Engineering Geology 327 (2023) 107336

Contents lists available at ScienceDirect

Engineering Geology
journal homepage: www.elsevier.com/locate/enggeo

Detecting gas upwelling hazards in coastal areas through integration of

active and passive electrical and seismic methods (Fiumicino, Central Italy)
Giorgio De Donno a, *, Alessandro Bosman b, Ettore Cardarelli a, Michele Cercato a,
Giuliano Milana c, Guido Penta de Peppo a
a
“Sapienza” University of Rome - DICEA, Via Eudossiana 18, 00184 Rome, Italy
b
Istituto di Geologia Ambientale e Geoingegneria (IGAG), Consiglio Nazionale delle Ricerche r.u. “Sapienza” University of Rome – DICEA, Italy
c
Istituto Nazionale di Geofisica e Vulcanologia (INGV), Sezione Roma 1, Rome, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: The accurate location of gas upwelling flows is still an open problem for non-invasive imaging techniques in
Electrical resistivity tomography populated areas. Gas blowouts of deep origin may represent a serious threat to human health in urban areas and
High-resolution seismic data should be correctly imaged with high-resolution for assessing the related hazards. In this work, we propose an
Ambient noise recordings
integration of active (electrical resistivity tomography and high-resolution sub-bottom profiling complemented
Self-potential
with the multibeam bathymetry) and passive (self-potential and ambient noise recordings) geophysical methods
Gas migration
Geological hazard to image gas upwelling flows in the coastal area of Fiumicino (Central Italy), where the gas presence is well-
documented by previous works. We demonstrate that merging seismic sub-bottom profiling and electrical re­
sistivity tomography has enormous diagnostic potential for gas detection, since they combine the high resolution
needed to correctly image the subsurface and the interfaces between different media with the high diagnostic
capability of electrical methods to detect anomalies associated with the gas emissions. Passive seismic methods
complement the analysis enabling an estimation of the shear-wave velocity through array measurements. Finally,
the reconstruction of the natural electrical sources, inferred from the inversion of self-potential data, confirms the
location of the near-surface gas upwelling flows assessed through the resistivity model. This work demonstrates
that the integration of high-resolution active and passive seismic and electrical methods can be an effective
choice for the accurate location of risk-prone areas by imaging the near surface gas pathways where borehole
drilling is strongly limited if not forbidden.

1. Introduction are available and non-destructive geophysical surveys are a cost-

effective choice for imaging the gas pathways. Geoelectrical methods,
During the last decades gas blowouts have represented a serious such as electrical resistivity tomography (ERT), can be diagnostic for gas
threat to human health in urban contexts or when planning future ur­ detection, as they can highlight the resistivity contrast between water-
banization (Hansell and Oppenheimer, 2004). Natural gases found in the saturated (conductive) and gas-saturated (more resistive) media. In
shallow subsurface can have different causes, even if the most abundant fact, the electric resistivity of porous sediments significantly increases
are undoubtedly carbon dioxide (CO2) and methane (CH4). These gases when electrically conductive brine is displaced by CO2 (e.g. Bergmann
are generally of deep origin and rise toward the surface through faults, et al., 2012). However, a decrease of resistivity can be observed
cracks and voids, but may also be conveyed to the surface by borehole depending on the phase of the CO2, in case dissolution in brine and
drilling (Barberi et al., 2007; Sella et al., 2014). uptake of dissolved solids occurs, although this effect is negligible for
For these reasons, previous works have focused on locating gas in the salinity between 20 and 160 g/l (Fleury and Deschamps, 2008), which is
shallow subsurface (e.g. Carcione et al., 2011; Johansson et al., 2011), the range likely encountered in coastal areas. Examples of field appli­
even if a standard procedure for identifying these gases through non- cation of ERT for gas detection are almost only restricted to monitoring
invasive investigations has not been established. In such geological CO2 storage or injection experiments (Bergmann et al., 2012 and ref­
scenarios where borehole drilling is limited, only a few geological data erences therein) or to image hydrothermal upwelling fluids in large-

* Corresponding author.
E-mail address: giorgio.dedonno@uniroma1.it (G. De Donno).

https://doi.org/10.1016/j.enggeo.2023.107336
Received 26 January 2023; Received in revised form 25 September 2023; Accepted 21 October 2023
Available online 7 November 2023
0013-7952/© 2023 Elsevier B.V. All rights reserved.
G. De Donno et al. Engineering Geology 327 (2023) 107336

scale surveys often located on volcanic areas (e.g. Gresse et al., 2017). is applied to the coastal area of Fiumicino prone to gas hazard as re­
High-resolution (HR) seismic reflection profiles executed on land or ported by many authors (see e.g. Carapezza et al., 2015) and located 25
in shallow water (rivers and channels) can complement and validate the km west of Rome (Italy), where we performed active (ERT and HR
electrical models, by inferring the subsoil layering down to significant seismic sub-bottom profiler complemented with the multibeam ba­
depths as well as identifying the gas upwelling flows often visible on the thymetry) and passive (ambient noise recordings and SP) investigations,
seismic sections as blank zones of low-amplitude signal levels (e.g. to reduce the ambiguities often arising when geophysical techniques are
Riedel et al., 2002). Furthermore, HR resolution multibeam bathymetry applied standalone.
is often employed alongside seismic reflection for the morphological
characterization of submerged areas and for pockmarks identification 2. Study area and geophysical measurements
(Bosman and Orlando, 2017).
In recent years, the acquisition of passive seismic data has become The study area is located on the Tiber delta in the Municipality of
routinary, due to the cost-effectiveness of this technique compared to Fiumicino (Rome, Italy), close to the Fiumicino International Airport
active seismic surveys (Bard and SESAME Team, 2004). Common ap­ (Fig. 1) and to the coastline. Four areas were investigated in a narrow
plications of single-station recordings and array measurements include range of 5 km up to a maximum of 80 m depths from the coast to the
the detection of bedrock surfaces (e.g. Lane et al., 2008) and faults (e.g. Tiber River (Fig. 1). The near-surface layering, inferred from scattered
Qian and Liu, 2020) rather than revealing degassing zones. However, borehole data, consists of three main geological units (Fig. 2), separated
low-frequency anomalies in spectral ratios of single-station microtremor by unconformity surfaces (Milli et al., 2013), from the bottom to the top:
measurements were observed in oil and gas fields in Austria (Lambert i) clay and silty clay of Lower Pleistocene, belonging to the Monte Mario
et al., 2009), even though their interpretation is still controversial. Sequence (MMS), ii) gravels and sandy gravels of the Middle Pleistocene
Self-potential (SP) passive signals have been often recorded in vol­ Ponte Galeria sequence (PGS), iii) clay and peaty clay of the Upper
canic areas, where a strong “W”-shaped signature (e.g. Barde-Cabusson Pleistocene to Holocene Tiber Depositional sequence (TDS).
et al., 2021) is often associated with upwelling flows and there are also The PGS formation is the most permeable layer hosting a ground
some applications of SP to small-scale problems, where the magnitude of aquifer where gas rising from depth may accumulate (Fig. 2).
degassing phenomena is supposed to be much lower (e.g. Byrdina et al., Conversely, the uppermost clays (TDS) act as an impervious cap rock for
2009 and references therein; Nickschick et al., 2017). Additionally, PGS gravels allowing gas pressurization (Carapezza et al., 2015).
some studies integrated SP with passive seismic methods as well as ERT Previous investigations in the study area included boreholes (Milli
for detection and monitoring of hydrothermal activities (e.g. Legaz et al., 2013; Sella et al., 2014) and soil gas surveys (Bigi et al., 2014;
et al., 2009). Ciotoli et al., 2016; Maffucci et al., 2022), as well as a low-resolution
In this work we present an integrated methodology that combines multichannel seismic refraction profile along the Tiber River (Bigi
electrical and seismic methods to retrieve an accurate image of the gas et al., 2014). PGS unit was found in these boreholes between 37.5 and
upwelling flows and the geological features (gas reservoir, saline 53 m below sea level (b.s.l.), although with significant variations of both
intrusion, stratigraphy) in coastal environments. The proposed approach depth and thickness throughout the study area (Table 1).

