You are on page 1of 10

Journal of Volcanology and Geothermal Research 260 (2013) 52–61

Contents lists available at SciVerse ScienceDirect

Journal of Volcanology and Geothermal Research


journal homepage: www.elsevier.com/locate/jvolgeores

Carbon dioxide in the urban area of Naples: Contribution and effects of


the volcanic source
Domenico Granieri a,⁎, Antonio Costa b, Giovanni Macedonio c, Marina Bisson a, Giovanni Chiodini c
a
Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Pisa, Via della Faggiola 32, 56126 Pisa, Italy
b
Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Bologna, via Donato Creti 12, 40128 Bologna, Italy
c
Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Napoli, via Diocleziano 328, 80124 Napoli, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Naples is a large city located between two active volcanic areas: Campi Flegrei to the W and Vesuvius to the SE. The
Received 8 January 2013 Solfatara crater, inside the caldera of the Campi Flegrei and nearest to the western quarters of the city, is a prodi-
Accepted 6 May 2013 gious source of natural CO2 with a mean emission rate of 1067 ton/d, i.e. seven times higher than that of Vesuvius
Available online 26 May 2013
(151 ton/d).
This study shows that the area around the Solfatara and part of the urban area of Naples are affected by the
Keywords:
CO2 dispersion
volcanic plume when atmospheric circulatory patterns are dominated by the locally frequent sea breezes.
Solfatara Under these conditions the CO2 content in the air increases above normal values, reaching more than
Gas hazard 1000 ppm in proximity to the Solfatara crater to a few tens of ppm several kilometres from the source. Al-
Campi Flegrei though these values do not indicate a health risk even under the most unfavourable atmospheric conditions,
the volcanic source contributes to the total CO2 burden from all urban emissions and hence to overall air
quality. An emission rate ten times higher than the present one would lead to an air CO2 concentration in ex-
cess of recommended health protection thresholds.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction area. A large part of the city lies downwind of the CF volcanic area. The
most active volcanic site in the CF is the Solfatara crater; within just a
Human activity is concentrated in urban areas, where the massive few hundred square metres the crater hosts two high-temperature fu-
use of fossil fuels, mainly for domestic heating and transport, produces maroles (140–160 °C), a few low-temperature fumaroles (95.0–
a significant amount of anthropogenic carbon dioxide. The buildup of 97.5 °C) and one large degassing mud pool. The bottom of the crater
CO2 over an urban area has been called an “urban CO2 dome” by Idso and the external eastern slope of the cone are affected by widespread dif-
et al. (1998), by analogy with the more well-known “urban heat island” fuse CO2 degassing. Carbon dioxide has a molecular weight greater than
(Goward, 1981; Oke, 1982). Several studies have documented its that of air (44 g/mol vs 29 g/mol of air) and therefore tends to accumu-
presence in the urban environment of large-sized cities such as London, late in low-lying areas, forming potentially lethal invisible traps
UK (Derwent et al., 1995), Vancouver, Canada (Reid and Steyn, 1997), (Carapezza et al., 2003; Costa et al., 2008; Chiodini et al., 2010a). This is
Phoenix, USA (Idso et al., 1998), Karachi, Pakistan (Ghauri et al., 1994) not the case of the Solfatara because the discharged gas is warmer than
and Nottingham, UK (Berry and Colls, 1990). None of these studies the surrounding air, so that the greater molecular weight of CO2 is com-
were undertaken in cities located near active volcanoes, where the natu- pensated by the decrease in density due to its higher temperature (Costa
ral source can increase urban CO2 levels. In this study we describe the et al., 2005). Under such conditions the CO2 plume is dispersed mainly
dispersion of CO2 discharged from the Solfatara volcanic area near the under the effect of wind and atmospheric turbulence (so-called passive
city of Naples. Naples is the most densely-populated city in Italy and dispersion) and, depending on the wind direction, it can move towards
one of the most densely-populated metropolitan areas in Europe, with the densely populated quarters of the city or its suburbs.
a density of 8566 inh/km2 downtown and 2612 inh/km2 in the suburbs To investigate the dispersion of the CO2 cloud we applied the DISGAS
(ISTAT, 2001). The urban area lies between two volcanoes: its western code (an improved version of the model presented in Costa et al., 2005,
quarters are located inside the Campi Flegrei (CF) caldera, whereas the see the Appendix), coupled with a mass-consistent wind model
southeastern side of the city extends up to the slopes of Vesuvius. The (Douglas and Kessler, 1990). Simulations were performed for the
city centre is surrounded by fairly high hills on three sides such that most significant local-scale wind and hydrodynamic conditions of the
the dominant sea breeze from the SW is channelled over the urban atmosphere derived from the analysis of micrometeorological observa-
tions. Results agree satisfactorily with field data collected in the Solfata-
⁎ Corresponding author. Tel.: +39 0508311958; fax: +39 0508311942. ra area by two automated stations that continuously measure the CO2
E-mail address: granieri@pi.ingv.it (D. Granieri). content in air. In the following we discuss the spatial and temporal

0377-0273/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jvolgeores.2013.05.003
D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61 53

evolution of the gas plume for the mean and maximum CO2 output rates where the overbar indicates the mean and (′) a deviation from the
measured in the current quiescent period of the volcano. Lastly, we used mean of the instantaneous measurement. High-frequency acquisition
the model to define the potential hazard to human health by increasing allows us to resolve small-scale wind features and hydrodynamic con-
the flow rate to ten times the current one. ditions at the surface such as friction velocity (u*) and Monin–Obukhov
length (L). Friction velocity, an index of the “drag” which the Earth's
2. Carbon dioxide source surface exerts on atmospheric flows, is defined as follows:

