CO2 Soil Flux at Vulcano Italy Comparison Between

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223173839

CO2 soil flux at Vulcano (Italy): Comparison between active and passive
methods

Article in Applied Geochemistry · January 2004


DOI: 10.1016/S0883-2927(03)00111-2

CITATIONS READS
67 452

2 authors:

Maria Luisa Carapezza Domenico Granieri


National Institute of Geophysics and Volcanology National Institute of Geophysics and Volcanology
68 PUBLICATIONS 1,569 CITATIONS 81 PUBLICATIONS 3,171 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Volcanogenic SO2 degassing View project

Central America magmatism View project

All content following this page was uploaded by Domenico Granieri on 25 October 2017.

The user has requested enhancement of the downloaded file.


Applied Geochemistry 19 (2004) 73–88
www.elsevier.com/locate/apgeochem

CO2 soil flux at Vulcano (Italy): comparison between


active and passive methods
Maria Luisa Carapezzaa,*, Domenico Granierib
a
INGV—Sezione Roma1, Via di Vigna Murata 605, 00143 Rome, Italy
b
INGV—Osservatorio Vesuviano, Via Diocleziano, 323, 80128 Naples, Italy

Received 26 September 2002; accepted 9 May 2003


Editorial handling by H Ármannson

Abstract
Carbon dioxide soil flux has been used for many years to monitor Italian active volcanoes and both the active
Dynamic Concentration (DCM) and the passive Accumulation Chamber (ACM) methods are employed. These two
methods have been compared by means of 218 simultaneous flux measurements carried out in the La Fossa area of
Vulcano Island, where a large variation of CO2 soil release occurs. Results indicate that DCM overestimates CO2 flux and
is proportional to it only in high flux zones (flux higher than 100 gm2 day1). Using ACM fluxes and the Stefan–Maxwell
equation, the measured CO2/depth curves in the soil could be reproduced. In high flux points CO2 is transported
mostly by viscous flow up to a very shallow depth and then by diffusive flow, which is the dominant gas transport
mechanism in low flux points. Carbon dioxide soil flux values are controlled by proximity to active gas releasing frac-
tures, by changes in the barometric pressure and by variations in soil permeability.
# 2003 Elsevier Ltd. All rights reserved.

1. Introduction therefore be established. The knowledge of these areas is


very important in active volcanoes, like Mt. Etna, where
Over the last years soil degassing, mainly of CO2, has high risk flank eruptions frequently occur and the geo-
been increasingly employed for the study of gas release chemical monitoring of degassing structures as potential
in volcanic and geothermal areas (Bergfeld et al., 2001) sites for opening of new eruptive fissures is a useful
and for the recognition of volcano–tectonic structures complement to seismic and ground deformation mon-
(Barberi and Carapezza, 1994; Giammanco et al., 1998). itoring (Badalamenti et al., 1994a).
Geochemical studies have shown that in both active and Also of great importance is the identification and
quiescent volcanoes, gas is released not only from the control of degassing structures in inhabited areas
craters and fumaroles, but also from particular zones on around quiescent volcanoes, like the island of Vulcano
the flanks and even at the base of volcanoes. The con- or Colli Albani, where high CO2 flux through the soil
tribution of this soil degassing to the total gas output is can produce a dense gas layer on the ground with a
significant (Allard et al., 1991; Chiodini et al., 1998). consequent lethal hazard, as indicated by episodes of
A deep origin (e.g. magmatic vs metamorphic) of the human and animal deaths in these zones (Baxter et al.,
released CO2 can be ascertained by isotopic analyses 1990; Badalamenti et al., 1991; Pizzino et al., 2002;
and the nature of diffuse degassing structures can Carapezza et al., 2003).
Geochemical monitoring of anomalous soil degassing
on the flank of volcanoes may represent, together with
* Corresponding author. Tel.: +39-06-5186-0420; fax: +39- remote sensing methods such as COSPEC and FTIR
06-5186-0549. (Malinconico, 1987; Francis et al., 1998; Bruno et al.,
E-mail address: carapezza@ingv.it (M.L. Carapezza). 1999; Burton et al., 2001), an important surveillance
0883-2927/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0883-2927(03)00111-2
74 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

tool when an eruption is impending and the monitoring al., 1997; Gioncada, 1997). It emerged from the sea
of crater fumaroles becomes impossible. about 120 ka ago and consists of several eruptive centres
All the cases mentioned require frequent measure- that migrated with time from SE to NW. There are two
ments of the CO2 flux or, better, continuous monitoring. active volcanic centres in the northern part of the island:
In Italy, two different methods to measure CO2 flux La Fossa and Vulcanello. The La Fossa cone was built
have been developed and extensively used in volcano up in the last 6 ka whereas Vulcanello emerged from the
monitoring: the accumulation chamber method (ACM) sea in 183 B.C. (Keller, 1980). Both had several histor-
(Chiodini et al., 1998) and the dynamic concentration ical eruptions, the most recent of which occurred at La
method (DCM) (Gurrieri and Valenza, 1988). Presently Fossa in 1888–1890 and gave name to the ‘‘vulcanian’’
there are 12 automatic stations monitoring CO2 flux on explosive eruption type (Mercalli and Silvestri, 1891;
Italian volcanoes: 4 (DCM) at Vulcano Porto, the other Walker, 1973). Since the end of the last eruption in
8 (ACM) are placed at Stromboli (2), Vesuvius (2), 1890, the La Fossa cone has been in a quiescent state,
Campi Flegrei (2) and Etna (2). In addition, monthly, or characterised by an intense fumarolic activity on the
more frequently in case of crisis, field surveys are carried rim, inside the crater and at the base of the volcanic
out on sampling networks at Vulcano and Etna (with edifice (Baia di Levante beach).
DCM) (Badalamenti et al., 1988; Diliberto et al., 1993, The fluids discharged by the La Fossa crater fumar-
2002; Giammanco et al., 1998), and at Vesuvius, Campi oles are the result of a mixing process between (a) a
Flegrei and Cava dei Selci site on Colli Albani with magmatic phase released by a shallow magmatic body,
ACM (Chiodini et al., 2001; Carapezza et al., 2003). The (b) a geothermal phase resulting from the vaporization of
DCM was first utilised in Italy and data collection goes a brine or of laterally infiltrated seawater and (c) a minor
back nearly 15 a (Badalamenti et al., 1988). It is still the surficial component of meteoric origin (Chiodini et al.,
only method systematically used in soil gas monitoring 1995; Nuccio et al., 1999; Tedesco and Scarsi, 1999).
at Vulcano and on the Etna flanks, along with the 2 new During the last century, several episodes of unrest
ACM stations installed in summer 2001. The ACM has occurred at the La Fossa that were not followed by
become progressively more popular for CO2 soil flux eruptions. The first occurred in 1916–1924 when the
measurements, both in Italy and elsewhere (see Bergfeld temperature of the crater fumaroles progressively
et al., 2001). increased from about 100 to 613  C (Sicardi, 1941). A
In order to make a close comparison of these two second increase of crater fumarole temperature and gas
methods they were both simultaneously used in detailed output began in 1977. The temperature reached the
field surveys on the island of Vulcano, over a large soil maximum value of 692  C in January 1993, then
degassing area with a wide range of CO2 fluxes. The decreased to 430  C in February 1996 and increased
field work was planned in order to investigate the dif- again to over 500  C in the same year (Barberi et al.,
ferences between the two methods and their limits and 1991; Capasso et al., 1997). Since then, both gas output
advantages. An attempt is made to quantify a conver- and maximum temperature have decreased slightly to
sion factor between the two data sets which may help in the present value of 340–360  C. During the progressive
obtaining comparable flux values that allow, in case of increase and expansion of the fumarolic activity no sig-
volcanic unrest, interpretation of the different data sets. nificant geophysical anomalies were recorded, apart
In addition to 218 simultaneous field flux measure- from some short seismic swarms and minor inflation
ments, soils were cored and analysed in the laboratory and deflation episodes (Barberi et al., 1991; La Volpe et
to check for the effects of their permeability and grain al., 1997).
size on the CO2 flux. Carbon dioxide concentration gra- The unrest, therefore, mostly consisted of an increase
dients in the soil were measured at some locations to (with fluctuations) in the steam and gas output and
investigate the gas transport mechanism. Finally some maximum temperature of the crater fumaroles. Chemi-
environmental parameters were simultaneously mea- cal and isotopic evidence, especially during rapid chan-
sured to check for their influence on the CO2 soil gas flux. ges, indicated an increasing input of hot magmatic
fluids. An increasing input of deep hot fluids into the
shallow groundwaters at the base of the volcano was
2. Geological and geochemical setting also observed (Chiodini et al., 1995; Tedesco, 1995;
Italiano et al., 1998; Nuccio et al., 1999; Capasso et al.,
Vulcano is the southernmost island of the Aeolian arc 1999, 2001). A magmatic component is indicated by the
that was formed by subduction of the African plate high gas temperature, the nature of the compositional
beneath the European plate (Barberi et al., 1973; Gas- variations and the high 3He/4He ratio (up to 6 times the
parini et al., 1982; Patacca et al., 1990). The island is air ratio: R/Ra=6; Tedesco and Scarsi, 1999; Italiano
entirely made of volcanic rocks, ranging in composition and Nuccio, 1997) a value falling in the compositional
from trachybasalts to shoshonites, latites, trachytes and range for the magmatic He of island arc volcanoes
rhyolites (Keller, 1980; Ellam et al., 1988; De Astis et (Craig et al., 1978; Welhan et al., 1988). Clinopyroxenes
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 75