Fig. 1. Satellite image of the surveyed areas and location of the geophysical investigations with available boreholes and anomalous gas emission points after Bigi
et al. (2014). The two main investigated sites, located at “Coccia di Morto” (Site 1, enlarged view in Fig. 3a) and near “Capo due Rami” (Site 2, enlarged view in
Fig. 3b) are within white rectangles.

2
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 2. Simplified stratigraphic column of deepest boreholes close to the investigated area: Pesce Luna Core (PLC, Milli et al., 2013) on the left and S0 (Sella et al.,
2014) on the right. TDS: Tiber Depositional Sequence (yellow), PGS: Ponte Galeria Sequence (light blue), MMS: Monte Mario Sequence (purple). See Fig. 1 for
borehole locations. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1
Location of gravel layer (PGS) from a piezocone test (CPTU) and boreholes: a. Milli et al., 2013, b. Technical report (private communication), c. Sella et al., 2014. All
depths are referred to the sea level. *Orthometric elevation of the wellhead inferred from Laser Imaging Detection and Ranging (mesh 0.5 m) provided by the Italian
Ministry of the Environment. **Value inferred from piezocone test results: tip resistance ~18 MPa and pore pressure > 600 kPa. See Fig. 1 for borehole locations.
Borehole/Piezocone test Elevation (wellhead)* (m a.s.l.) Gravel roof (m a.s.l.) Gravel bed (m a.s.l.)

PLCa 1.0 − 48 − 53
CPTUb − 6.5 − 37.5** not found
S0c 1.8 − 42 − 52
S4c 1.8 − 40 − 47
S5c 1.3 − 40.5 − 50.5

The maximum gas concentrations (CO2 and CH4) were found at two m). In this site, gas emissions can be expected due to the closeness to the
selected areas (Fig. 1): “Coccia di Morto” (Site 1), where the main main degassing event, although not still systematically recorded by soil
degassing vent (Fiumicino Gas Vent) occurred in 2013 and near “Capo gas surveys. Therefore, we only integrate ERT and passive seismic data
due Rami” (Site 2), on the right bank of the river, where the soil gas to improve the knowledge of deep geological layering.
surveys recorded the maximum concentrations (Bigi et al., 2014; Ciotoli Then, the fully integrated approach is applied at Site 2 (Fig. 3b),
et al., 2016; Maffucci et al., 2022). where gas concentrations were clearly highlighted from previous mea­
We carried out preliminary geoelectrical investigations along two surements (Bigi et al., 2014; Ciotoli et al., 2016; Maffucci et al., 2022).
profiles (L1 and L3), almost parallel to the central Tyrrhenian shoreline At Site 2, we take advantage of the high-resolution (HR) seismic
(Fig. 1) and close to the available boreholes (PLC, SO and S4) and reflection profile (purple line A-B in Fig. 1) acquired along the Tiber
reasonably far from the known gas anomalies from previous measure­ River (Fiumara Grande branch), together with a high-resolution multi­
ments (e.g. Bigi et al., 2014). This initial survey aims to reconstruct the beam bathymetry of the riverbed. The ERT data were acquired on a long
geological layering down to the TDS/PGS interface (depth of investi­ (700 m) and deep (DOI ~ 80 m) profile (L4 in Fig. 3b) along the levee of
gation - DOI ~ 45 m), the expected resistivity range of the geological the Tiber River. We also used passive geophysical methods (self-poten­
formations and the location of the saltwater-freshwater interface, since tial and ambient noise recordings) to complement the active methods
the geoelectrical models can be validated by borehole data. thus reducing the uncertainty in the interpretation of geophysical
Then, we applied integrated geophysical surveys at the two selected models. Passive seismic measurements were carried out with a single-
sites (Site 1 and 2 in Fig. 1). For Site 1 (Fig. 3a) we carried out a 2D array station technique, due to the limited space available on the levee.
of ambient noise recordings, together with a deeper ERT line (DOI ~ 80

3
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 3. Detail of the geophysical survey. (a) Site 1, (b) Site 2. The white line represents the projection of the array center to the L2 line. High-resolution multibeam
bathymetry of the Tiber River is superimposed on the site map. The reader is referred to Fig. 1 for the large-scale location of the investigated sites.