2 2 2
An impressive amount of CO2 is released continuously at the u ¼ ðu′ w′ Þ þ ðv′ w′ Þ : ð4Þ
Solfatara crater through diffusion from the soil (Chiodini et al.,
2001, 2010b). From 1998 to 2008 twelve extensive field surveys The magnitude of u* is a function of the roughness and slope of the
were performed in order to monitor CO2 emission rates. Each survey site, as well as the wind speed.
covers roughly the same area of ~ 1 km2 centred over the Solfatara The Monin–Obukhov length relates the turbulent energy generated
crater and includes about 400 carbon dioxide soil diffusion measure- by the mechanical mixing of gas in the atmosphere with the buoyancy-
ments (Chiodini et al., 2010b). The average CO2 emission (ACE) of all induced turbulent energy:
the surveys is 1067 ton/d (Fig. 1) and the maximum CO2 emission
(MCE) is ~ 1800 ton/d (Chiodini et al., 2010b). For comparison, note 3
L ¼ −T v u =kg F b0 ð5Þ
that the average discharge of CO2 through diffuse degassing at Vesuvius
is seven times smaller than that at the Solfatara crater (i.e., 151 ton/d where k is the dimensionless von Karman constant, Tv is the virtual po-
(Frondini et al., 2004) vs 1067 ton/d). tential air temperature and Fb0 is the virtual potential temperature flux
at the surface due to air density differences determined by temperature
3. Micrometeorological measurements and experiment design and humidity fluctuations. L is zero for neutral stratifications (L values
at Solfatara of around zero typically occur during day–night transitions, under over-
cast skies or with strong winds), negative for unstable stratifications
The surface boundary layer (SBL), the lowest layer of the tropo- (small negative values indicate a highly unstable atmosphere, large
sphere, is almost always turbulent and is influenced by surface friction negative ones a slightly unstable atmosphere) and is positive for stable
due to vegetation and topography (Stull, 1988). Within the SBL, atmo- stratifications.
spheric trace gases are rapidly diffused to (or from) the surface by irreg- Our micrometeorological measurements were performed inside the
ular or random motions generated by wind shear and buoyancy forces Solfatara crater using a 3.3 m high tower (Fig. 2) equipped with a
(Baldocchi et al., 1988; Stull, 1988). Micrometeorology is based on the three-axis sonic anemometer (Campbell Scientific, Model CSAT3-3D),
high-frequency acquisition of these turbulent fluctuations generated one barometer (Campbell Scientific, Model CS106) and one fine-wire
by small short-lived eddies. The high-frequency acquisition of data thermocouple (Campbell Scientific, Model FW05). An ancillary
over a time frame provides a means for splitting variables such as open-path gas analyser was used to derive H2O and CO2 fluxes (Licor,
wind into a mean part and a turbulent part. For example, the turbulent Model LI-7500).
component of wind in the longitudinal direction (u′) is: The micrometeorological survey lasted 2 years, from June 2004 to
June 2006, and valid data were collected approximately every three

u′ ¼ u− u ð1Þ days (207 days over a two-year period). The three components of
wind speed, atmospheric pressure and air temperature were sampled
where u is the instantaneous value of wind along the x axis and ū is the at a 10-Hz rate and averaged over 10-min intervals. We calculated
mean longitudinal velocity over the period of time chosen to average more than 12,700 10-min averages, corresponding to about 76 million
the raw measurements. We can likewise divide the latitudinal (v) and observational readings.
vertical (w) components of wind into their mean and turbulent parts: The micrometeorological survey allowed us to resolve in great de-
tail the dominant wind conditions and hydrodynamic features of the
− atmospheric surface layer at the Solfatara crater. Acquired data re-
v′ ¼ v− v ð2Þ
vealed the following features: 1) onshore breezes (westerly and

southwesterly) are present more than 50% of the year, mainly during
w′ ¼ w− w ð3Þ spring and summer (Fig. 3); 2) typical spring/summer meteorological

Fig. 1. Soil CO2 flux measured in the Solfatara area during 12 extensive field surveys in the 1998–2008 period (Chiodini et al., 2010b). Vertical bars represent the variation in each
survey value (95% confidence interval). The horizontal dashed line is the average CO2 emission (ACE, 1067 ton/d) of all the surveys, whose variation is represented by the grey area
(790–1600 ton/d, 95% confidence interval). MCE is the maximum CO2 emission rate (MCE, 1800 ton/d in the range 1260–2900 ton/d) measured in the ten-year period. Vertical
dashed lines indicate the days used in the procedure of the code validation (see Section 4.1).
54 D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61

Fig. 2. Micrometeorological station at Solfatara. The tower is 3.3 m high. Inset: red dot indicates the location of the micrometeorological station inside the crater facing the mud pool
(Fangaia). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

conditions are characterised by a weak synoptic circulation (mainly


in summertime due to the presence of the Azores anticyclone over
Europe) and a well developed breeze regime controlled by the daily
thermally-induced sea–land air circulation; 3) diurnal breeze cycles
are characterised by very weak air circulation during the night (land
breeze) and an increase in wind intensity in the day (sea breeze). In
summer/spring, the sea breeze cycle starts at around 9:00–10:00
Local Time, LT (about 3–4 h after sunrise) and ends in the evening
at about 21:00 LT, with the maximum intensity during the warmer
hours of the day (about 3–3.5 m/s at 15:00 LT, Fig. 4a). Diurnal sea
breezes initially blow from ~270° (W) and then veer to ~190° (SSW)
in broad daylight (Fig. 4b). Wind measurements suggest that air circu-
lation is quite weak at night (mean speed of 0.45 m/s, Fig. 4a), so that
the expected land breeze (winds from the NE and NNE) is often
overpowered by the synoptic SW wind; 4) in summer/spring, the
values of the Monin–Obukhov length (L) at Solfatara reveal that atmo-
spheric conditions range from highly unstable (small negative L from Fig. 3. Wind rose plot for the Solfatara crater (data refer to 207 days in a two-year period).
9:00 to 18:00 LT) to weakly unstable (from 19:00 to 22:00 LT), The scale of the axes is from 0 to 15% (frequency).
D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61 55

are added to the first step field, and an objective analysis scheme is
used to produce a new gridded field (U, V, W). The scheme is designed
so that the observations are used to define the wind field within a
user-specified radius of influence while the first step (U, V, W) field is
used in sub-regions in which observations are unavailable. Finally, a
divergence-minimization procedure is iteratively applied until velocity
divergence is smaller than an arbitrarily user defined small number. The
final product of DWM is an approximately null-divergence wind field
consistent with the observations. Dispersion is assumed to be governed
by the wind and atmospheric turbulence, since the passive plume con-
dition (Mohan et al., 1995):