of the 1888–1890 eruption products have a significantly


lower R/Ra ratio (3.8; Magro, 1997) suggesting that the
present He is being released from a more primitive, and
possibly deeper, magma body than the 1888–1890
magma, which also shows crustal contamination
(Gioncada, 1997). The positive correlation between CO2
and He and the CO2 isotopic composition indicate that
CO2 also derives from magma degassing. The varia-
tions observed in fumarolic discharge during unrest
crises are attributed to variability in the vertical per-
meability induced by earthquakes and self-sealing
hydrothermal processes, that modify the mixing pro-
portion between the magmatic and the hydrothermal
gas components (Chiodini et al., 1992, 1995; Tedesco,
1995). Degassing at different depths by depressurisation
of a saturated magma has also been postulated (Nuccio
et al., 1999).
Soil gas measurements began at Vulcano as part of a
geothermal exploration project (Bertrami et al., 1984).
Carbon dioxide and He concentration anomalies were
found, and shown to be correlated with the main struc-
tural features of the island. In 1984 a systematic mon-
itoring of soil CO2 was initiated. The CO2 flux was
measured at least monthly by DCM on a network of 51
points at the base of the La Fossa cone and in the
inhabited area of Vulcano Porto (Badalamenti et al.,
1988) (Fig. 1). Since 1989, CO2 and He concentrations
at 50 cm depth were also measured monthly using the
same network (Carapezza and Diliberto, 1993; Car-
apezza, 1994, 1996; Diliberto and Gurrieri, 1994;
D’Alessandro and Parello, 1997). Results revealed the
Fig. 1. Location of the CO2 soil flux measurement points.
existence of zones, particularly at the southern base of
Dashed areas indicate the fumarolic fields of the La Fossa cra-
the La Fossa cone (Palizzi zone) and at Baia di ter and Baia di Levante. Coordinates are in UTM 33- ED 50.
Levante (Fig. 1), with strong soil emission of CO2 and Digital Elevetion Model of Vulcano Island by Pareschi et al.
He rich gas, chemically and isotopically similar to the (1999).
gas of the nearest fumaroles (crater and Baia di
Levante) and showing synchronous variations with
them. In 1988 the ACM was first employed at the La soil CO2 flux measurements. In 1994 the authors carried
Fossa (Baubron et al., 1990) and the existence of out a detailed comparative study of the ACM and DCM
strong soil CO2 and He degassing areas with magmatic techniques at this site.
inprint was confirmed.
In the early 1990s, the temperature and gas output
from the La Fossa crater fumaroles increased so strongly 3. Techniques for the measurement of CO2 soil flux
that it became dangerous to sample the fumaroles. As an
alternative, soil gas sampling at the base of the volcano, Many CO2 soil flux measurement techniques have
with particular attention paid to monitoring the CO2 been applied for geological and agricultural aims and
flux, was intensified. Continuous automatic stations include both direct and indirect methods (e.g. Reiners,
using DCM were installed at the base of the La Fossa 1968; Kucera and Kirkham, 1971; Kanemasu et al.,
cone (Badalamenti et al., 1994b) for recording the CO2 1974; Parkinson, 1981). Direct methods are called either
soil flux. active or passive, whether or not a flux of air is used to
These studies have shown a wide variation of CO2 extract gas from the soil. In the present study, the
flux from the soil around the La Fossa cone, ranging authors have utilised the active ‘‘dynamic concen-
from 0.2 to 2900 gm2 day1 (Chiodini et al., 1998), i.e. tration’’ (DCM) and the passive ‘‘accumulation cham-
from values typical of vegetated soil to those of actively ber’’ (ACM) methods. The measurement of CO2
degassing volcanic zones. This area therefore appeared concentration profiles in the soil, in some of the loca-
to be an ideal site to test the different techniques for tions where CO2 flux by DCM and ACM had been
76 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

measured, allows an estimate of the prevailing gas pump) and on soil permeability. For the loose pyro-
transport regime (Evans et al., 2001). clastic-alluvium Vulcano soil, a constant M value of
The DCM (Gurrieri and Valenza, 1988) is an ‘‘active’’ 3.8*102 cm/s was adopted assuming a constant poros-
method shown schematically in Fig. 2. The gas is ity coefficient (P=0.34) and a constant soil permeability
extracted from the soil at about 50 cm depth by means (24 darcy) (Gurrieri and Valenza, 1988; Badalamenti et
of a pump. The aspiration flow of the pump is constant al., 1988). It will be shown that this assumption is
and if CO2 flux from the soil is lower than that of the incorrect, as the Vulcano soil permeability varies by
pump, some outside air enters the system, mixes with over one order of magnitude and the need to estimate
the soil gas and the mixture is brought to the analyser. the factor M at each measurement point appears to be a
This should maintain unaltered the natural flow condi- strong limitation to the use of DCM.
tions around the measuring site. Once pumping, the DC Witkamp and Frank (1969) have shown that the CO2
measurement is obtained when the CO2 concentration flux tends to be overestimated when DCM is used and
of the analysed gas–air mixture reaches a constant according to Kanemasu et al. (1974) and Falcone (1994)
value. According to Gurrieri and Valenza (1988) the results are strongly affected by the physical modifi-
dynamic concentration is directly proportional to CO2 cations induced by pumping under different flux
flux by the following relationship: regimes. In the opinion of Tonani and Miele (1991)
t ¼ M CD ð1Þ DCM would be proportional to flux for high CO2 flux
values (5103 cm/s or 1690 gm2 day1) only.
where t is the CO2 flux (in cm/s), CD is the measured The ACM (see Fig. 3 for technical details) is a direct,
CO2 dynamic concentration (expressed in mole fraction) passive method based on the measurement of the initial
and M is a factor depending on the experimental device CO2 concentration increase inside a chamber of known
(geometry and length of the probe; flow rate of the volume, placed on the soil. This increase is directly pro-
portional to the CO2 flux (Tonani and Miele, 1991). The
method is based on the mass balance of the gas in the
volume of the chamber:
HdcðtÞ=dt ¼ t ð1  cðtÞÞ ð2Þ

Fig. 2. Device for the measurement of the CO2 dynamic con- Fig. 3. The accumulation chamber for CO2 soil flux measure-
centration (modified after Gurrieri and Valenza, 1988). The 50 ment (modified after Chiodini et al., 1998). The 10 cm high and
cm long and 1.3 cm wide steel tube (a), laterally perforated in 40 cm diameter circular chamber is placed on the ground. The
the lowest 15–20 cm, is rammed into the ground. A small Cu soil gas is pumped to an IR spectrometer after the removal of
probe (b) is inserted into it and a rubber plug (c) seals the sys- water vapour by a filter filled with Mg(ClO4)2 and it is then
tem. The soil gas inside the tube is connected to the atmosphere reinjected into the chamber. The spectrometer signal is
through a second small tube inserted into the plug (d). The gas acquired by the AD converter and transmitted to the palmtop
is brought to an IR spectrometer analyser (IRS) by a pump (e). computer.
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 77