3. Data acquisition, processing and inversion 3.2. Acoustic survey: HR seismic reflection sub-bottom profiling and
multibeam bathymetry
3.1. Electrical Resistivity Tomography (ERT)
The high-resolution (HR) seismic profiling was performed using a
The ERT profiles were acquired using the IRIS Instruments Syscal Pro Benthos Chirp III source with a sweep between 2 and 7 kHz and a single-
resistivimeter with 48 stainless steel electrodes spaced 5 m (L1 and L3 in channel zero-offset configuration (Bosman and Orlando, 2017) on board
Fig. 1) and 10 m (L2 and L4 in Fig. 3a and b respectively) apart, a small vessel. GNSS positioning of the single-channel seismic was
depending on the external limitations. Multiple gradient array is conducted in Real Time Kinematic (RTK) referred to as the “Mose” base
employed for ERT acquisition, using a maximum dipole length a = 5 and station located at “Sapienza” University of Rome (permanent GNSS
a maximum separation factor s = 9 (738 data points for each baseline), network, frame ETRF2000 epoch 2008). The baseline is approximately
as it combines consistent signal strength in high conductive environ­ 22 km and provided a centimetric accuracy for this mono-channel
ments with good resolution and depth of investigation. For the L4 line, seismic survey. Raw data were processed through Geo Suite All Works
the ERT dataset was acquired using the roll-along technique by over­ software, and we set for the time-depth conversion a value of 1470 m/s
lapping 40 electrodes for each baseline (240 new points for each new in water (measured in situ by a sound velocity profiler sensor) and 1550
section and 100% coverage). We use a 12 V - 110 Ah external battery for m/s for the underlying sediment, estimated as a mean value for un­
power supply recording satisfactory current levels (1–2.5 A) along the consolidated silty/clayey sediments at shallow depths (Hamilton, 1979).
line, mainly due to the relatively low resistivity of the sediments because Although there are some recent attempts to filter out multiples e.g.
of closeness to the sea. Conversely, the extremely conductive environ­ through gapped deconvolution (Vesnaver et al., 2021), we prefer to only
ment leads to a low signal-to-noise ratio for the L1 line (executed on the note them wherever occur in the acoustic record to avoid distortion of
shore), due to the low voltage drop recorded. the seismic signals or attenuation of underlying weaker signals.
We filtered raw data for negative apparent resistivity values and for The HR bathymetry survey of the Tiber River was performed using
clear isolated points, but we decided to keep in the datasets also points the Teledyne Reson SeaBat 7125 echo-sounder (400 kHz) using the
with high percentage standard deviations. In fact, the percentage error multibeam transducers in standard mode (look down) and rotating the
fails to properly assess the reliability of a measurement for zero or close head 30◦ to investigate the riverbanks up to hydrographic zero level.
to zero observations (Sanders, 1997), such as those recorded in the The vessel positioning was supplied in real-time by an Applanix Position
highly conductive coastal environments where voltage drops are often and Attitude System (POS/MV wave master V5) using RTK corrections
very low. Consequently, we reported all errors in the following ERT received by a GNSS master base station belonging to the GNSS National
models as both percentage and absolute values. Dynamic Network (MOSE http://www.igmi.org/rdn/). Data were also
Apparent resistivity data were inverted using the VEMI algorithm re-processed with post-processing kinematic (PPK) techniques by means
(De Donno and Cardarelli, 2017), where the two-dimensional forward POSPac MMS software for very highly accurate positioning of the
solution is achieved using a finite element approach with quadrangular soundings. Multibeam bathymetry data were processed using Caris Hips
elements, while a Gauss-Newton iterative formulation, based on the & Sips 9.1 hydrographic software to generate a high-resolution Digital
minimization of the l1-norm (the so-called robust or blocky inversion), is Elevation Model (DEM) with a 0.2 m cell size. The processing workflow
used for data inversion (Loke et al., 2003). The robust inversion is highly consisted of replacing GNSS positions processed in PPK mode, sound
indicated for this case study since it is less sensitive to data points with velocity refraction editing, patch test, tide correction based on PPK
larger errors and it can enhance the sharp transitions between different GNSS/IMU data techniques and application of statistical and geomet­
media (Loke et al., 2003). Although a priori information can be intro­ rical filters to remove coherent/incoherent noise (Bosman et al., 2015).
duced into the inversion process using VEMI, we made no preliminary
assumption on the soil layering.

4
G. De Donno et al. Engineering Geology 327 (2023) 107336

3.3. Ambient noise recordings curves are manually rejected from the software. For the computation, we
used a frequency band from 0.4 to 20 Hz.
Site 1 was investigated by a 2D seismic array based on 12 standalone
seismic stations equipped with a 3-component velocity sensor Lennartz
LE3D-5 s with an eigenfrequency of 0.2 Hz. Signals were continuously 3.4. Self-potential
recorded during a 2-h time window using a high-resolution 24-bit
Reftek130 datalogger at a sampling rate of 250 samples/s. Array ge­ Self-potential (SP) data were acquired with an offset of 180 m to the
ometry was quite sparse with a maximum aperture of about 150 m and a first electrode of the L4 line (Fig. 3b), by a string of 10 non-polarizable
minimum spacing of about 18 m. Stations were positioned using a GNSS electrodes (Cu/CuSO4), spaced 5 m apart, rolled along the investigated
Leica 1200 receiver (DGPS) to reduce the position errors to <5 cm. profile by overlapping an electrode pair for each baseline. At each sta­
Signals were processed using the open-source software Geopsy tion, a small hole (~10 cm deep) was dug to improve the electrical
(Wathelet et al., 2020) to obtain Rayleigh waves ellipticity (horizontal- contact between the electrode and the ground. The electrodes are con­
to-vertical spectral ratio - HVSR) and dispersion curves along with site nected to the Syscal Pro via the same equipment used for ERT survey.
resonance frequency (f0). The HVSR data were analyzed following the For each line, we performed from 5 to 10 repetitions every 2 min starting
guidelines and recommendations of the SESAME project (Bard and from 20 min after the electrode plug-in, to check the robustness and
SESAME Team, 2004). The dispersion curve was obtained by applying consistency of the measurements. SP data were filtered for outliers and
both conventional and high-resolution FK analysis (Capon, 1969; inverted using the SP2DINV software (Souied et al., 2013), achieving a
Ohrnberger et al., 2004) along with the Modified Spatial Autocorrela­ current density model directly related to the underground sources. Data
tion (MSPAC) technique (Bettig et al., 2001) to extend the investigated inversion was carried out using Tikhonov approach with a depth
frequency range toward the low-frequency band. weighting matrix, computing the regularization parameter with the
For Site 2 we used a three-component Sara seismic sensor with a generalized cross-validation (GCV) method (e.g. Jardani et al., 2008).
natural period of 5 s; the duration of the seismic noise records was set to
45 min employing a sampling frequency of 100 Hz. Time alignment of 4. Results
samples and positioning were guaranteed by a GNSS receiver (DGPS).
Also in this case we processed the ambient noise signals using Geopsy 4.1. ERT preliminary survey
with the SESAME recommendations to obtain the HVSR curves and the
resonance frequency (f0), and to extract the directivity information of The ERT inverted models for the L1 and L3 lines are shown in Fig. 4a
the signals by plotting the spectral ratio as a function of both frequency and Fig. 4b, respectively. In both cases, the electrode spacing is 5 m so
and azimuth. Data processing was performed with a 25 s time windows that the ultra-shallow weathered layer is not imaged at all, and we can
adopting the following processing parameters: short-term average consider the water level coinciding with the ground surface. In one case
(STA) = 1, long-term average (LTA) = 30, with min and max STA/LTA (L1), the surface aquifer is salt water, whereas for the L3 line a fresh-
thresholds between 0.2 and 2.5. A 5% Tukey window function was water aquifer overlies salt water. The relatively high percentage error
applied to the raw signal and the curves were smoothed using the Konno on L1 (14.3%) is due to the extremely low conductive environment
and Ohmachi (1998) method, with a smoothing constant of 40. Outlier (mean apparent resistivity ~1 Ωm) but is satisfactory if evaluated in
absolute terms (~ 0.2 Ωm).