   
MCO2 44 15 −3
Δρ ¼ −1  C ≈ −1  C ¼  C b 0:001 kg m ð1Þ
Mair 29 29

is legitimate for diluted and warm gases (up to 162 °C) of Solfatara at
few centimetres above the ground level (Δρ is the density contrast be-
tween gas plume and air, M CO2 and Mair are the molecular weights of
CO2 and air and C is the CO2 concentration, expressed in kgm−3).
DISGAS inputs include topography, roughness length, meteorological
data (wind velocity and direction, air temperature), atmospheric stabil-
ity information (Monin–Obukhov length, friction velocity), and gas flow
rates from the source. The model yields CO2 concentrations expressed
as values in excess of background CO2 levels in the air at heights select-
Fig. 4. Wind in the Solfatara crater during summer 2004. (a) wind speed; (b) wind direction
with respect to the north. Each dot represents a 10-min interval.
ed by the user. For all simulations, the computational domain extended
over an area of 27 × 27 km2 that was discretised using a square grid
with 180 × 180 cells in the x and y directions. Vertical grid steps ranged
indicating that during the daytime the turbulence induced by buoyancy from Δz = 10 m in the uppermost level (100 m) to Δz = 0.5 m in the
exceeds mechanical mixing (Fig. 5). During nighttime the atmosphere lowermost ground level so as to better detect near-surface gradients. A
is stable (positive L from 23:00 to 8:00), implying low turbulence preliminary test demonstrated the results of the simulations do not vary
effects. substantially when reproducing the wind field over a thicker domain
(e.g., assigning the altitude of 500 m at the top of the domain). The tran-
4. The numerical model: validation and applications sition sea–land is associated with the change of the roughness height
from 0.01 m for the sea (relative to the marine surface in coastal
The numerical simulations presented in this study were performed areas) to 0.5 m for land (typical for rolling or moderate relief covered
using the DISGAS code (Costa et al., 2005), an Eulerian model based by shrubs, rare tree and thin buildings).
on advection–diffusion equations, coupled with the mass-consistent di-
agnostic wind model (DWM, Douglas and Kessler, 1990), which can de- 4.1. Model validation
scribe wind fields over complex terrains such as those surrounding
Solfatara. The DWM generates wind components (U, V, W) in a In order to validate the model, simulated concentrations were
terrain-following coordinate system. The model needs topographic compared with CO2 concentrations in the air measured directly inside
data, average wind on the computational domain, atmospheric stability the Solfatara crater. We chose to simulate the daily cycles (1:00–
information within the scale of the domain (i.e. the temperature gradi- 24:00 LT) recorded by the micrometeorological station on 29 June
ent dT/dz), and, when available, local surface and upper-air wind mea- 2004 and 24 December 2004, which are considered to be representa-
surements. In a first step the domain-mean wind is adjusted for the tive of summertime and wintertime cycles respectively. Meteorolog-
kinematic effects of terrain (lifting and acceleration of the airflow over ical data and parameters relative to the hydrodynamic conditions of
terrain obstacles), thermodynamically generated slope flows, and the atmosphere employed in the validation of numerical simulations
blocking effects. In a second step, wind observations, when available, are reported in Table 1.
The carbon dioxide flow rate was assumed to be the mean be-
tween the ACE (1067 ton/d) and the August 2004 survey value
(684 ton/d) in the summer (Fig. 1) and the March 2005 survey
value (572 ton/d) in the winter. Simulated values were compared
with CO2 concentrations measured at the station #1 (3.3 m above
soil level) and station #2 (0.5 m a.s.l.), located in two different
areas of the Solfatara crater and with an acquisition periodicity of
1 h and 2 h respectively. There is less correlation between modelled
and observed CO2 concentrations at the beginning of the series
(Fig. 6), when the day/night transition produces abrupt changes in
wind direction and intensity. In the middle and at the end of the con-
sidered period the two series are remarkably similar, both in order of
magnitude and in the trend over time. As a whole, the correlation co-
efficients (Table 2) suggest that the correlation is higher for station
Fig. 5. Variations in the Monin–Obukhov length (L) at Solfatara in summer 2004 (about
#1, and lower but significant at the 95% confidence level for station
2400 10-min intervals); grey areas define the theoretical range of L over a daily cycle in #2. Correlation is better in the summer, when the wind regime is
fair weather conditions over land (Stull, 1988). dominated by breezes, than in the winter.
56 D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61

Table 1 Table 2
Input data for the numerical simulation. Correlation coefficient (R) between measured and simulated CO2 concentrations.