where H is the height of the AC, dc(t)/dt is the increase on the ground with its center at 30–70 cm distance from
of CO2 concentration with time within the chamber, t the DCM probe. At the other points DCM flux was
is the total CO2 flux entering from the soil into the measured by placing the probe at the centre of the sur-
chamber, c(t) is the CO2 concentration at time t. face covered by the ACM chamber.
Integration of (2) gives: At 45 points on the permanent grid a soil core was
ðt =HÞ drilled to about 50 cm depth (to 100 cm depth in 5 high
cðtÞ ¼ 1  e ð20 Þ
flux points, with a recovery of 72–83 cm). All the cores
and assuming that ct=0=0 were collected in the proximity (less than 50 cm dis-
dc=dt ¼ t=H ð3Þ tance) of the measurement points, and analysed in the
laboratory for the physical characteristics of the soil.
In the field, the gas–air mixture within the ACM is
continuously analysed by an IR spectrometer and the
concentrations are recorded on a palmtop computer, 5. Laboratory analysis of soil samples
where the concentration/time curve may be seen. A
program gives the flux value (in cm/s or ppmV/s) once In order to not disturb the soil samples, permeability
the operator has selected the portion of the curve for the was measured directly using the Al casing of the 45
initial concentration gradient calculation. The ACM cores, whose length varied from 33.5 to 59 cm, but was
does not require any specific assumption and flux can be of about 50 cm in most samples. Water permeability
measured without knowing the soil characteristics. The was measured by a variable load permeameter, after
method has been tested by several authors under con- saturation in water. Results are shown in Table 1 and in
trolled laboratory conditions. Chiodini et al. (1998) the histogram of Fig. 4.
measured values within 15% of the imposed flux, The variation is very wide (0.1–59.9 darcy). Large
whereas Evans et al. (2001) recognized a systematic differences are found also for cores drilled only a few
underestimation of the flux (average of 12.5%), over dozens of meters apart, indicating important lateral
an imposed flux range of 200–12,000 gm2 day1. variations in the Vulcano soil characteristics even on a
In a field ACM reproducibility test carried out at two small scale. In all the 5 longer cores, the permeability of
points with high and low CO2 flux and consisting the layers below 50 cm depth (22–33 cm thick, depend-
respectively of 20 and 12 consecutive measurements on ing on recovery) was found to be much lower than in the
the same site, the authors found an uncertainty of 12% upper 50 cm. It is therefore clear that a constant per-
(high flux, avg=997 gm2 day1) and of 24% (low flux, meability value cannot be assumed to convert DC into
avg=5.7 gm2 day1). flux values as used by Gurrieri and Valenza (1988),
Badalamenti et al. (1988) and Diliberto et al. (1993 and
2002).
4. Field work Grain size analysis was made on 25 soil cores. In the 5
longest cores, analysis was carried out separately above
The field work was carried out in July–mid September and below 50 cm, so that a total of 30 grain size analyses
1994. Measurement points were distributed over a 2.2 were carried out, representative of 25 different CO2 flux
km2 area, at Vulcano Porto and at the base of the La measurement points.
Fossa cone (Fig. 1). It consisted of 218 simultaneous
measurements of CO2 soil flux by DCM and ACM and
particularly of:

A survey on 47 points of the permanent network


currently used for monthly measurements of the CO2
flux by the DCM (Badalamenti et al., 1988; Diliberto
et al., 1993 and 2002) (Fig. 1);
Ten surveys on 11 selected points at the base of the La
Fossa cone to monitor the flux variation with time (9 of
the permanent network and 2, 4M and C4, newly iden-
tified in the highly degassing Palizzi zone) and several
additional measurements on the Palizzi zone points;
Two surveys to investigate short time and small scale
lateral variations. Fig. 4. Frequency histogram of Vulcano soil permeability
measured in 50 cm cores. Asterisk indicates the value (24 darcy)
On the network, DCM flux was measured in the per- used by Gurrieri and Valenza (1988) for the conversion of the
manent probes whereas the ACM equipment was placed CO2 dynamic concentration into flux.
78 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

Table 1
Soil permeability and average and variation range of ACM and DCM CO2 soil fluxesab

Point K DCM DCM c ACM Meas. Point K DCM DCM c ACM Meas.
No. Darcy No.c No. Darcy Avg. No.c
Min Max Avg. Avg Min Max Avg. Min Max Avg. Min Max Avg.

2 2.8 71 167 99 12 16 85 49 14 29 5.5 10 2 4 1


3 34.2 276 610 394 562 32 95 54 13 30 7.5 13 4 5 1
4 17.1 1048 6135 2811 2003 122 624 276 23 32 3.7 13 2 3 1
4M 30.4 6118 27,547 16,709 21,165 465 2265 1252 20 33 6.4 42 11 19 1
C4 32.4 1555 20,111 6405 8620 238 1301 780 12 34 4.6 13 2 1 1
6 8.1 19 55 35 12 1 14 7 12 35 6.0 42 10 32 1
7 15.0 132 82 21 1 36 6.0 42 10 24 1
8 8.0 13 64 25 8 10 21 16 13 38 4.2 64 11 18 1
9 55.3 13 48 23 54 3 16 8 11 39 4.2 71 12 29 1
10 3.0 51 135 81 10 8 33 20 12 40 12.8 39 21 23 1
12 4.6 87 161 119 23 37 96 75 12 41 8.3 10 3 9 1
14bis 59.9 257 641 47 1 41bis 2.0 35 3 14 1
15 41.4 61 225 143 246 13 51 32 2 42 1.9 10 1 8 1
16 6.0 32 51 42 10 4 22 13 2 44 26.7 276 307 23 1
17 55.0 32 74 15 1 45 16.8 19 13 12 1
19 2.6 71 8 94 1 46 20.4 103 151 127 108 64 85 74 2
20 17.7 51 38 82 1 47 14.2 83 50 20 1
21 2.6 35 4 18 1 $0 12.7 10 13 11 7 2 11 6 2
22 0.1 13 0 18 1 81 6.7 48 13 35 1
23 6.0 19 5 21 1 82 35.6 51 76 33 1
24 2.6 3 0 4 1 83 11.4 22 31 27 13 14 17 15 2
25 7.4 19 6 3 1 84 13.0 58 31 19 1
26 2.5 6 1 4 1 85 8.0 19 6 14 1
27 16.2 19 13 15 1 86 8.0 13 295 135 45 10 75 34 10
28 5.5 35 8 15 1
a
Flux values are in gm2 day1
b
DCM c=corrected values
c
Meas. No.=Number of measurements.

Samples are all made up of loose volcaniclastic sedi- reference to the fact that Kt depends on a factor (Co)
ments. According to the Wentworth (1922) classification which likely differs from soil to soil. For this reason, in
all samples fall into the sand field: 3 fine sands,17 med- the following discussion the measured K values will be
ium size sands, 8 coarse sands and 2 very coarse sands. considered.
Soil samples have an average grain diameter ranging The intrinsic (water) permeability measured in the soil
from 0.8 to 2.37 phi (avg=1.04; =1.61). Sorting samples can be used to estimate the apparent (gas) per-
ranges from 0.9 to 2.57 phi (avg=1.74; =0.5); 2 sam- meability by means of the Klinkenberg equation (see
ples are moderately sorted, 15 poorly sorted and 13 very Eq. 16 in Appendix). Thorstenson and Pollock
poorly sorted. Skewness ranges from 0.38 to 0.48 (1989a, b) estimated the Klinkenberg parameter for air
(avg=0.03 and =0.18) with nearly symmetrical dis- (bair) using the following relation with the intrinsic per-
tributions for most samples. meability (K):
The textural characteristics of the analysed samples
bair ¼ 0:11K 0:39 ð5Þ
were used to evaluate their permeability using the fol-
2 2
lowing relation (Krumbein and Monk, 1951): where bair is expressed in N/m and K in m .
2 ð1:31 Þ For a generic gaseous species i, the Klinkenberg
Kt ¼ Co ðDm Þ e ð4Þ
parameter bi can be calculated from:
where Kt is the intrinsic permeability in darcy, Co is a
bi ¼ ðMi =Mair Þ1=2 ðair =i Þ bair ð6Þ
constant (760 darcy/mm2), Dm is the average dimension
of the grains (mm) and  is the sorting in phi units. where MiMair, and airi are the molecular weights
Kt values calculated by Eq. (4) were compared with and the viscosities respectively of i and air.
the measured permeability values (K), neglecting the Using Eq. (5) and the measured intrinsic permeability
samples without grain size uniformity (i.e. presence of values, Eq. (6) can be used for CO2. Results are shown
cm-sized pebbles). The two sets of data show a good in Table 2. Average CO2 flowing pressures have then
correlation (r2=0.796), with Kt usually higher than K been calculated for a set of variable initial pressures.
(only in one case it is lower and in 6 cases the values are Using these data and applying Eq. (16) (see Appendix)
very similar). These differences can be explained with the graphs of Fig. 5 can be obtained. In agreement with
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 79