Fig. 4. Resistivity model for L1 (a). Error (AE): 14.3% (0.2 Ωm). Resistivity model for L3 (b) Absolute Error (AE): 5.9% (0.2 Ωm). TDS: Tiber Depositional Sequence,
PGS: Ponte Galeria Sequence. See Fig. 1 for location.

5
G. De Donno et al. Engineering Geology 327 (2023) 107336

The L1 line (close to the coastline) detects a slightly resistive layer was not possible to retrieve information at higher frequencies. The good
(3–5 Ωm) down to 5–7 m b.s.l., while resistivity decreases to 0.5–1.5 Ωm stability of data in the 1–3 Hz frequency range and the range of depth to
between 8 and 35 m b.s.l. This effect is almost only related to a lithology be investigated suggest using this part of the ellipticity curve to be
change between shallow dune sands and deeper clays and silts (both jointly inverted with the dispersion curve to obtain the shear-wave ve­
belonging to TDS) since the saltwater level is close to the surface (0 m a. locity (Vs) profile at the investigated site. The inversion was performed
s.l.). Then, we observed a slight resistivity increase in the gravel PGS using again the Geopsy software package based on the Conditional
formation (z > 35 m b.s.l.), which is known to host the gas reservoir, Neighbourhood Algorithm (Wathelet, 2008; Wathelet et al., 2008). The
from borehole data in the adjacent CPTU (offshore) and PLC (Fig. 2). The fit of the inversion and the Vs model are shown in Fig. 7, where the
depth of the reconstructed surface varies between 30 and 40 m along the ellipticity curve is calculated in terms of the arctangent, expressed in
investigated line, thus demonstrating the variability of the lower degrees, of the H/V spectral ratio. The dispersion curve is associated to
boundary of TDS already seen between CPTU and PLC (Table 1). ERT the fundamental mode of Rayleigh waves. The initial model used in the
models do not show significant effects of gas upwelling since resistivity inversion process was based on the results of the PLC borehole (Fig. 2).
remains approximately constant within the silt/clay layer (Fig. 4a). A The inverted velocity models are represented at a maximum depth of 80
similar layering is also reconstructed for the L3 line (1.5 km far from the m since near surface characterization is the focus of this study.
seashore), although the first layer is thicker (0–8 m) and more resistive The shear-wave velocity model (Fig. 7c) shows a low-velocity
(8–30 Ωm). In fact, here the shallow ERT layering mainly reflects shallow layer (Vs between 200 and 250 m/s) with a first interface at
changes in water salinity rather than in lithology, since the distance depth of about 25 m where the velocity increases to 350–400 m/s. A
from the seashore increased. The freshwater-saltwater interface is found major impedance contrast is found at depths between 50 and 60 m
at 8 m b.s.l. and the ERT model detects the gravel roof at 35–40 m, where where Vs reaches the value of 800 m/s, likely due to the presence of the
resistivity increase (~ 5 Ωm) is likely due to gas saturation within the gravel layer as indicated by ERT.
gravel layer (Fig. 4b).
4.3. Site 2
4.2. Site 1
4.3.1. HR seismic reflection
4.2.1. ERT The HR seismic profile A-B collected along the Tiber River (Fig. 8a),
The inverted model of the L2 line (Fig. 5) with electrodes spaced 10 highlights two main reflections characterized by sharp bottom echoes,
m apart, reached deeper zones (DOI ~ 80 m). The electrical layering is due to the riverbed (red line in Fig. 8b) and the unconformity articulated
similar to the L3 model (Fig. 4b), even though the shallow resistive layer surface (green line in Fig. 8b) between TDS finer sediments and PGS
extends down to 15 m b.s.l.. Here the resistivity abruptly decreases to gravels (estimated depth = 40–65 m along the profile) together with
0.5–1.5 Ωm between 15 and 55 m b.s.l., due to the presence of saline several blank zones, likely associated with accumulations of trapped gas.
intrusion inland, while we observed a resistivity increase in the gravel Conversely, there are no pieces of evidence of major faults along the
PGS formation. A slight increase in the resistivity of the middle layer (ρ investigated A-B section, as well as on the riverbed (Fig. 8a) since hy­
>1.5 Ωm, x = 210 and 310 m) can be likely attributed to the presence of perbolas and/or layer discontinuities are not visible on the HR profile.
gas upwelling flows if compared with the L1 and L3 models, where this Although the riverbed is characterized by bedforms (Figs. 3b and 8c),
effect is not visible. the HR multibeam bathymetry does not show the presence of faults or
steps on the bottom of the river. However, it highlights the presence of
4.2.2. Array measurements depressed morphologies caused by localized erosion due to the hy­
HVSR data collected in the site 1 (Fig. 3a) show a complex behavior draulic narrowing of anthropogenic features along the levees (parking
(Fig. 6a) with a peak at a very low frequency (0.2 Hz), very common in spaces for cars), commonly found in this area of the Tiber River.
the Roman area (Marcucci et al., 2019) and a secondary one at about 1.5
Hz. These features are quite common at all recording sites. At fre­ 4.3.2. ERT
quencies higher than 3 Hz station’s behavior is no longer homogeneous In Fig. 9, we show the ERT model of the L4 line (DOI ~ 80 m),
suggesting some lateral variation in the velocity properties of the very executed along the levee (height ~ 5 m), where the presence of gas vents
shallow soil layers. was well documented (Bigi et al., 2014). The electrical layering, already
The dispersion curve (Fig. 6b) is well-defined in the 0.55–7.5 fre­ seen on L2 (Fig. 5) is confirmed also for the L4 line, with the location of
quency range. Due to the last observation and the large array aperture, it the unconformity surface in agreement with the A’-B′ seismic sub-

Fig. 5. Resistivity model for the L2 line (Site 1), where interfaces between different layers are marked with dotted lines. Absolute Error (AE): 18.8% (0.8 Ωm).
Vertical exaggeration is 1.4. TDS: Tiber Depositional Sequence, PGS: Ponte Galeria Sequence. See Fig. 1 for location.

6
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 6. a) H/V ratio for all the 2D array stations; b) Rayleigh fundamental mode dispersion curve obtained from the array data.

Fig. 7. (a) Fit of experimental (black line) and inverted ellipticity curve; (b) fit of experimental (black line) and inverted dispersion curve; (c) inverted Vs veloc­
ity model.

bottom profile collected on Tiber River. We detect six anomalous zones, effect of directionality is visible between 0 and 45◦ N and 160–180◦ N
elongated in the vertical direction, likely due to gas upwelling flows, of only for HV5 (Fig. 10g), while there are no significant effects on the
which the most significant is located between 380 and 410 m. other stations (Fig. 10e, f and h).

