfer and transcurrent faults, respectively, and are characterized by deep vertical extensions, affecting a deeper portion of the crust [Acocella and Funiciello, 2006; Faccenna et al., 1996]. In particular, the Alban Hills volcanic district is located at the intersection of two main N-S and NE-SW regional fault systems (Figure 1) [Faccenna et al., 1994a; Marra et al., 2003; Frepoli et al., 2010]. The Santa Maria delle Mole area, characterized by significant CO2 emission, is located along another important N-S fault zone [Marra, 2001], which induces moderate seismicity (Ml 2–2.5) [Frepoli et al., 2010]. Conversely, the Solforata hydrothermal manifestation is located along the NE-SW trending faults controlling the half-graben structure of the Ardea Basin [Faccenna et al., 1994b]. In contrast to the abovementioned sites, deep gas emissions in the coastal alluvial plain of the Tiber River are unexpected. The package of recent undeformed sediments would represent an ideal impermeable cover preventing fluids from reaching the surface. The widespread gas emission in this area implies the existence of an active magmatic fluid circulation beneath the Tiber valley between the Latium volcanic districts (see Figure 1) and the presence of a system of deep faults and fractures. These allow the deep CO2 to ascend and accumulate either at the interface between Holocene and Plio-Pleistocene sediments, as well as within pools in the sandy, permeable layers at the base and toward the top of the alluvial cover. Remarkably, the location of the delta and the terminal tract of the Tiber River appear to be strictly controlled by a NE-SW to ENE-WSW trending deep structure (Figure 1). Moreover, the riverbed in this area displays a series of N-S, en-echelon tracts which, similar to other sectors of the Roman area [Faccenna et al., 1994a], possibly reflect the presence of buried N-S fault segments which control the main course of the Tiber River in Rome (Figure 1). The occurrence of these lineaments, coupled with the sudden inversion of the river course close to Fiumicino, suggests the presence of a network of active structural lineaments, which may represent the cause for the observed uprising of deep fluids. In the absence of deformation in the Holocene cover, however, the gas emissions may be triggered by drilling operations, disrupting the impermeable cover and allowing the pressured gas to reach the surface. Future drilling and ground excavations in the Tiber delta (including the airport) should be based on accurate knowledge of gas distribution in the geologic substrate.
 Two anonymous reviewers helped to improve the manuscript. Thanks are due to Artur Ionescu for sample handling.
 The Editor thanks Patrick Dobson and an anonymous reviewer for their assistance in evaluating this paper.