Table 2 much higher than the intrinsic one and very


Klinkenberg parameter (b) for CO2 in the measured intrinsic difficult ‘‘on site’’ measurements of gas P
permeability range should be made in order to convert K into Kg;
The measured K values used for the conversion
Permeability bAir bCO2
(darcy) (Pa) (atm)
of DC into CO2 flux therefore represent the
‘‘minimum’’ permeability for CO2 of the system,
0.1 12923.873 0.1901 whose real value depends on the unknown
1 5264.931 0.0775 average P of the flowing CO2.
5 2810.570 0.0413
10 2144.829 0.0316
15 1831.121 0.0269
6. Results and discussion
20 1636.783 0.0241
25 1500.362 0.0221
30 1397.383 0.0206 6.1. Comparison between ACM and DCM flux values
35 1315.849 0.0194
40 1249.077 0.0184 Carbon dioxide soil flux values simultaneously mea-
45 1192.998 0.0176 sured by ACM and DCM are reported in Table 1,
50 1144.971 0.0168 together with the (water) permeability (K value) mea-
55 1103.192 0.0162 sured in the nearest soil core. DCM fluxes calculated
60 1066.384 0.0157 using the constant M conversion factor of Gurrieri and
65 1033.609 0.0152 Valenza (1988) (see Eq. 1) are compared, in the same
70 1004.164 0.0148
table, with values obtained by introducing a correction
factor (K/24) accounting for the measured permeability.
As previously discussed, the soil gas permeability is cer-
the Klinkenberg equation, the permeabilities for gas are tainly higher than the water permeability, but it cannot
always higher than those for liquid, and for increasing be estimated if the average pressure of flowing CO2 is
values of the average flowing gas pressure, this differ- not known. Diliberto et al. (2002), presenting 10 a of
ence tends to zero (see Eq. 18 in Appendix). DCM CO2 flux measurements at Vulcano, report that
The following conclusions are reached: no change in the M converting factor has been experi-
mentally observed using a sand with permeability of 32
In most laboratory measurements, gas flux darcy instead of the (24 darcy) sand of the initial Gur-
pressure is 1 bar or higher. Such a pressure gives rieri and Valenza (1988) experiment. Unfortunately the
apparent permeability values close to those of papers do not report how permeability was measured,
intrinsic (water) permeability; or if they refer to water or CO2 permeability, and which
In natural conditions average flowing pressure was the imposed average CO2 flowing pressure. It is
may be very low (Natale et al., 2000; Evans et al., clear from the K values in Table 1 that in most cases
2001) and therefore the apparent permeability is corrected DCM fluxes are lower than uncorrected ones,

Fig. 5. Gas permeability (Kg) as a function of the intrinsic permeability K for different CO2 pressures at depth. The 0.001 curve has to
be read on the right axis, all the others on the left one.
80 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

because the measured K values are lower than the con- in low permeability soils. As there is no reason to use a
stant permeability (24 darcy) assumed for the calcula- single soil permeability value to compute DCM fluxes,
tion of uncorrected fluxes. Consequently uncorrected in the following corrected DCM values will be referred
DCM fluxes are almost always higher than the ACM to.
fluxes measured at the same site (Fig. 6a). However, In the Log–Log diagram for the whole data set
many corrected DCM fluxes in low permeability med- (Fig. 6) a relatively good linear correlation is found
ium–low flux zones (Log ACM flux < 2) are lower than between ACM and DCM fluxes (r2 of 0.86 and 0.74, for
the corresponding ACM flux values (Fig. 6b). Such uncorrected and corrected DCM values, respectively).
behaviour may either reflect the different soil perme- This correlation is influenced by the wide variation field
ability of the surface layer compared with 50 cm depth and it is clear from Fig. 6 that data can be subdivided
(where DCM measurements are made); or result from into a high flux group and a medium-low flux group
severe CO2 distribution disturbance caused by pumping with a separating threshold at a Log ACM flux value

Fig. 6. Comparison of the CO2 soil flux Log values obtained by the ACM with uncorrected (A) and corrected (B) Log flux values
obtained by the DCM at the same points. Straight line is x=y Eq.
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 81

of 2. All values in the high flux group refer to measure- permeability one order of magnitude higher (e.g. points
ments carried out in the Palizzi zone at the southern 9, 17, 82). This indicates that although obviously soil
base of the La Fossa cone (points n. 4, 4M, and C4; permeability affects the CO2 flux at the surface, this
Fig. 1 and Table 1). All the remaining points belong to depends also, and to a relevant extent, on the quantity
the medium–low flux group. In the high CO2 flux group of CO2 that is locally released from the deep source.
(Fig. 7a) DCM values are on average one order of From a geological viewpoint low permeability-medium
magnitude higher than ACM fluxes but the two sets of flux sites are thus as interesting as high permeability-
data are relatively well correlated (r2=0.74). No corre- high flux sites, since both indicate a significant CO2
lation exists between ACM and DCM data in the med- degassing at depth and the presence of gas releasing
ium–low flux group (Fig. 7b). cracks.
It is clear from Table 1 that although high flux values A strong variation of flux with time is observed at all
correspond to soils with a medium–high permeability points where measurements have been repeated 10 or
(17–32 darcy), there is no correlation between soil per- more times in a time span of 2 weeks (see Table 1). Both
meability and CO2 flux. There are low permeability ( < 5 DCM and ACM fluxes vary by a factor of 2–6, with iso-
darcy) soils (e.g. points 2, 10, 12, 19) with a CO2 flux lated peaks up to 22.7 (DCM) and 14 (ACM). The var-
higher than that measured in other soils which have a iations can be only partly explained by barometric

Fig. 7. Log–Log diagram of ACM and corrected DCM CO2 soil fluxes for high (A) and medium-low values (B). Straight line is x=y Eq.
82 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

pressure changes (see Section 6.3) and largely reflect by a large uncertainty even with a dense distribution of
changes in the regime of deep degassing of the La Fossa measurement points.
volcanic system. ACM and DCM CO2 fluxes were measured every 2 h
In conclusion both DCM and ACM identify the same for 24 h, at 3 high flux points in the Palizzi zone (4, 4M,
high CO2 flux sites where Log fluxes (gm2 day1) are C4; Fig. 1). DCM flux was measured on the permanent
respectively higher than 2 (ACM) and 3 (DCM) probes and ACM flux always in the same site at a dis-
(Fig. 7a). In the medium–low flux zone point dispersion tance from the DCM probe of 40, 30 and 32 cm respec-
is large and for the same Log ACM flux a range of Log tively. Results are shown in Table 3. Random variations
DCM flux values over 1.5 order of magnitude can be were observed (16–19% for ACM; 10–12% for DCM)
observed and vice-versa (Fig. 7b). In low flux zones, gas that probably reflect the methods’ uncertainties. No
transport is mostly by diffusion driven by the [CO2] correlation exists between the two sets of data at any
gradient (see Section 7). The different response of DCM point, with DCM fluxes much higher, as usual, than
and ACM could be explained considering that ACM ACM ones. No correlation exists either with the simul-
chambers may have a blocking effect that lowers the taneously recorded environmental parameters. This
[CO2] gradient, whereas DCM pumping favours the indicates that, at least in this case, the possible effects
diffusion of CO2 within the soil. due to variations in barometric pressure (1005–1008
mbar), air temperature (24.5–35.0  C) and air relative
6.2. Small scale and short time variation in CO2 flux humidity (35.0–63.7%) are lower than the measurement
uncertainty.
In the Palizzi area, between the two highly emissive
points 4 and 4M (Fig. 1), 14 ACM and DCM CO2 flux 6.3. Influence of environmental parameters
measurements with 1 m spacing were carried out in 56
min at constant atmospheric pressure, over an area of Soil gas flux and concentration are affected by envir-
only 7 m2. ACM fluxes varied from 68 to 1960 gm2 onmental conditions (atmospheric T, P and humidity,
day1 (avg 473;  541); DCM fluxes from 1010 to 21,130 soil T and moisture, wind speed) (e.g. Reimer, 1980;
gm2 day1 (avg 6760;  5920). The ACM and DCM Hinkle, 1994). The field work was carried out in dry
values are well correlated (r2=0.77), the second being over summer months with stable weather, so only variations
one order of magnitude higher than the first. In the same of atmospheric pressure (atm-P) have to be considered.
zone 24 ACM flux measurements were made in 1 h, moving In high flux zones like Palizzi CO2 is transported mainly
the chamber so that the covered surfaces were adjacent by viscous flow, that is governed by the gas pressure
to each other (40 cm distance between the accumulation gradient (see Eq. 11 and following in Appendix). It is
chamber centers in two successive measurements). The therefore logical that in such zones atmospheric pressure
following results were obtained (gm2 day1): affects soil gas fluxes, and in fact a negative correlation
is found between the two parameters (Table 4). The
min ¼ 169 max ¼ 507 average ¼ 287  ¼ 86: pressure effects are surprisingly somewhat higher on the
DCM flux, based on gas extraction from 50 cm depth,
Results suggest that the surface release of CO2 is than on the ACM flux which is measured at the soil sur-
confined to microcracks that are much narrower than face. Clearly atmospheric pressure changes also affect CO2
the ACM-chamber (40 cm diameter). These data con- at 50 cm depth (see Table 4). No correlation between
firm the large variability of CO2 soil flux even on a very atmospheric pressure and DCM flux is observed for the
small scale and from an apparently homogenous soil. whole set of data at point 4M, but the expected negative
Values differ over one order of magnitude at 4 m dis- correlation is found if the values are divided into two
tance and by a factor of 3–4 at only 40 cm distance. groups corresponding to lower (4 Mb) and higher (4 Ma)
Estimates of total CO2 soil output are therefore affected pressure respectively (Table 4). Chiodini et al. (1998) also