4.3.3. Single-station ambient noise recordings 4.3.4. Self-potential

The H/V spectra for Site 2 (Fig. 10) exhibit only moderate amplifi­ Self-potential observed data are shown in Fig. 11a (gray line and
cation effects, as the maximum spectral ratio is around 2.1–2.2. The circles) together with the fitting (predicted data) at the last iteration of
main resonant peak is located at 1.8–2.0 Hz for the first stations (Fig. 10a the inversion process (black line). Although inversion fails to properly
and b), even though it becomes less significant or barely visible for the address some minor anomalies (i.e. x = 200–230 m), significant in­
last stations (Fig. 10c and d). This peak is likely associated to the seismic creases in current density between 290 m and 320 m and mostly be­
impedance contrast between the TDS finer sediments and the underlying tween 360 and 420 m, are clearly highlighted, consistently with the
gravels, as shown for the Site 1. For the last stations (HV5–7, Fig. 10c resistivity anomalies shown by the ERT model. The maximum DOI is
and d) a secondary peak at 4.5–5 Hz becomes prevalent, as well as a around 15 m (10 m b.s.l.), as proof that the SP method can enlighten in
high-frequency response (10–15 Hz) is also observed for HV5. A clear this case only the shallow portion of the subsurface.

7
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 8. (a) HR seismic sub-bottom profile A-B collected on the Tiber River. Alongside the A’-B′ branch, electrical and passive seismic measurements were performed
on land. (b) Interpretation of the profile A-B: red solid line – riverbed; green solid line – main reflector. Black arrows indicate the gas accumulation zone which
corresponds to the blanking areas in the A’-B′ branch. See Fig. 1 for location. (c) HR resolution multibeam bathymetry collected along the Tiber River shows bedforms
and large depressions produced by anthropogenic hydraulic narrowing. Bathymetric vertical datum: multibeam data were collected in ellipsoid elevation and then
transformed to orthometric elevation using the Italgeo2005 model. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)

4.4. Data integration the left part of the section, likely due to the first-approximation value of
velocity chosen for the time-depth conversion. The main resonant fre­
Data integration for the two investigated sites is reported in Fig. 12. quency of single-station recordings is 1.8–2 Hz, slightly higher than that
For Site 1 (Fig. 12a), the deep interface, associated to the transition retrieved by the array measurements, likely due to the shallower posi­
between TDS (sand, silt and silty clay) and PGS (gravel layer) identified tion of the TDS/PGS interface compared to Site 1 (50–60 m) or to minor
in the resistivity model of Fig. 5, is detected at approximately the same lithological changes between the two sites. The resistivity model high­
depth (~ 55 m b.s.l.) of the shear velocity profile reconstructed from lights six main gas plumes upwelling from the reservoir (black arrows),
inversion of array measurements (superimposed to the ERT model in defined as the lower boundary of TDS. Three of these (located approx. at
Fig. 12a), where there is a strong seismic impedance contrast (approx. 300, 400 and 480 m) match exactly the current density increases in the
From 400 to 800 m/s), thus strengthening the validity of the proposed SP model, even though this evidence is limited to the shallow portion of
interpretation. Additionally, data integration confirmed that the inter­ the subsurface. The main anomaly is located approx. at x = 400 m, also
face at approximately 15 m b.s.l. is related to a transition between media validated by surface gas measurements made by previous works (Bigi
having different salinity and not to a lithological change. et al., 2014; Ciotoli et al., 2016), highlighting both CO2 and CH4 emis­
Also at Site 2, the integration of the HR seismic data and the ERT sions at this position. Additionally, the station which shows a clear
profile shows a good correspondence between the increase of resistivity directional effect (HV5) is located where active and passive electrical
due to the gas reservoir and the main reflector (depth ranging from 45 to methods show the presence of the most significant gas rise.
55 m b.s.l. along the investigated profile), with minor discrepancies in

8
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 9. Resistivity model for the L4 line (Site 2), where interfaces between different layers are marked with dotted lines. Absolute Error: 4.6% (0.3 Ωm). Vertical
exaggeration is 2. TDS: Tiber Depositional Sequence, PGS: Ponte Galeria Sequence. See Fig. 1 for location.

Fig. 10. Site 2: directionality of ambient noise recording for HV1 (a), HV3 (b), HV5 (c) and HV7 (d) stations (see Fig. 3 for the location of stations).

5. Discussion previous works for the Fiumicino area (Bigi et al., 2014; Ciotoli et al.,
2016; Maffucci et al., 2022), is the best option to highlight the risk-prone
Gas detection using geophysical investigation is a relatively new areas and to focus the geophysical survey on selected areas, thus
research topic if restricted to near-surface low emissions in urban areas, improving the cost-effectiveness on the whole survey. Despite the
while extended literature is available for imaging of deep gas emissions extreme environmental conditions encountered in this study (ρ < 1 Ωm
in natural scenarios (Barde-Cabusson et al., 2021). In this context, for large areas), the ERT method has been demonstrated to be effective
drilling boreholes is strictly limited if not forbidden due to the risk of and highly diagnostic for imaging gas upwelling flows, where resistivity
explosion of flammable gases and expensive operating procedures that approximately doubles compared to the neighboring zones (Fig. 12).
have to be implemented when drilling. Therefore, surface geophysical The geophysical investigation proposed in this work, where ERT is
methods, together with geochemical data are the only option to image complemented by ambient noise recordings, SP and HR sub-bottom
the gas upwelling flows. In this respect a large preliminary screening profiling, could be effectively employed for quantitative hazard assess­
performed with geochemical measurements, already examined in ment at a local scale, providing adequate resolution for moderate depth

9
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 11. Site 2: Inversion of SP data: (a) fitting between observed (gray line with open circles) and predicted (black line) data; (b) current density model.