Hour Wind speed Wind u*b Lc Td Season Station #1 Station #2


(m/s) directiona (m/s) (m) (°C) (3.3 m) a (0.5 m) a
(°)
Summer +0.86 +0.55
Summer/ Summer/ Summer/ Summer/ Summer/ Winter +0.68 +0.29
winter winter winter winter winter a
The value in brackets is the height at which CO2 concentrations were measured.
1 0.36/0.35 213/246 0.06/0.06 −4.36/2.12 18.6/3.8
2 0.36/0.22 215/273 0.04/0.10 9.13/3.93 18.4/3.8
3 0.40/0.25 216/248 0.03/0.11 1.76/10.73 18.3/4.1
4.2. Model applications
4 0.37/0.36 220/234 0.05/0.13 4.12/−3.30 18.1/4.5
5 0.35/0.24 218/150 0.04/0.15 3.78/7.76 17.9/4.2
6 0.27/0.32 189/188 0.05/0.15 −1.88/0.54 17.8/3.9 Having described the evolution of the wind field through the prog-
7 0.27/0.08 179/260 0.05/0.08 7.87/17.89 18.3/4.9 nostic DWM, we now report the results of two scenarios during a typical
8 0.60/0.11 195/297 0.11/0.11 4.56/−1.94 18.9/5.9 spring/summer period accounting for two different CO2 emission rates.
9 1.09/0.53 232/38 0.19/0.15 −11.32/0.41 21.8/7.6
In the first scenario the carbon dioxide flow rate is assumed to be
10 1.37/0.58 278/53 0.11/0.11 −1.05/−11.54 25.9/9.4
11 1.83/1.43 273/48 0.18/0.14 −3.68/−8.20 28.2/10.5 1067 ton/d (ACE), i.e. the average of 12 surveys in the 1998–2008 peri-
12 1.72/0.75 273/46 0.19/0.23 −3.17/−46.15 29.0/11.7 od (Chiodini et al., 2010b). In terms of mean value and range of varia-
13 1.69/0.64 244/305 0.25/0.21 −7.9/−57.39 28.9/12.2 tion, the most representative survey is the October 2005 survey
14 2.42/1.30 228/237 0.26/0.19 −6.91/−47.33 28.7/12.6
(Fig. 1). The second scenario assesses the consequences of a CO2 feeding
15 2.68/0.90 217/214 0.26/0.16 −8.36/−142.62 29.0/12.3
16 1.75/1.13 256/192 0.19/0.20 −4.05/27.80 29.1/11.8
rate equal to the maximum value actually measured in the ten-year pe-
17 1.90/0.78 277/178 0.16/0.12 −1.94/45.13 29.4/11.7 riod (MCE ~1800 ton/d in July 2000, Fig. 1). Fig. 7 reports the spatial dis-
18 2.46/1.09 263/174 0.20/0.19 −3.14/84.64 29.8/11.6 tribution of the CO2 source in the Solfatara area for the ACE value
19 2.21/1.18 211/166 0.25/0.14 −20.95/38.74 29.2/11.7 (Fig. 7a) and the MCE value (Fig. 7b). CO2 flux values (about 400 CO2
20 1.65/1.16 217/185 0.22/0.19 −18.85/64.89 27.8/11.7
soil flux measurements for each survey) were mapped using the se-
21 0.79/0.93 206/193 0.18/0.28 −36.98/128.05 25.1/11.7
22 1.15/1.03 175/193 0.14/0.27 −191.18/184.48 22.6/11.6 quential Gaussian simulation approach (Cardellini et al., 2003;
23 0.93/0.70 172/188 0.13/0.27 29.09/149.43 21.3/11.9 Deutsch and Journel, 1998). The resulting maps highlight strong CO2
24 0.45/0.24 181/52 0.05/0.27 6.43/189.24 20.6/12.2 degassing in the Solfatara crater and in the neighbouring areas;
a
Wind direction is the direction from which the wind blows, reported in degrees degassing is mainly controlled by the NW–SE trending fractures
from the north. (Chiodini et al., 2001; Granieri et al., 2010). Meteorological data and pa-
b
u* is the friction velocity. rameters relative to the hydrodynamic conditions of the atmosphere
c
L is the Monin–Obukhov length.
d are reported in Table 1 (summer column). Simulated CO2 concentrations
T is the air temperature measured 3.3 m above ground level.
were calculated at a height of 1.5 m, this being the typical breathing
height of humans.
Despite the limitations of using the Eulerian approach to reproduce
dispersion in the proximity of the gas source (Boybeyi and Raman, 5. Results and discussion
1995), the numerical simulations we performed provide a satisfactory
representation of the natural system. Furthermore, the average CO2 The wind field was reproduced using the diagnostic wind model
concentration for 28 gas plume samples collected inside the Solfatara (Douglas and Kessler, 1990); data from two meteorological stations lo-
on 7 July 2008 is 1606 (±947) ppm (Chiodini et al., 2011). This value cated in different sites of the computational domain were used as
is similar to the simulated value inside the crater during a typical sum- input. The results of the diagnostic wind model indicate that during
mer day (about 1100 ppm) at a height of 0.5 m to 3.3 m. the night (Fig 8a) the wind regime in the lower layers is characterised

Fig. 6. CO2 concentrations measured at stations #1 (a–b) and #2 (c–d) compared with DISGAS code simulation values. The stations are equipped with IR non-dispersive gas ana-
lyzers (Licor LI-7500 at station #1 and Dräger Polytron IR-CO2 at station #2 with 1% and 3% accuracy respectively). The local background CO2 concentration (380 ppm, Chiodini et
al., 2011) was added to the simulated values.
D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61 57

Fig. 7. Maps of the soil CO2 flux measured in a) October 2005 and b) July 2000. CO2 degassing from soil is controlled by the first-order NW–SE fractures affecting the Solfatara crater
and the surrounding areas. Pisciarelli is an area of intense fumarolic activity on the external NE flank of the crater.

by a dominant circulation from the W, with weak winds (mean Assuming an ACE rate, the CO2 dispersion model suggests that car-
velocity = 0.27 m/s) that flow around the main morphological obsta- bon dioxide reaches higher concentrations during the night, when
cles in the domain. During the day, the winds pick up (mean winds are weak and the atmosphere is very stable. In this situation
velocity = 2.33 m/s), shift from the W to the SW and penetrate deep the gas accumulates and CO2 concentrations at breathing level in the
inland (Fig 8b). These results agree with the considerations of Barone source area reach up to 1400 ppm (Fig. 9) in excess of local background
et al. (2000) on wind circulation patterns over the city of Naples during values (380 ppm, Chiodini et al., 2011). Values decrease to a few tens of
summer: “During the early morning a shift in the wind direction (from ppm 2–3 km from the source. Daytime atmospheric conditions are un-
westerly to southerly) is observed at the Naples monitoring station, stable and wind circulation is more intense, so that the gas is rapidly di-
where the wind direction changes by about 90°; the transition between luted in proximity to the source (Solfatara), where maximum values of
the night and day circulation is due to the sea breeze, which develops 300–350 ppm in excess are estimated. The gas plume stretches more
in the morning hours and lasts until late afternoon.” than 3 km from the source in a SW–NE direction (Fig. 9); CO2