Table 3
Variation of CO2 flux within 24 h

Point Accumulation chamber (gm2 day1) Dynamic concentration corr. (gm2 day1)

Min Max Avg s Min Max Avg s

4 225 385 309 48 2914 4359 3801 397


4M 1205 2434 2035 390 17,299 27,902 23,980 2977
C4 1104 1960 1542 288 1490 2647 2081 388
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 83

Table 4 DCM CO2 soil fluxes were simultaneously measured


Relationship between CO2 soil flux and atmospheric pressure (Table 5). The shape of the CO2-depth curves depends
on the flux. Three representative profiles are shown in
Point Avg DCM flux— ACM flux— CO2a—
Fig. 8. The profile changes from nearly linear at point 8
ACM flux Atm P Atm P Atm P
(gm2 day1) (r2) (r2) (r2)
(CO2 flux=18 gm2 day1) to moderately curved at
point 4 (CO2 flux=473 gm2 day1) to highly curved at
4 287 0.76 0.30 0.69 point 4M (CO2 flux=1893 gm2 day1). The CO2 pro-
4M 1546 0.08 0.43 0.27 file shapes in the Vulcano natural sandy soil are very
4M a 1798 0.73 0.44 similar to those found by Gurrieri and Valenza (1988)
4M b 1326 0.65 0.50 and Evans et al. (2001) in laboratory tests using homo-
C4 1177 0.53 0.49 no data
genous sand and a range of imposed CO2 fluxes. Gur-
a
Concentration of CO2 measured at 50 cm depth. rieri and Valenza (1988) noted that the change in the
profile shape from curved to linear represents the tran-
sition from mainly viscous to mainly diffusive CO2 flux.
observed at Palizzi in 1994 a variation of the P-flux Evans et al. (2001) showed that the total viscous flow of
relation with time. Both the authors’ data and those of gas (a mixture of CO2 and air) does not change close to
Chiodini et al. (1998) indicate that CO2 soil flux for the the soil surface, where CO2 transport becomes diffusive.
same point varies by a factor 4–5 with atmospheric In other words a viscous CO2 flow at depth becomes a
pressure variations of 5–9 mbar. Variations in the flux viscous flow of mainly air near the soil surface and the
by 3–4 times are however also observed for the same pressure gradient persists up to the soil surface.
atmospheric pressure value, indicating that factors The evaluation of the relative contribution of the dif-
others than atmospheric pressure (probably related to the fusive and viscous components of the CO2 flux is diffi-
gas source at depth) govern the rate of the CO2 output. cult, mostly because the pressure gradient in the soil is
Under severe weather conditions CO2 flux values are not known (see the review in Appendix of the funda-
perturbated by rapid changes in the atmospheric pres- mental laws governing gas transport in porous media).
sure and by the high soil moisture and wind velocity It is known (Thorstenson and Pollock, 1989b; Evans et
(Hinkle, 1994). Therefore data must be corrected for the al., 2001) that small pressure gradients of only a few Pa
influence of environmental parameters in order to can generate viscous fluxes that overwhelm diffusive
obtain reliable information on the CO2 release from fluxes. Following the examples provided by Thorstenson
depth. DCM and ACM automatic stations can even and Pollock (1989b) the authors used the Stefan–Max-
operate under a thick snow cover which, however, well equation and the measured CO2 flux in order to
modifies the CO2 concentration in the soils and hence calculate the CO2 distribution with depth and compare
the flux (Badalamenti et al., 1994a; Gerlach et al., 2001). the values obtained with those measured in the field. In
order to do this it must be assumed that the numerical
effect of the Knudsen diffusivity and the total pressure
7. CO2 concentration profiles and gas transport gradients in the constitutive equation are negligible,
mechanism in the Vulcano soil whereas the Klinkenberg parameter may be estimated
by Eq. (6). Input data and results are shown in Fig. 8.
Carbon dioxide concentration profiles were obtained The computed CO2 values fit the measured ones well
at 3 points (4, 4M and 8; Fig. 1) by analysing the gas with the exception of the deepest part at point 4M. Here
extracted from 50, 75 and 100 cm depth (also from 25 the gas composition is almost pure CO2, transport is
cm at 4M) using a probe inserted into the soil with 3 (4 dominated by viscous flux and thus the Stefan–Maxwell
in 4M) small channels (4 mm diameter). The ACM and equation is no longer adequate to describe the system.

Table 5
Variation of CO2 concentration with depth compared with ACM and DCM soil fluxes

Point Range of CO2 concentration at various depths (vol%) Range of Range of Measur.
No. ACM flux DCMc flux No.a
25 cm 50 cm 75 cm 100 cm (gm2 day1) (gm2 day1)

4 21.6–50.9 30.8–67.9 38.2–76.9 155–624 747–4371 32


4M 42.4–90.1 54.1–96.1 68.3–100 68.3–100 465–1983 7985–34850 16
8 0.4–1.5 0.6–3.0 1.2–4.6 1.2–4.6 10–21 10
a
Data refer to the number of simultaneous measurements of all parameters.
84 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

Fig. 8. CO2 measured at different depths in the soil of points 4, 4M and 8 (large full symbols). Dashed lines are the CO2/depth profiles
computed with the Stefan–Maxwell equation. In all cases DO2CO2=DN2-CO2=1.64e–5 m2/s. For each point the values of ACM flux
(F in gm2 day1), atmospheric pressure (P in mbar) and temperature (T in  C) and of nt are indicated.

Note that at the low flux point 8 the measured CO2 8. Conclusions
gradient is perfectly reproduced by assuming that the
soil is permeated by air down to a depth of 30 cm. Carbon dioxide soil flux values measured with the
Using the DGM (Dusty Gas Model see Eq. 9 in DCM are frequently over one order of magnitude
Appendix) Evans et al. (2001) were able to calculate the higher than those measured with the ACM. In high flux
ratio of various flux components (diffusive vs viscous) to zones they are relatively well correlated whereas no
total molar flux and showed that the percentage of CO2 correlation exists in medium-low flux zones (AC flux
transported by viscous flux is at any depth approxi- < 100 gm2 day1). In high flux zones the prevailing gas
mately equal to its concentration at that depth. Such a transport mechanism is by viscous flow, and the CO2 in
relation holds for medium to coarse dry soils at low the soil at any depth is approximately equal to the per-
elevations where abiogenic CO2 flux predominates over centage of CO2 transported by viscous flow (Evans et
soil respiration flux. This is certainly the case in the al., 2001). This can explain the good correlation found
Palizzi zone, and from Fig. 8 it can be inferred that at between DCM and ACM fluxes in high flux zones, as
point 4M, where the highest surface flux is recorded, DCM values are related to CO2 at 50 cm depth, and
CO2 transported by viscous flow is nearly 100% at 100 hence to the efficiency of gas transport toward the sur-
cm depth, and still over 80% at 25 cm depth. The per- face. In low flux zones the gas is instead transported
centage of CO2 transported by viscous flow decreases as mainly by diffusion and the relation CO2-total flux does
the measured surface flux decreases. It is still high (75%) not hold anymore. In these zones the DCM pumping
at 100 cm depth at point 4, whereas it is only a few creates a pronounced dilution effect that results in a
percent at the same depth at the low flux point 8, where forced in-diffusion of CO2 at the base of the probe.
CO2 transport is mainly by diffusion. These results confirm that DCM overestimates CO2 flux
Using the DCM (corrected) flux instead of the ACM and that DCM is proportional to flux only for high flux
ones in the Stefan–Maxwell equation, there is no way of values, in agreement with the finding of Witcamp and
reproducing the CO2 gradient measured at point 4M, Frank (1969) and Tonani and Miele (1991). An indica-
whereas an unrealistic porosity value (0.80) has to be tion that ACM fluxes correspond to real soil fluxes
assumed at point 4. This confirms that the DCM over- better than DCM ones comes from the CO2 curves
stimates CO2 flux. Although there are no CO2 values for measured at some high flux points. These in fact are well
depths lower than 25 (point 4M) and 50 cm (point 4) the reproduced using the Stefan–Maxwell equation and the
profile shapes confirm the indication by Evans et al. measured ACM fluxes, but cannot be reproduced using
(2001) that CO2 transport which is mainly viscous at the DCM flux values. Although DCM is based on gas
depth becomes diffusive close to the surface. extraction from 50 cm depth and ACM from surface
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 85