targets (maximum DOI in this case ~80 m). In Table 2 we summarized Self-potential positive signals are strongly correlated to gas upwell­
the main targets theoretically achievable by the proposed methods, with ing flows in volcanic areas (Zlotnicki and Nishida, 2003), even if with a
the limitations discussed below. higher magnitude compared to this case study where the maximum SP
The loss of resolution with depth, intrinsic to the ERT method, can be data is approximately 20 mV. Nevertheless, the SP level observed at the
mitigated by using HR seismic reflection methods in shallow water, Fiumicino site is comparable to that recorded in similar terrestrial sce­
since for deep zones the location of a resistive target cannot be evaluated narios (Nickschick et al., 2017), where SP was also applied in combi­
accurately by ERT standalone, and deviations of some meters are com­ nation with ERT. As in that case, anomalous SP zones are marked with
mon (e.g. Cardarelli and De Donno, 2017). Additionally, a high resistive fluctuating higher and lower magnitudes, similar to the W-shaped sig­
anomaly (gas reservoir) might be reconstructed in the bottom pixels of natures observed for gas flows in volcanic environments (Barde-Cabus­
the ERT model with lower resistivity values compared to those expected, son et al., 2021). The current density model, reconstructed from the
due to the loss of resolution with depth (e.g. Gélis et al., 2010), as in this inversion of SP data, can delineate the gas accumulation in the shallow
case where the resistivity increases only by few Ωm. For land surveys far portion of the subsurface (DOI ~ 15 m), by complementing the ERT
from coastal areas or rivers, the HR single-channel seismic acquisition model. In fact, gas detection in the near surface through ERT can be
can be conveniently replaced by a multi-channel reflection, even with an biased by the increase of resistivity due to the decrease of salinity and/or
increasing effort for data acquisition and processing and higher costs. saturation observed in the study area.
Alternatively, ambient noise recordings could be a low-budget option to As an ancillary result, the ERT models also permitted an assessment
have a rough estimate of the reservoir depth, provided that the shear of the freshwater (FW)-saltwater (SW) interface, even though only
wave velocity is properly estimated by inversion of array data. Addi­ restricted to a few lines and with a lower resolution compared to that
tionally, information about the presence of faults can also be inferred by achievable with a high-resolution shallower survey only focused on SW
the analysis of the signal directivity polarization of H/V spectra. Po­ detection. We chose a resistivity value around 3 Ωm as a threshold for
larization transversal to the strike direction has been previously locating the FW-SW interface, which was previously used for silty-clayey
observed for normal and strike-slip faults, as a result of stiffness saturated sediments in similar coastal scenarios (e.g. Attwa et al., 2011;
anisotropy in the fault zone (Pischiutta et al., 2013). In this case, the Goebel et al., 2017). In the north-western part of the study area, the
polarization as well as the blanking of seismic reflection signal observed position of the FW-SW was found approximately at 0 m b.s.l. for L1
in the high emission area can be used only as a general indication of (Fig. 4a, located 100 m far from the coastline), 15 m for L2 (Fig. 5, 1.1
anomalous zones, since their quantitative interpretation needs further km) and 8 m b.s.l. for L3 (Fig. 4b, 1.4 km). Therefore, the SW level is not
theoretical and experimental investigations which are beyond the scope only correlated to the distance from the coastline, but also to the hy­
of this paper as well as the detection of the causes of degassing. In this drogeology (groundwater preferential pathways) and to anthropogenic
regard, previous studies (Bigi et al., 2014; Maffucci et al., 2022) sug­ causes (i.e. water abstraction for irrigation of the farms widespread in
gested that gas emissions are both natural and human-induced, since the study area) which can in turn favor the SW intrusion inland. The
faults control the deeper fluid migration pathways, allowing the low effect of hydrogeological and anthropogenic factors is magnified by the
permeability levels of the Lower Pleistocene to be supplied from depth, SW level (around 13 m b.s.l.) reconstructed for L4 line (located 3.6 km
while human activities (primarily drilling) that cut the cover or reduce far from the coastline), which is comparable to those detected for L2 and
the lithostatic head, can allow the pressured gas to reach the surface. L3 despite the increased distance from the sea. As a further confirmation

10
G. De Donno et al. Engineering Geology 327 (2023) 107336

Fig. 12. Data integration at (a) Site 1 and (b) Site 2 where imaged gas upwelling flows are marked with black arrows. Site 1: resistivity model for the L2 line (Fig. 5),
where the Vs model from inversion of array measurements is superimposed. Site 2: resistivity model for the L4 line (Fig. 9), where HR seismic sub-bottom profile A’-B′
and current density model (only values > 0.05 mA/m2) are superimposed. ERT and HR seismic profiles are spaced 90 m apart on average. The dashed white line
indicates the bottom of the embankment. Green pins indicate HVSR stations, while red pins indicate CO2 (black-filled circles) and CO2 + CH4 (black-filled star)
anomalies after Bigi et al. (2014). TDS: Tiber Depositional Sequence, PGS: Ponte Galeria Sequence. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)

Table 2
Main targets theoretically achievable by the proposed methods for gas-prone areas in coastal environments. ✓: detectable, ⨯: not detectable.
Method Gas upwelling flows Reservoir location Near-surface geology Saline intrusion

ERT ✓ ✓ ✓ ✓
Sub-bottom profiling ✓ ✓ ✓ ⨯
Ambient noise (array) ⨯ ✓ ✓ ⨯
SP ✓ ⨯ ⨯ ⨯

of the latter result, similar SW levels, even if slightly shallower (8–10 m emissions in coastal environments, like the one encountered in the
b.s.l.), were also found by a previous ERT survey in the adjacent Ostia Fiumicino area, where borehole drilling is strongly limited if not
Antica archaeological area (Cardarelli et al., 2016). These findings can forbidden. The seismic sub-bottom profile in the Tiber River com­
pave the way for a future large-scale campaign focused on the assess­ plemented with the multibeam bathymetry, gives an HR image of the
ment of the SW intrusion inland in the Fiumicino coastal area. subsurface down to a depth of ~60 m, locating the unconformity surface
between Pleistocene/Holocene clayey sediments and PGS gravels and
6. Conclusions highlighting several blank zones likely associated with the gas
emissions.
This work demonstrated the diagnostic potential of integrating Through the combination of ERT (DOI ~ 80 m) and SP (DOI ~ 15 m)
active (ERT and HR sub-bottom seismic profiler with multibeam ba­ methods on a selected site, we reconstructed a three-layer model, where
thymetry) and passive (SP and HVSR) geophysical data for imaging gas local increases in resistivity in the middle clayey layer are related to