Fig. 8. Simulated wind fields in the first layer of the computational grid (2 m above the ground) at a) 3:00 and b) 13:00 LT on 29 June 2004. The reference vector is the wind velocity,
solid (blue) dots indicate the position of the meteorological stations. The background layer is a high-resolution orthophoto from the 2000 Italy flight (Bisson et al., 2007).
58 D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61

Fig. 9. Typical nighttime (3:00, LT) and daytime (13:00, LT) CO2 concentrations 1.5 m above the ground. Concentrations are expressed as values in excess of background CO2 con-
centrations in air. The urban area of Naples, affected by volcanic CO2 dispersion, has the highest population density in Italy (8566 inh/km2 in the city vs 189 inh/km2 in Italy, ISTAT,
2001). The shaded relief with cell size of 10 m is shown in the background (Tarquini et al., 2007).

concentrations in the northwesterly quarters of Naples reach up to a of CO2 in air ranging from 200 ppm to 1100 ppm were estimated in
few tens of ppm. The map of CO2 concentrations averaged over a com- the populated area around Solfatara (Fig. 10b).
plete 24-h cycle shows a roughly circular pattern, with maximum Considering the MCE rate measured in the last ten years
values of about 500 ppm above background levels inside the Solfatara (~1800 ton/d in July 2000, Fig. 1), the model predicts that CO2 concen-
crater (Fig. 10a). The plume is slightly elongated toward the NE, proba- trations at a height of 1.5 m increase by 100% compared to the previous
bly due to the effect of the SW–NE marine breezes during daytime case, reaching maximum values of about 1300 ppm in an area near to
hours. This tongue of gas covers the northwestern area of Naples, in the gas source (Fig. 11a) and values of up to 300–400 ppm in the area
which the ‘residential areas’ of Agnano (10,000 residents) and Bagnoli NE of the Solfatara crater. In this area the model calculates values rang-
(28,600 residents) are found. At the lowest wind speeds and under con- ing from 300 ppm to 2400 ppm during nighttime, when gas accumula-
ditions of atmospheric stability (2:00–7:00, LT), excess concentrations tion is favoured by stagnant atmospheric conditions (Fig. 11b).

Fig. 10. Air CO2 concentration maps assuming an ACE rate (1067 ton/d). a) Map averaged over a complete diurnal cycle (24 h) and b) nocturnal map (2:00–7:00, LT).
D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61 59

Fig. 11. Maps of CO2 concentrations in air assuming the maximum CO2 emission rate measured in the 1998–2008 period (MCE rate ~1800 ton/d). a) Values averaged over a whole
diurnal cycle (24 h) and b) nighttime values (2:00–7:00, LT).

Even considering the maximum CO2 emission rate measured in (Berry and Colls, 1990; Ghauri et al., 1994; Derwent et al., 1995; Reid
the last ten years and the most unfavourable atmospheric conditions and Steyn, 1997; Idso et al., 1998). The phenomenon has been called
(i.e. onshore wind and atmospheric stability), results indicate that an urban CO2 dome by Idso et al. (1998), by analogy with the more
there is no serious risk to populations at the present CO2 emission well-known urban heat island (Goward, 1981; Oke, 1982). Considering
rates, since concentrations are below the “hazardous” threshold of the scattered disposition of the CO2 sources (primarily domestic
5000 ppm of CO2 (Table 3). However, it is important to consider heating and vehicular traffic), the CO2 dome generally mantles the
that this threshold refers to a selected population (working-age peo- whole city and is responsible for a general moderate increase (from
ple in good health) exposed to the gas for a limited time period (8 h). several ppm to a few tens of ppm) in CO2. Although a thorough and
The carbon dioxide originating from the active volcanic area of Solfa- well-documented survey of CO2 concentrations in air has not been com-
tara adds to the huge amount of CO2 produced by anthropogenic activ- pleted in the urban area of Naples, the “dome” phenomenon can be rea-
ity in the city, primary by the localised burning of fossil fuels. Naples is a sonably assumed to exist considering the high degree of urbanisation
densely-populated metropolitan area that hosts an industrial area and and the intense traffic in the city.
one of the most highly trafficked seaports (passengers and containers) This study focussed on the dispersion of volcanic CO2 plume emis-
in the Mediterranean (data from Naples Port Authority, http://www. sions at Solfatara. Although they do not constitute a human health haz-
porto.napoli.it/en/statistiche/stat1.php, accessed Jan. 10, 2012). Follow- ard at the current emission rate, volcanic CO2 emissions contribute to
ing official inventories submitted by Italy in conformity with the guide- build up the load of carbon dioxide in the urban environment, particu-
lines of the Intergovernmental Panel on Climate Change (IPCC, 2007), larly during the summer months, when high concentrations of primary
the CO2 emitted by the metropolitan (city plus province) area of Naples pollutants (principally NOx) and photochemical smog stagnate near the
(Table 4) was approximately constant during the 2000–2005 period ground (Barone et al., 2000). At present, volcanic CO2 represents on
(22,062 ton/d equivalent to 8.05 Mton/y in 2000 and 22,105 ton/d average 4.8% of the entire production of CO2 in the metropolitan area
equivalent to 8.07 Mton/y in 2005). of Naples (about 22,000 ton/d in the 2000–2005 period), with a maxi-
The emission of anthropogenic CO2 determines a buildup of CO2 mum contribution of 8.1% in July 2000 and a minimum contribution of
over the urban environment. Several studies have documented this 2.6% in March 2005. In the SNAP97 classification this natural emission
phenomenon in some large-sized cities throughout the world where, falls in the “Source 11”, which includes volcanic activity. Source 11 has
depending on human activity cycles as well as trace gas mixing a negative value (namely CO2 consumption) in all Italian provinces ex-
dynamics in the atmospheric boundary layer, CO2 concentrations are cept Catania (Sicily), where it has been ascribed to the positive contri-
persistently greater than background levels in rural and remote sites bution of volcanic CO2 from Etna (ISPRA, 2007). Official inventories

Table 3
Threshold values for CO2 concentrations in the air (from Costa et al., 2008, modified).