measurements of CO2 increase within a chamber, baro- If the two species i and j have different molecular
metric pressure has the same relevant influence on both weights (Mi >Mj), an initial flux of j toward i occurs.
DCM and ACM fluxes, that vary by a factor of 4–5 This generates an infinitesimal increase of pressure in
with variations of atmospheric pressure of 5–9 mbar. the region occupied by the i species and as a con-
This simply indicates that atmospheric pressure varia- sequence a viscous flux of i versus j arises. In a free sys-
tions affect the CO2 at 50 cm depth as well as the surface tem, without walls, the pressure increase is immediately
CO2 flux and there is some evidence for this from the dissipated and the system remains in a steady state.
CO2 measurements in the soil. Diffusion through natural soil requires, however, a
Comparison between the two sets of data is difficult more complex model accounting for two important
because of the lack of data on the soil permeability for additional factors.
CO2 that is required to convert DCM into flux at each The coefficient Dij has to be corrected for the porosity
measurement point. The data show that the (water) (n) and tortuosity (t) of the soil (Penman, 1940). The
permeability is highly variable even in apparently corrective factor is called diffusibility (Qm) and modifies
homogeneous soils like the ones of Vulcano. However the term Dij in Deij:
CO2 soil permeability cannot be evaluated as the aver- Deij ¼ Qm Dij ð8Þ
age CO2 flowing pressure is not known. This appears to
be a strong limitation for DCM because of the difficulty Abu-El-Shar’s and Abriola (1997) found a value of
of carrying out reliable in situ soil gas permeability Qm=0.435n for sands similar to the Vulcano soils. For
measurements, which are not needed for ACM. porosity typical of sands (0.30–0.40), Deij is much smaller
Automatic continuous stations of either DCM than Dij (0.13–0.17).
(Badalamenti et al., 1994b) and ACM (Rogie et al., A second difficulty arises from the physical structure
1998) have proved to be reliable and a long term com- of soil, which can be described as a multitude of wide-
parison between them is under way at Stromboli. spread channels, constituting a system with walls. In this
Large variability of CO2 soil flux is observed even on case the gas moving through the capillary cannot dis-
a very small scale (up to over one order of magnitude in sipate the increment of pressure and the viscous flux is
points with a spacing of only a few meters). For the not negligible. To account for this, Mason et al. (1967)
purpose of volcano surveillance it is therefore con- proposed the Dusty Gas Model (DGM) which considers
venient either to use continuous recording stations or to the soil–gas system as a multi-component gaseous mix-
establish a permanent network to be periodically reoc- ture where soil grains are assimilated to stagnant gas
cupied for the measurement of CO2 flux, as it is cur- giant molecules. For an isothermal and isobaric system
rently done at some high flux Italian volcanoes. In any of n components, the DGM general equation, expressed
case the control of CO2 soil flux is by now confirmed as in terms of diffusive molar fluxes, is given by (Thor-
a powerful tool for investigating and monitoring volca- stenson and Pollock, 1989a, b):
nic, seismic and geothermal areas.      D
X xi ND j xj NDi DGM N rPi
DGM
e  i kDGM ¼
j¼1j6¼i
D ij D i RT ð9Þ
Acknowledgements
i ¼ 1; :::;
This work was supported by the National Volcanic
Group of Italy (GNV). The review of Dr. Niels where xi and xj are the mole fraction of gases i and j, ND i
Oskarsson provided several useful suggestions. e ND e
j are the total molar diffusive flux, Dij represents the
effective binary diffusion coefficient of species i in j given
by Eq. (8) and Dki is the Knudsen diffusivity. Note that
Appendix. Gas transport in natural porous media Eq. (9) is referred to an isobaric system (Ptot =cost) and
the term related to the viscous flux is still not considered.
Fick’s first law is generally used to describe the diffu- The Knudsen diffusivity arises from collisions
sion of a binary gaseous system. Under isothermal and between moving gaseous molecules and soil particles.
isobaric conditions, neglecting external forces (gravity, From a physical point of view the second term on the
electrical field, etc.), the diffusive F D i flux can be left of Eq. (9) represents the momentum transferred
expressed by: from gaseous species to the soil grains. The first term on
the left represents the momentum lost through mole-
FD
i ¼ nDrx ð7Þ
ij i cule–molecule collisions.
Two extreme cases are possible:
where n is the total mole fraction; Dij the binary diffu-
(a) Di k >> Di e
sion coefficient of species i in j; xi the mole fraction of
component i. (b) Di k <<Di e.
86 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

In case (a) the Knudsen diffusivity is negligible and or in the form:


Eq. (9) reduces to the Stefan–Maxwell equation (Thor-  k
T Di  P
stenson and Pollock, 1989a, b). This condition occurs N ¼ þK rP ð13cÞ
P RT
with high soil permeability and/or high gas pressure.
In case (b) the major contribution is given by mole-
cule–particle collisions. This occurs when the pore dia- Eq. (13c) suggests that the total molar flux can be
meter is small with respect to the gas molecule size and/ seen as a viscous flux governed by Darcy’s law (Eq. 12)
or for low gas pressure. where the intrinsic permeability (K) is replaced by
Under non-isobaric condition, the Pi term of Eq. (9) apparent, or gas phase, permeability (Kg):
becomes:
D ki 
rPi ¼ Prxi þ xi rP ð10Þ Kg ¼  þK ð14Þ
P
where P represents the total pressure of the system; the
term rP is the pressure gradient and suggests the and the total molar flux becomes:
appearance of a viscous flux. 
The viscous flux is expressed by Darcy’s law : T P
N ¼ Kg rP ð15Þ
RT
K
q ¼ rP ð11Þ

Equation (14) shows that the gas (apparent) perme-
where q is the volumetric flux, K is the intrinsic perme- ability is higher than the liquid (intrinsic) permeability
ability e  is the dynamic viscosity of the gas. for a slipping effect impressed on the gas molecules
For an ideal gas Eq. (11) can be written as: along the porous medium walls. This phenomenon is
P K known as the Klinkenberg effect or ‘‘slip flux’’.
NV ¼  rP ð12Þ The original relation (Klinkenberg, 1941) is:
RT  
bi
where Nv is the molar viscous flux. Considering that the Kg ¼ K 1 þ  ð16Þ
total flux is given by the sum of viscous and diffusive P
fluxes (NTi=Ndi+xiNv), Eq. (9) becomes: where Kg and K are the gas and liquid permeability, bi is
    the Klinkenberg parameter which depends on the kind of
X xi NTj xj N Ti DGM N T  gas, on the characteristics of the solid medium, on tem-
DGM
 i kDGM perature and pressure and P is the gas average pressure.
j¼1 j6¼i
Deij Di
Eq. (16) is equal to Eq. (14) if the following relation
  ð13Þ is considered (Thorstenson and Pollock, 1989a and
Prxi kP xi rP
¼ þ 1þ k i ¼ 1; :::; 1989b):
RT Di  RT
Dik ¼ K bi =i ð17Þ
that is the DGM general equation under isothermal
conditions (Thorstenson and Pollock, 1989a, b). An increase of P determines a Kg decrease until it
For a single component gas phase, the molecular dif- approaches the liquid permeability:
fusivity vanishes, the molar fraction is equal to 1, and lim Kg ¼ K: ð18Þ
Eq. (13) reduces to: 
P !1