11
G. De Donno et al. Engineering Geology 327 (2023) 107336

upwelling gas flows from the underlying gravel layer (gas reservoir). Bosman, A., Orlando, L., 2017. Bedforms on the Lowermost Reach of the Tiber River
(Rome, Italy): Preliminary Results from Integrated Geophysical Surveys and
The analysis of ambient noise recordings highlights the seismic
Samplings. Atlas of Bedforms in the Western Mediterranean. ISBN: 978-3-319-
impedance contrast between the TDS finer sediments and the underlying 33938-2. https://doi.org/10.1007/978-3-319-33940-5_16.
gravels, while a clear directional effect is seen nearby the main gas Bosman, A., Casalbore, D., Anzidei, M., Muccini, F., Carmisciano, C., 2015. The first
emissions. ultra-high resolution Digital Terrain Model of the shallow-water sector around Lipari
Island (Aeolian Islands, Italy). Ann. Geophys. 58 (2), 2015. https://doi.org/
The integration of these methodologies also provides a more accu­ 10.4401/ag-6746.
rate reconstruction of the distribution of gases in the shallow subsurface Byrdina, S., Revil, A., Pant, S.R., Koirala, B.P., Shrestha, P.L., Tiwari, D.R., Gautam, U.P.,
down to 80 m, highlighting six main points of ascent. This information is Shrestha, K., Sapkota, S.N., Contraires, S., Perrier, F., 2009. Dipolar self-potential
anomaly associated with carbon dioxide and radon flux at Syabru-Bensi hot springs
useful for local hazard assessment, which represents a critical issue in in Central Nepal. J. Geophys. Res. 114, B10101.
the management of urban areas affected by gas emissions. Capon, J., 1969. High-resolution frequency-wavenumber spectrum analysis. In:
Proceedings IEEE, 57, pp. 1408–1418.
Carapezza, M.L., Tarchini, L., Granieri, D., Martelli, M., Gattuso, A., Pagliuca, N.,
CRediT authorship contribution statement Ranaldi, M., Ricci, T., Grassa, F., Rizzo, A., Pizzino, L., Sciarra, A., 2015. Gas blowout
from shallow boreholes near Fiumicino International Airport (Rome): gas origin and
hazard assessment. Chem. Geol. 407, 54–65.
Giorgio De Donno: Conceptualization, Investigation, Methodology,
Carcione, J.M., Gei, D., Picotti, S., Michelini, A., 2011. Cross-hole electromagnetic and
Project administration, Software, Supervision, Writing – original draft, seismic modeling for CO2 detection and monitoring in a saline aquifer. J. Pet. Sci.
Writing – review & editing. Alessandro Bosman: Data curation, Eng. 162–172.
Investigation. Ettore Cardarelli: Conceptualization, Project adminis­ Cardarelli, E., De Donno, G., 2017. Multidimensional electrical resistivity survey for
bedrock detection at the Rieti Plain (Central Italy). J. Appl. Geophys. 141, 77–87.
tration, Supervision, Funding acquisition, Investigation. Michele Cer­ https://doi.org/10.1016/j.jappgeo.2017.04.012.
cato: Conceptualization, Investigation, Validation. Giuliano Milana: Cardarelli, E., De Donno, G., Scatigno, C., Oliveti, I., Martinez, M.P., Prieto-Taboada, N.,
Data curation, Funding acquisition, Investigation, Supervision. Guido 2016. Geophysical and geochemical techniques to assess the origin of rising damp of
a Roman building (Ostia Antica archaeological site). Microchem. J. 129, 49–57.
Penta de Peppo: Data curation, Software. Ciotoli, G., Etiope, G., Marra, F., Florindo, F., Giraudi, C., Ruggiero, L., 2016. Tiber delta
CO2-CH4 degassing: a possible hybrid, tectonically active sediment-hosted
geothermal system near Rome. J. Geophys. Res. Solid Earth 121 (1), 48–69.
Declaration of Competing Interest De Donno, G., Cardarelli, E., 2017. VEMI: a flexible interface for 3D tomographic
inversion of time- and frequency-domain electrical data in EIDORS. Near Surf.
Geophys. 15, 43–58.
The authors declare that they have no known competing financial Fleury, M., Deschamps, H., 2008. Electrical conductivity and viscosity of aqueous NaCl
interests or personal relationships that could have appeared to influence solutions with dissolved CO2. J. Chem. Eng. Data 53 (11), 2505–2509.
Gélis, C., Revil, A., Cushing, M.E., Jougnot, D., Lemeille, F., Cabrera, J., De Hoyos, A.,
the work reported in this paper.
Rocher, M., 2010. Potential of Electrical Resistivity Tomography to Detect Fault
zones in Limestone and Argillaceous Formations in the Experimental Platform of
Data availability Tournemire, France. Pure Appl. Geophys. 167, 1405–1418.
Goebel, M., Pidlisecky, A., Knight, R., 2017. Resistivity imaging reveals complex pattern
of saltwater intrusion along Monterey coast. J. Hydrol. 551, 746–755.
The data that support the findings of this study are available from the Gresse, M., Vandemeulebrouck, J., Byrdina, S., Chiodini, G., Revil, A., Johnson, T.C.,
corresponding author upon reasonable request. Ricci, T., Vilardo, G., Mangiacapra, A., Lebourg, T., Grangeon, J., Bascou, P.,
Metral, L., 2017. Three-dimensional electrical resistivity tomography of the Solfatara
crater (Italy): Implication for the multiphase flow structure of the shallow
Acknowledgments hydrothermal system. J. Geophys. Res. Solid Earth 122 (11), 8749–8768.
Hamilton, E.L., 1979. Vp/Vs and Poisson’s ratios in marine sediments and rocks.
J. Acoust. Soc. Am. 66 (4), 1093–1101.
This work was founded by “Sapienza” University of Rome, Italy,
Hansell, A., Oppenheimer, C., 2004. Health hazards from volcanic gases: a systematic
Grant no. 171/D/2016 and INGV, Italy, Grant. 961 - Principal In­ literature review. Archiv. Environ. Health Intern. J. 59 (12), 628–639.
vestigators: Ettore Cardarelli (2016-18) and Giorgio De Donno (2019- Jardani, A., Revil, A., Bolève, A., Dupont, J.P., 2008. Three-dimensional inversion of self-
potential data used to constrain the pattern of groundwater flow in geothermal
22). The authors wish to thank Francesco Pugliese (“Sapienza” Univer­
fields. J. Geophys. Res. Solid Earth 113 (B9).
sity of Rome) and MSc and BSc students Francesco Giaconi and Michele Johansson, S., Rosqvist, H., Svensson, M., Dahlin, T., Leroux, V., 2011. An alternative
Tomassetti (“Sapienza” University of Rome) for their help during field methodology for the analysis of electrical resistivity data from a soil gas study.
acquisition. Geophys. J. Int. 186 (2), 632–640. https://doi.org/10.1111/j.1365-
246X.2011.05080.x.
Konno, K., Ohmachi, T., 1998. Ground-motion characteristics estimated from spectral
References ratio between horizontal and vertical components of microtremor. Bull. Seismol.
Soc. Am. 88 (1), 228–241. https://doi.org/10.1785/BSSA0880010228.
Lambert, M., Schmalholz, S.M., Saenger, E.H., Steiner, B., 2009. Low-frequency
Attwa, M., Günther, T., Grinat, M., Binot, F., 2011. Evaluation of DC, FDEM and IP
microtremor anomalies at an oil and gas field in Voitsdorf, Austria. Geophys.
resistivity methods for imaging perched saltwater and a shallow channel within
Prospect. 57 (3), 393–411. https://doi.org/10.1111/j.1365-2478.2008.00734.x.
coastal tidal flat sediments. J. Appl. Geophys. 75 (4), 656–670.
Lane, J.W., White, E.A., Steele, G.V., Cannia, J.C., 2008. Estimation of Bedrock Depth
Barberi, F., Carapezza, M.L., Ranaldi, M., Tarchini, L., 2007. Gas blowout from shallow
using the Horizontal-to-Vertical (H/V) Ambient-Noise Seismic Method. In:
boreholes at Fiumicino (Rome): induced hazard and evidence of deep CO2 degassing
Symposium on the Application of Geophysics to Engineering and Environmental
on the Tyrrhenian margin of Central Italy. J. Volcanol. Geotherm. Res. 165 (1–2),
Problems, 490-502. https://doi.org/10.4133/1.2963289.
17–31, 15 August 2007.
Legaz, A., Revil, A., Roux, P., Vandemeulebrouck, J., Gouédard, P., Hurst, T., Bolève, A.,
Bard, P.Y., The SESAME Team, 2004. Guidelines for the implementation of the h/v
2009. Self-potential and passive seismic monitoring of hydrothermal activity: a case
spectral ratio technique on ambient vibrations measurements, processing and
study at Iodine Pool, Waimangu geothermal valley, New Zealand. J. Volcanol.
interpretation. In: SESAME European research project EVG1-CT-2000-00026,
Geotherm. Res. 179 (1–2), 11–18.
pp. 1–62 available at: http://sesame.geopsy.org/Papers/HV_User_Guidelines.pdf.
Loke, M.H., Acworth, I., Dahlin, T., 2003. A comparison of smooth and blocky inversion
Barde-Cabusson, S., Finizola, A., Grobbe, N., 2021. A practical approach for self-potential
methods in 2D electrical imaging surveys. Explor. Geophys. 34 (3), 182–187.
data acquisition, processing, and visualization. Interpretation 9 (1). https://doi.org/
Maffucci, R., Ciotoli, G., Pietrosante, A., Cavinato, G.P., Milli, S., Ruggiero, L., Sciarra, A.,
10.1190/int-2020-0012.1.
Bigi, S., 2022. Geological hazard assessment of the coastal area of Rome (Central
Bergmann, P., Schmidt-Hattenberger, C., Kiessling, D., Rücker, C., Labitzke, T.,
Italy) from multi-source data integration. Eng. Geol. 106527.
Henninges, J., Gunther, B., Schütt, H., 2012. Surface-downhole electrical resistivity
Marcucci, S., Milana, G., Hailemikael, S., Carlucci, G., Cara, F., Di Giulio, G.,
tomography applied to monitoring of CO2 storage at Ketzin, Germany. Geophysics
Vassallo, M., 2019. The deep bedrock in Rome, Italy: a new constraint based on
77 (6), B253–B267.
passive seismic data analysis. Pure Appl. Geophys. 176 (6) https://doi.org/10.1007/
Bettig, B., Bard, P.Y., Scherbaum, F., Riepl, J., Cotton, F., Cornou, C., Hatzfeld, D., 2001.
s00024-019-02130-6.
Analysis of dense array noise measurements using the modified spatial
Milli, S., D’Ambrogi, C., Bellotti, P., Calderoni, G., Carboni, M.G., Celant, A., Di Bella, L.,
autocorrelation method (SPAC): application to the Grenoble area. Boll. Geofis. Teor.
Di Rita, F., Frezza, V., Magri, D., Pichezzi, R.M., Ricci, V., 2013. The transition from
Appl. 42, 281–304.
wave-dominated estuary to wave-dominated delta: the late Quaternary stratigraphic
Bigi, S., Beaubien, S.E., Ciotoli, G., D’Ambrogi, C., Doglioni, C., Ferrante, V.,
architecture of Tiber River deltaic succession (Italy). Sediment. Geol. 284, 159–180.
Lombardi, S., Milli, S., Orlando, L., Ruggiero, L., Tartarello, M.C., Sacco, P., 2014.
Mantle-derived CO2 migration along active faults within an extensional basin
margin (Fiumicino, Rome, Italy). Tectonophysics 637, 137–149.