CO2 thresholds for human health Principal effect

5000 ppm (0.5%) TWA (time weighted average — 8 h/day for 5 days per week) a Slight increase in breathing rate
15,000 ppm (1.5%) (10 min max exposure time in working area) b Breathing deeper and more frequent
30,000 ppm (3%) STEL (short term exposure limit — 15 min max time exposure Breathing increases to twice normal rate, weak narcotic effect, headache for
four times a day) a long time exposure
100,000 ppm (10%) c Respiratory distress with loss of consciousness in 10–15 min
120,000–150,000 ppm (12–15%) c,d Lethal concentration, exposure to levels above this is intolerable
a
National Institute of Occupational Safety and Health (NIOSH), 1997.
b
International Volcanic Health Hazard Network (IVHHN), 2005.
c
Le Guern et al. (1982).
d
Baxter et al. (1999).
60 D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61

Table 4 for the scaled concentration C may be re-written in a generalised form as


Carbon dioxide emitted by the metropolitan area of Naples (ISPRA, 2007). (Toon et al., 1988; Jacobson et al., 1996; Park and Kim, 1999):
Sourcea 2000 2005    
(ton/y) (ton/y) ∂C ∂C ∂C ∂C ∂ ∂C ∂ ∂C
þU þV þW ¼ Kh þ Kh
01 Combustion in energy and transformation industries 698,698 881,989 ∂t ∂x ∂y ∂z ∂t ∂x ∂y ∂y
02 Non-industrial combustion plants 1,245,025 1,515,737 ∂ ∂C
þ Kz þ Q ð1Þ
03 Combustion in manufacturing industry 617,276 308,836 ∂z ∂z
04 Production processes 24,909 22,707
06 Solvent and other product use 42,739 45,429
where t is time, (U, V, W) are the scaled wind speeds, Kh and Kz are the
07 Road transport 4,393,564 4,400,325
08 Other mobile sources and machinery 1,292,317 1,215,360 diagonal scaled diffusion coefficients and Q* the source term in the gen-
09 Waste treatment and disposal 13,133 19,528 eralised coordinate. If we consider the simple transformation changing
11 Other sources and sinks (nature) −275,051 −341,466 only the vertical coordinate as: z* = z − h(x,y), the Jacobian is equal to
Total 8,052,612 8,068,445 ∂z
the unit (J ¼ ∂z ¼ 1). Concerning the turbulent diffusion tensor, we ne-
a 
Source nomenclature according to SNAP97 (Selected Nomenclature for sources of Air
glect the off-diagonal horizontal and vertical turbulent diffusion compo-
Pollution). SNAP97 categories were developed as part of the CORINAIR (CORe INventory
of AIR emissions) project (EMEP/CORINAIR, 2006) for distinguishing emission source sec- nents (Jacobson et al., 1996). For a more general treatment that include
tors, sub-sectors and activities. SNAP97 is fully consistent with the IPCC (Intergovernmental the off-diagonal diffusion components see Byun and Chin (1999).
Panel on Climate Change) nomenclature (IPCC, 2007). Emissions are categorised under The advective terms are discretised according to the second-order
eleven sources (or macro-sectors), each source including sub-sectors and activities. Sources
Lax–Wendroff scheme (e.g., Lax and Wendroff, 1960; Ewing and Wang,
05 (Extraction and distribution of fossil fuels and geothermal energy) and 10 (Agriculture)
do not contribute to the total CO2 budget in the area: they are therefore not present in the
2001). For instance, the advective term in the x-direction we have:
inventory. Source 11 mainly includes the effects of managed and unmanaged forests, with  
positive values for CO2 emission and negative ones for CO2 consumption by photosynthesis ∂C n Cr  n n

−Δt U ¼− C iþ1;j;k −C i−1;j;k
in green plants. In this category the specific activity “Volcanoes” would include “emissions ∂x i;j;k 2
Cr  n 
from eruptive and non-eruptive volcanoes”. 2
n n
þ C iþ1;j;k −2C i;j;k þ C i−1;j;k ð2Þ
2
submitted by Italy in the years 2000 and 2005 did not take into account
Δt
for the CO2 originating from the active volcanic area of Solfatara. where Cr ¼ U Δx denotes the Courant number (Δx is the grid spacing
The exposure to Solfatara CO2 emissions would not be negligible if and Δt the computational time step), n is an index for time
fluxes were to increase. CO2 concentrations should not consistently ex- discretisation, (i, j, k) for the spatial grid. In order to reduce the numer-
ceed 15,000 ppm (1.5%), a threshold which our simulations suggest ical over- and under-shooting that commonly affects high-order
could happen in some locations if emissions increased by at least schemes near discontinuities one can use slope limiter methods. In
ten-fold (up to ~10,000 ton/d). A recent study, based on a dendrochrono- fact, these methods preserve the monotonicity of the solution while
logical methodology, showed that the last (1982–1984) bradyseismic the accuracy remains higher than the first order upwind methods
crisis at Campi Flegrei was probably accompanied by a higher-than- (Sweeby, 1984; Ewing and Wang, 2001; Wang and Hutter, 2001). For
the-present release of deep 14C-depleted CO2 into the atmosphere instance, applying the minmod slope-limiter method and considering
(Mostacci et al., 2009). Enhanced release lasted for several years. It is the general case with non-uniform velocity, the Eq. (2) becomes:
thus likely that dangerous threshold levels have been sporadically
 
attained in the past around the Solfatara and in the inhabited areas NE ∂C n Δt  n n n n