KP rP rP References
NT ¼ NV þ ND ¼   DK
i : ð13aÞ
i RT RT
Abu-El-Shar’s, W., Abriola, L.M., 1997. Experimental assess-
Note that, even in the absence of a gas concentration ment of gas transport mechanisms in natural porous media:
gradient, a diffusive flux can exist as a consequence of parameter evaluation. Water Resour. Res. 33–4, 505–516.
the Knudsen effect. Allard, P., Carbonnelle, J., Dajlevic, D., Le Bronec, J., Morel,
P., Robe, M.C., Maurenas, J.M., Faivre-Pierret, R., Martin,
The terms KP/i and DK i represent, respectively, the
D., Sabroux, J.C., Zettwoog, P., 1991. Eruptive and diffuse
viscous and diffusive component in the total flux equa-
emission of CO2 from Mount Etna. Nature 351, 387–391.
tion and P represents the average pressure of gas (P). Badalamenti, B., Gurrieri, S., Hauser, S., Valenza, M., 1988.
Eq. (13a) can be written as follows (Mason and Mal- Ground CO2 output in the island of Vulcano during the
inauskas, 1983): period 1984–1988: gas hazard and volcanic activity surveil-
 lance implications. Rend. Soc. It. Mineral. Petrol., Car-
Dki PK
NT ¼  þ rP ð13bÞ apezza Mem. Vol 43, 893–899.
RT RT
M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88 87

Badalamenti, B., Gurrieri, S., Nuccio, P.M., Valenza, M., 1991. degassing. In: Data related to eruptive activity, unrest phe-
Gas hazard on Vulcano island. Nature 350, 26–27. nomena and other observation on the Italian active volca-
Badalamenti, B., Capasso, G., Carapezza, M.L., D’Alessandro, noes in 1991. Acta Volcanol. 3, 273–276.
W., Di Gangi, F., Diliberto, I. S., Giammanco, S., Gurrieri, Carapezza, M.L., Badalamenti, B., Cavarra, L., Scalzo, A.,
S., Nuccio, P. M., Parello, F., Valenza, M., 1994a. Soil gas 2003. Gas hazard assessment in a densely inhabited area of
investigations during the 1991–1993 Etna eruption. Acta Colli Albani volcano (Cava dei Selci, Roma). J. Volcanol.
Vulcanol 4, 135–141. Geotherm. Res. (in press).
Badalamenti, B., Gurrieri, S., Valenza, M., 1994b. Continuous Chiodini, G., Cioni, R., Falsaperla, S., Montalto, A., Guidi,
monitoring (temperature, CO2 in soil gases and reducing M., Marini, L., 1992. Geochemical and seismological inves-
capacity)(Vulcano). In: Data related to eruptive activity, tigations at Vulcano (Aeolian Island) during 1978–1989. J.
unrest phenomena and other observation on the Italian Geophys. Res. 97, 11025–11032.
active volcanoes in 1992. Acta Vulcanol. 6, 46-48. Chiodini, G., Cioni, M., Marini, L., Panichi, C., 1995. Origin
Barberi, F., Carapezza, M.L., 1994. Helium and CO2 soil gas of the fumarolic fluids of Vulcano Island, Italy, and implica-
emission from Santorini (Greece). Bull. Volcanol 56, 335– tions for the volcanic surveillance. Bull. Volcanol 57, 99–110.
342. Chiodini, G., Cioni, R., Guidi, M., Marini, L., Raco, B., 1998.
Barberi, F., Gasparini, P., Innocenti, F., Villari, L., 1973. Vol- Soil CO2 flux measurements in volcanic and geothermal
canism of the Southern Tyrrhenian Sea and its geodynamical areas. Appl. Geochem. 13, 543–552.
implications. J. Geophys. Res. 78, 5221–5232. Chiodini, G., Frondini, F., Cardellini, C., Granieri, D., Marini,
Barberi, F., Neri, G., Valenza, M., Villari, L., 1991. 1987–1990 L., Ventura, G., 2001. CO2 degassing and energy release at
unrest at Vulcano. Acta Vulcanol 1, 95–106. Solfatara volcano, Campi Flegrei, Italy. J. Geophys. Res.
Baubron, J.C., Allard, P., Toutain, J.P., 1990. Diffuse volcanic 106-B8, 16213–16221.
emission of carbon dioxide from Vulcano Island, Italy. Nat- Craig, H., Lupton, J.E., Horibe, Y., 1978. A mantle helium
ure 344, 51–53. component in circum-Pacific volcanic gases: Hakone, the
Baxter, P., Tedesco, D., Miele, G., Baubron, J.C., Cliff, K., Marianas and Mt. Lassen. In: Alexander, Ozima (Eds.),
1990. Health hazards of volcanic gases. The Lancet July 21, Terrestrial Rare Gases. Japan Sci. Soc. Press, Tokyo, pp. 9–
176. 16.
Bergfeld, D., Goff, F., Allard, P., (Eds.), 2001. High CO2 flux D’Alessandro, W., Parello, F., 1997. Soil gas propection of He,
222
measurements in volcanic and geothermal areas, methodolo- Rn and CO2: Vulcano Porto area, Aeolian Islands, Italy.
gies and results. Chem. Geol. 177. Appl. Geochem. 12, 213–224.
Bertrami, R., Ceccarelli, A., Lombardi, S., 1984. L’elio nei gas De Astis, G., La Volpe, L., Peccerillo, A., Civetta, L., 1997.
del suolo nella prospezione geotermica. Rend. Soc. It. Volcanological and petrological evolution of Vulcano island
Mineral. Petrol 39, 341–342. (Aeolian Arc, southern Thyrrenian Sea). J. Geophys. Res.
Bruno, N., Caltabiano, T., Romano, R., 1999. SO2 emissions at 102, 8021–8050.
Mt. Etna with particular reference to the period 1993–1995. Diliberto, I.S., Gurrieri, S., 1994. Soil CO2 degassing in the
Bull. Volcanol 60, 405–411. Island of Vulcano during 1992. Acta Vulcanol 6, 35.
Burton, M. R., Allard, P., Murè, F., 2001. FTIR detection of Diliberto, I.S., Gurrieri, S., Valenza, M., 1993. Vulcano: gas
chemical changes in volcanic gas emissions before and during geochemistry. CO2 flux from the ground. Acta Volcanol. 3,
the 2001 flank eruption of Mt. Etna. In: GNV Assemblea del 272–273.
1 anno, Abst., 216. Diliberto, I.S., Gurrieri, S., Valenza, M., 2002. Relationship
Capasso, G., Favara, R., Inguaggiato, S., 1997. Chemical fea- between diffuse CO2 emissions and volcanic activity on the
tures and isotopic composition of gaseous manifestations on island of Vulcano (Aeolian Island, Italy) during the period
Vulcano Island (Aeolian Islands, Italy): an interpretative 1984–1994. Bull. Volcanol 64, 219–228.
model of fluid circulation. Geochim. Cosmochim. Acta 61 Ellam, R.M., Menzies, M.A., Hawkesworth, C.J., Leeman,
(16), 3425–3440. W.P., Rosi, M., Serri, G., 1988. The transition from calc-
Capasso, G., Favara, R., Francofonte, S., Inguaggiato, S., alkaline to potassic orogenic magmatism in the Aeolian
1999. Chemical and isotopic variations in fumarolic dis- Islands, Southern Italy. Bull. Volcanol 50, 386–398.
charge and thermal waters at Vulcano Island (Aeolian Evans, W.C., Sorey, M.L., Kennedy, B.M., Stonestrom, D.A.,
Islands, Italy) during 1996: evidence of resumed volcanic Rogie, J.D., Shuster, D.L., 2001. High CO2 emissions
activity. J. Volcanol. Geotherm. Res. 88, 167–175. through porous media: transport mechanisms and implica-
Capasso, G., D’Alessandro, W., Favara, R., Inguaggiato, S., tions for flux measurement and fractionation. Chem. Geol
Parello, F., 2001. Interaction between the deep fluids and the 177, 15–29.
shallow groundwaters on Vulcano island (Italy). J. Volcanol. Falcone, G., 1994. Il flusso di CO2 come precursore dell’attività
Geotherm. Res. 108, 187–198. vulcanica. Thesis. Univ. Pisa.
Carapezza, M.L., 1994. Helium and CO2 soil gas concen- Francis, P., Burton, M.R., Oppenheimer, C., 1998. Remote
tration. In: Unrest at Vulcano. A collection of preliminary measurements of volcanic gas compositions by solar occul-
contributions for the IPG-WOVO Meeting (Guadalupe, tation spectroscopy. Nature 396, 567–570.
December 1993) CNR-GNV, 177–187. Gasparini, C., Iannaccone, G., Scandone, P., Scarpa, R., 1982.
Carapezza, M.L., 1996. Soil gas investigation in volcanic areas. Seismotectonics of the Calabrian arc. Tectonophysics 110,
In: Barberi, F., Casale, R., (Eds.), The mitigation of volcanic 59–78.
hazard, EUR 16804 EN, Brussels, 403–432. Gerlach, T.M., Doukas, M.P., McGee, K.A., Kessler, R., 2001.
Carapezza, M.L., Diliberto, I.S., 1993. Helium and CO2 soil Soil efflux and total emission rates of magmatic CO2 at the
88 M.L. Carapezza, D. Granieri / Applied Geochemistry 19 (2004) 73–88