12
G. De Donno et al. Engineering Geology 327 (2023) 107336

Nickschick, T., Flechsig, C., Meinel, C., Mrlina, J., Kämpf, H., 2017. Architecture and Sella, P., Billi, A., Mazzini, I., De Filippis, L., Pizzino, L., Sciarra, A., Quattrocchi, F.,
temporal variations of a terrestrial CO2 degassing site using electric resistivity 2014. A newly-emerged (August 2013) artificially-triggered fumarole near the
tomography and self-potential. Int. J. Earth Sci. 106 (8), 2915–2926. Fiumicino airport, Rome, Italy. J. Volcanol. Geotherm. Res. 280, 53–66.
Ohrnberger, M., Schissele, E., Cornou, C., Bonnefoy-Claudet, S., Wathelet, M., Souied, Ahmed A., Jardani, A., Revil, A., Dupont, J.P., 2013. SP2DINV: a 2D forward and
Savvaidis, A., Scherbaum, F., Jongmanns, D., 2004. Frequency wavenumber and inverse code for streaming potential problems. Comput. Geosci. 59, 9–16.
spatial autocorrelation methods for dispersion curve determination from ambient Vesnaver, A., Busetti, M., Baradello, L., 2021. Chirp data processing for fluid flow
vibration recordings. In: Proceedings of the 13th World Conference on Earthquake detection at the Gulf of Trieste (northern Adriatic Sea). Bull. Geophys. Oceanogr. 62,
Engineering, Vancouver, Canada, paper no. 946 available at: https://www.iitk.ac. 365–386.
in/nicee/wcee/article/13_946.pdf. Wathelet, M., 2008. An improved neighborhood algorithm: parameter conditions and
Pischiutta, M., Anselmi, M., Cianfarra, P., Rovelli, A., Salvini, F., 2013. Directional site dynamic scaling. Geophys. Res. Lett. 35, L09301. https://doi.org/10.1029/
effects in a non-volcanic gas emission area (Mefite d’Ansanto, southern Italy): 2008GL033256.
evidence of a local transfer fault transversal to large NW–SE extensional faults? Phys. Wathelet, M., Jongmans, D., Ohrnberger, M., Bonnefoy-Claudet, S., 2008. Array
Chem. Earth Parts A/B/C 63, 116–123. performances for ambient vibrations on a shallow structure and consequences over
Qian, R., Liu, L., 2020. Imaging the active faults with ambient noise passive seismics and Vs inversion. J. Seismol. 12, 1–19.
its application to characterize the Huangzhuang-Gaoliying fault in Beijing Area, Wathelet, M., Chatelain, J.L., Cornou, C., Di Giulio, G., Guillier, B., Ohrnberger, M.,
northern China. Eng. Geol. 268, 105520 https://doi.org/10.1016/j. Savvaidis, A., 2020. Geopsy: a user-friendly open-source tool set for ambient
enggeo.2020.105520. vibration processing. Seismol. Res. Lett. 91 (3), 1878–1889. https://doi.org/
Riedel, M., Spence, G.D., Chapman, N.R., Hyndman, R.D., 2002. Seismic investigations of 10.1785/0220190360.
a vent field associated with gas hydrates, offshore Vancouver Island. J. Geophys. Res. Zlotnicki, J., Nishida, Y., 2003. Review on morphological insights of self-potential
Solid Earth 107 (B9). EPM-5. anomalies on volcanoes. Surv. Geophys. 24 (4), 291–338.
Sanders, N.R., 1997. Measuring forecast accuracy: some practical suggestions. Prod.
Invent. Manag. J. 38 (1), 43.

13