−Δt U ¼− U C −U i−1;j;k C i−1;j;k
of the crater (western sector of Naples), where gases are preferentially ∂x i;j;k Δx iþ1;j;k iþ1;j;k
Δx h n  n

n n

n

n
i
transported by the dominant marine breezes during spring and − Cri;j;k 1−Cr i;j;k σ i;j;k −Cri−1;j;k 1−Cr i−1;j;k σ i−1;j;k
2
summer.
ð3Þ
6. Conclusions n Δt n
where Cri;j;k ¼ U i;j;k Δx , and (Ewing and Wang, 2001):
In order to investigate the spatial and temporal evolution of the volca- ( n )
n C iþ1;j;k −C ni;j;k C ni;j;k −C ni−1;j;k
nic CO2 plume emitted by Solfatara, several simulations were performed σ i;j;k ¼ min mod ; ð4Þ
Δx Δx
using a numerical code. Simulations were performed for the most signif-
icant local wind and atmospheric conditions and were validated by mea- .
surements. Even considering the maximum CO2 emission rate measured with min modða; bÞ ¼ 1 ½ sgnðaÞ þ sgnðbÞ minðjaj; jbjÞ. The introduc-
2
in the last ten years and the more unfavourable atmospheric conditions tion of alternative limiter methods is straightforward.The diffusion terms
(i.e. onshore wind and atmospheric stability), simulated values do not ex- are evaluated using a central difference scheme for the general case with a
ceed the recommended threshold values. Nevertheless, volcanic CO2 adds not uniform turbulent diffusivity tensor, e.g. in the x− direction:
to the amount of CO2 produced by anthropogenic activity in Naples and
  "
contributes to the deterioration of urban air quality, particularly during ∂ ∂C 1  i;j;k i−1;j;k
 
iþ1;j;k i;j;k i−1;j;k

the summer months when the city of Naples is often affected by Kh ¼ 2 Kh þ Kh C i−1;j;k − K h þ 2K h þ K h C i;j;k
∂x ∂x Δx
#
photosmog episodes or high concentrations of primary pollutants. The  
iþ1;j;k i;j;k
magnitude of this very local source of CO2 indicates that it may be impor- þ Kh þ K h C iþ1;j;k : ð5Þ
tant in the regional budget of CO2 emissions and should be included in the
official inventory submitted by the Italian authorities in the framework of
The stability of the numerical scheme is ensured by using a time
actions to quantify greenhouse gas emissions.
step Δt lower than the critical. As established by Hindmarsh et al.
(1984), an explicit scheme for the multidimensional advection diffu-
Appendix
sion equation is numerically stable if the following condition is veri-
fied:
Model DISGAS: numerical solution method

1
Considering a null-divergence wind field, in a terrain following coordi- Δt ≤ : ð6Þ
nate system (Douglas and Kessler, 1990), the mass conservation equation
U
Δx þ Δx
U
þ Δx
U Kh
þ 2 Δx2 þ
Kh
Δy2
þ ΔzK z 2