Horseshoe Lake tree kill, Mammoth Mountain, California, Nuccio, P.M., Paonita, A., Sortino, F., 1999. Geochemical
1995–1999. Chem. Geol 177, 101–116. modeling of mixing between magmatic and hydrothermal
Giammanco, S., Gurrieri, S., Valenza, M., 1998. Anomalous gases: the case of Vulcano Island, Italy. Earth Planet. Sci.
soil CO2 degassing in relation to faults and eruptive fissures Lett. 167, 321–333.
on Mount Etna (Sicily, Italy). Bull. Volcanol 60, 252–259. Pareschi, M.T., Ranci, M., Valenza, M., Graziani, G., 1999.
Gioncada, A., 1997. L’attività eruttiva degli ultimi 50.000 anni The assessment of volcanic gas hazard by means of numer-
di Vulcano (Eolie): aspetti vulcanologici e magmatologici. ical models: an example from Vulcano Island (Sicily). Geo-
Doct. thesis, Univ. Pisa. phys. Res. Lett. 26, 10. 1405-1408.
Gurrieri, S., Valenza, M., 1988. Gas transport in natural por- Parkinson, K.J., 1981. An improved method for measuring soil
ous mediums: a method for measuring CO2 flows from the respiration in the field. J. Appl. Ecol 18, 221–228.
ground in volcanic and geothermal areas. Rend. Soc. It. Patacca, E., Sartori, R., Scandone, P., 1990. Tyrrhenian basin
Mineral. Petrol 43, 1151–1158. and apenninic arcs: kinematic relations since late tortonian
Hinkle, M.E., 1994. Environmental conditions affecting con- times. Mem. Soc. Geol. It 45, 425–451.
centrations of He, CO2, O2 and N2 in soil gases. Appl. Geo- Penman, H.L., 1940. Gas and vapour movement in the soil. I.
chem. 9, 53–63. The diffusion of vapours through porous solids. J. Agric. Sci.
Italiano, F., Nuccio, P. M., 1997. Variazioni del rapporto iso- 30, 437–462.
topico dell’elio nelle esalazioni fumaroliche di Vulcano. In: Pizzino, L., Galli, G., Mancini, C., Quattrocchi, F., Scarlato,
La Volpe, Dellino, Nuccio, Privitera, Sbrana (Eds.), Progetto P., 2002. Natural gas hazard (CO2, 222Rn) within a quiescent
Vulcano: Risultati dell’attività di ricerca 1993–1995. Felici volcanic region and its relations with tectonics: the case of
Editore, Pisa, 124–127. the Ciampino-Marino area, Alban Hills volcano, Italy. Nat-
Italiano, F., Pecoraino, G., Nuccio, P.M., 1998. Steam output ural Hazards 00, 1–35.
from fumaroles of an active volcano: tectonic and magmatic- Reimer, G.M., 1980. Use of soil-gas helium concentrations for
hydrothermal controls on the degassing system at Vulcano earthquakes prediction: limitations imposed by diurnal var-
(Aeolian arc). J. Geophys. Res. 103-B12, 29829–29842. iations. J. Geophys. Res. 85-B, 3107–3114.
Kanemasu, E.T., Powers, W.L., Sij, J.W., 1974. Field chamber Reiners, W.A., 1968. Carbon dioxide evolution from the floor
measurement of CO2 flux from soil surface. Soil Sci. 118- 4, of three Minnesota forests. Ecol 49, 471–483.
233–237. Rogie, J.D., Kerrick, D.M., Chiodini, G., Sorey, M.L., Virgili,
Keller, J., 1980. The island of Vulcano. Rend. Soc. It. Mineral. G., 1998. Continuous monitoring of diffuse CO2 degassing,
Petrol 36, 489–584. Horseshoe Lake, Mammoth Mountain, California. EOS
Klinkenberg, L. J., 1941. The permeability of porous media to Trans. Am. Gephys. Union 79, 942.
liquids and gases. In: Drilling and Production Practice. Sicardi, L., 1941. Il recente ciclo dell’attività fumarolica dell’i-
American Petroleum Institute, New York, 200-213. sola di Vulcano. Bull. Volcanol 7, 85–140.
Krumbein, W. C., Monk, G. D., 1951. Permeability as a func- Tedesco, D., 1995. Fluid geochemistry at Vulcano island: a
tion of the size parameters of unconsolidated sand. A.I.M.E. change in the volcano regime or continuous fluctuation in the
Trans., 151. mixing of different systems? J. Geophys. Res. 100, 4157–
Kucera, C.L., Kirkham, D.R., 1971. Soil respiration studies in 4167.
tallgrass praire in Missouri. Ecology 52, 912–915. Tedesco, D., Scarsi, P., 1999. Intensive gas sampling of noble
La Volpe, L., Dellino, P., Nuccio, P.M., Privitera, E., Sbrana, gases and carbon at Vulcano island. J. Geophys. Res. 104-
A. (Eds.), 1997. Progetto Vulcano: Risultati dell’attività di 55, 10499–10510.
ricerca 1993–1995. Felici Editore, Pisa, Italy. Thorstenson, D.C., Pollock, D.P., 1989a. Transport in unsatu-
Magro, G., 1997. La composizione isotopica dell’He e dell’ Ar rated porous media: the adequacy of Fick’s law. Rev. Geo-
nei prodotti di Vulcano: inclusioni fluide e gas fumarolici. In: phys 27, 61–78.
La Volpe, Dellino, Nuccio, Privitera, Sbrana (Eds.), Progetto Thorstenson, D.C., Pollock, D.P., 1989b. Gas transport in
Vulcano: Risultati dell’Attività di Ricerca 1993–1995. Felici unsaturated zones: multicomponent systems and the ade-
Editore, Pisa, pp. 128–132. quacy of Fick’s laws. Water Resour. Res. 25- 3, 477–507.
Malinconico, L.L., 1987. On the variation of SO2 emission Tonani, F., Miele, G., 1991. Methods for measuring flow of
from volcanoes. J. Volcanol. Geotherm. Res. 33, 231–237. carbon dioxide through soils in the volcanic setting. Internat.
Mason, E.A., Malinauskas, A.P., Evans III., R.B., 1967. Flow Conf. Active Volcanoes and Risk Mitigation. Napoli. 27
and diffusion of gases in porous media. J. Chem. Phys. 46, August–1 September 1991.
3199–3216. Walker, G.P.L., 1973. Explosive volcanic eruptions-a new
Mason, E. A., Malinauskas, A. P. 1983. Gas transport in por- classification scheme. Geol. Rundsch 62, 431–446.
ous media: the dusty gas model. Chem. Eng. Monog. 17, 194, Welhan, J.A., Poreda, R.J., Rison, W., Craig, H., 1988. Helium
Elsevier, New York. isotopes in geothermal and volcanic gases of the Western
Mercalli, G., Silvestri, O., 1891. Le eruzioni dell’isola di Vul- United States, II. Long Valley Caldera. J. Volcanol. Geo-
cano incominciate il 3 agosto 1988 e terminate il 22 marzo therm. Res. 34, 201–209.
1890. Relazione scientifica, Ann. Uff. Cent. Meteor. Geodin. Wentworth, C.K., 1922. A scale of grade and class terms for
10-4, 213. clastic sediments. J. Geol 30, 377–392.
Natale, G., Hernandez, P., Mori, T., Notsu, K., 2000. Pressure Witkamp, M., Frank, M.L., 1969. Evolution of CO2 from
gradient measurements in volcanic diffuse gas emanations. litter, humus and subsoil of a pine stand. Pedobiol 9,
Geophys. Res. Lett. 27–24, 3985–3987. 358–366.

View publication stats

You might also like