D. Granieri et al. / Journal of Volcanology and Geothermal Research 260 (2013) 52–61 61

References Ghauri, B., Salam, M., Mirza, M.I., 1994. An assessment of air quality in Karachi, Pakistan.
Environmental Monitoring and Assessment 32, 37–45.
Baldocchi, D.D., Hicks, B.B., Meyers, T.P., 1988. Measuring biosphere–atmosphere ex- Goward, S.N., 1981. Thermal behaviour of urban landscapes and the urban heat island.
changes of biologically related gases with micrometeorological methods. Ecology Physical Geography 2, 19–33.
69, 1331–1340. Granieri, D., Avino, R., Chiodini, G., 2010. Carbon dioxide diffuse emission from the soil:
Barone, G., D'Ambra, P., di Serafino, D., Giunta, G., Murli, A., Riccio, A., 2000. Application ten years of observations at Vesuvio and Campi Flegrei (Pozzuoli), and linkages
of a parallel photochemical air quality model to the Campanian region (southern with volcanic activity. Bulletin of Volcanology 72, 103–118.
Italy). Environmental Modelling & Software 15, 503–511. Hindmarsh, A., Gresho, P., Griffiths, D., 1984. The stability of explicit Euler integration for
Baxter, P.J., Baubron, J.-C., Coutinho, R., 1999. Health hazards and disaster potential of certain finite difference approximations of the multidimensional advection–diffusion
ground gas emissions at Furnas volcano, Sao Miguel, Azores. Journal of Volcanology equation. International Journal for Numerical Methods in Fluids 4, 853–897.
and Geothermal Research 92 (1–2), 95–106. Idso, C.D., Idso, S.B., Balling Jr., R.C., 1998. The urban CO2 dome of Phoenix, Arizona.
Berry, R.D., Colls, J.J., 1990. Atmospheric carbon dioxide and sulphur dioxide on an Physical Geography 19, 95–108.
urban/rural transect — I. Continuous measurements at the transect ends. Atmo- Intergovermental Panel and on Climate Change (IPCC), 2007. Climate change 2007: the
spheric Environment 24A, 2681–2688. physical science basis. In: Solomon, S., et al. (Ed.), Working Group I Contribution to
Bisson, M., Fornaciai, A., Mazzarini, F., 2007. SITOGEO: a geographic database used for the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
GIS applications. Il Nuovo Cimento C 30C (3). http://dx.doi.org/10.1393/ncc/ Cambridge, United Kingdom and New York, NY, USA.
i2007-10243-9. International Volcanic Health Hazard Network (IVHHN), 2005. Gas and Aerosol Guidelines:
Boybeyi, Z., Raman, S., 1995. Simulation of elevated long-range plume transport using a Carbon Dioxide (CO2). http://www.ivhhn.org/index.php?option=com_content&
mesoscale meteorological model. Atmospheric Environment 29, 2099–2111. view=article&id=84 (accessed Jan. 10, 2012).
Byun, D., Chin, J., 1999. Science algorithms of the EPA Model-3 Community Multiscale Italian Istituto Superiore per la Ricerca e la Protezione dell'Ambiente (ISPRA), 2007. Data
Air Quality (CMAQ) modeling system. In: Byun, D.W., Chin, J.K.S. (Eds.), Volume III, downloaded from http://www.sinanet.isprambiente.it/it/inventaria/disaggregazione_
EPA-600/R-99-030. Washington, DC. prov2005/Emissioni%20provinciali/view (accessed Jan. 10, 2012).
Carapezza, M.L., Badalamenti, B., Cavarra, L., Scalzo, A., 2003. Gas hazard assessment in Italian Institute of Statistics (ISTAT), 2001. XIV Italian Census of Population. http://dawinci.
a densely inhabited area of Colli Albani Volcano (Cava dei Selci, Roma). Journal of istat.it/MD/dawinciMD.jsp?a1=m0GG0c0I0&a2=mG0Y8048f8&n=1UH90009OG0&
Volcanology and Geothermal Research 23, 81–94. v=1UH0DH09OG00000 (accessed Jan. 10, 2012).
Chiodini, G., Frondini, F., Cardellini, C., Granieri, D., Marini, L., Ventura, G., 2001. CO2 Jacobson, M., Lu, R., Turco, R., Toon, O., 1996. Development and application of a new air
degassing and energy release at Solfatara Volcano, Campi Flegrei, Italy. Journal of pollution modeling system. Part I: gas-phase simulations. Atmospheric Environ-
Geophysical Research 106 (B8), 16213–16221. ment 30, 1939–1963.
Chiodini, G., Granieri, D., Avino, R., Caliro, S., Costa, A., Minopoli, C., Vilardo, G., 2010a. Lax, P.D., Wendroff, B., 1960. Systems of conservative laws. Communications on Pure
Non-volcanic CO2 Earth degassing: case of Mefite d'Ansanto (southern Apennines), and Applied Mathematics 13, 217–237.
Italy. Geophysical Research Letters 37 (L11303). http://dx.doi.org/10.1029/ Le Guern, F., Tazieff, H., Faivre Pierret, R., 1982. An example of health hazard: people
2010GL042858. killed by gas during a phreatic eruption, Dieng Plateau (Java, Indonesia), February
Chiodini, G., Caliro, S., Cardellini, C., Granieri, D., Avino, R., Baldini, A., Donnini, M., 20th, 1979. Bulletin of Volcanology 45, 153–156.
Minopoli, C., 2010b. Long term variations of the Campi Flegrei (Italy) volcanic sys- Mohan, M., Panwar, T., Singh, M., 1995. Development of dense gas dispersion model for
tem as revealed by the monitoring of hydrothermal activity. Journal of Geophysical emergency preparedness. Atmospheric Environment 29, 2075–2087.
Research 115 (B03205), 531–542. Mostacci, D., Chiodini, G., Berti, C., Tinazzi, O., 2009. Carbon-14 as a marker of seismic
Chiodini, G., Caliro, S., Aiuppa, A., Avino, R., Granieri, D., Moretti, R., Parello, F., 2011. First activity. Radiation Effects and Defects in Solids 164 (5), 376–381.
13 12
C/ C isotopic characterisation of volcanic plume CO2. Bulletin of Volcanology 73. National Institute of Occupational Safety and Health (NIOSH), 1997. Pocket Guide to Chem-
http://dx.doi.org/10.1007/s00445-010-0423-2. ical Hazard, DHHS (NIOSH). Publ. N.97-140, U.S. Gov. Print Office, Washington, DC.
Costa, A., Macedonio, G., Chiodini, G., 2005. Numerical model of gas dispersion emitted Oke, T.R., 1982. The energetic basis of the urban heat island. Quarterly Journal of the
from volcanic sources. Annals of Geophysics 48 (4/5), 805–815. Royal Meteorological Society 108, 1–24.
Costa, A., Chiodini, G., Granieri, D., Folch, A., Hankin, R.K.S., Caliro, S., Cardellini, C., Avino, R., Park, S., Kim, C., 1999. A numerical model for the simulation of SO2 concentrations in
2008. A shallow layer model for heavy gas dispersion from natural sources: applica- the Kyongin region, Korea. Atmospheric Environment 33, 3119–3132.
tion and hazard assessment at Caldara di Manziana, Italy. Geochemistry, Geophysics, Reid, K.H., Steyn, D.G., 1997. Diurnal variations of boundary-layer carbon dioxide in a coastal
Geosystems 9 (Q03002). http://dx.doi.org/10.1029/2007GC001762. city — observations and comparison with model results. Atmospheric Environment 31,
Derwent, R.G., Middleton, D.R., Field, R.A., Goldstone, M.E., Lester, J.N., Perry, R., 1995. 3104–3114.
Analysis and interpretation of air quality data from an urban roadside location in Stull, R.B., 1988. An Introduction to Boundary Layer Meteorology. Kluwer Academic
central London over the period from July 1991 to July 1992. Atmospheric Environ- Publishers, Dordrecht, The Netherlands.
ment 29, 923–946. Sweeby, P., 1984. High resolution schemes using flux limiter for hyperbolic conservation
Deutsch, C.V., Journel, A.G., 1998. GSLIB: Geostatistical Software Library and Users laws. SIAM Journal on Numerical Analysis 21 (5), 995–1011.
Guide. Oxford Univ. Press, New York, 369 pp. Tarquini, S., Isola, I., Favalli, M., Mazzarini, F., Bisson, M., Pareschi, M.T., Boschi, E., 2007.
Douglas, S., Kessler, R., 1990. In: Carr, L. (Ed.), User's Manual for the Diagnostic Wind TINITALY/01: a new Triangular Irregular Network of Italy. Annals of Geophysics 50
Model (San Rafael, CA, 1990) vol. III, EPA-450/4-90-007C. (3), 407–425.
EMEP/CORINAIR, 2006. Emission Inventory Guidebook — 2006. European Environment Toon, O., Turco, R.D.W., Malone, R., Liu, M., 1988. A multidimensional model for aerosols:
Agency Publishers, Copenhagen, Denmark (12 December 2006). http://www.eea. description of computer analogs. Journal of the Atmospheric Sciences 45, 2123–2143.
europa.eu/publications/EMEPCORINAIR4 (accessed Jan. 10, 2012). Wang, Y., Hutter, K., 2001. Comparisons of numerical methods with respect to
Ewing, R., Wang, H., 2001. A summary of numerical methods for time-dependent convectively dominated problems. International Journal for Numerical Methods
advection-dominated partial differential equations. Journal of Computational and in Fluids 37, 721–745. http://dx.doi.org/10.1002/fld.197.
Applied Mathematics 128, 423–445.
Frondini, F., Chiodini, G., Caliro, S., Cardellini, C., Granieri, D., Ventura, G., 2004. Diffuse
CO2 degassing at Vesuvio, Italy. Bulletin of Volcanology 66 (7), 642–651.

You might also like