Carbon dioxide emissions from Specchio di Venere

Bull Volcanol (2016) 78:29 DOI 10.1007/s00445-016-1023-6

RESEARCH ARTICLE

Carbon dioxide emissions from Specchio di Venere, Pantelleria, Italy Mariana P. Jácome Paz 1 & Salvatore Inguaggiato 2 & Yuri Taran 1 & Fabio Vita 2 & Giovanella Pecoraino 2

Received: 8 June 2015 / Accepted: 8 March 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract We have mapped the diffuse CO2 efflux from the Specchio di Venere Lake area using the accumulation chamber method. We calculated a CO2 emission of 43 ± 5 t day−1 for the area studied, accounting for both diffuse degassing from soil and bubbling through the lake. We also present data on the water composition of Specchio di Venere Lake, the Polla 3 spring, and Liuzza well. On the basis of water chemistry, two physical-chemical processes, evaporation and mineral precipitation of carbonate species, are invoked to explain the CO2 degassing for the lake area. Keywords Specchio di Venere Lake . Pantelleria . CO2 fluxes . Mineral precipitation . CO2 degassing

Introduction Carbon dioxide (CO2) is a valuable chemical tracer of subsurface magma degassing owing to its low solubility in silicate melts at low to moderate pressure (Gerlach and Graber 1985). Measurements on volcanoes can define diffuse degassing

Editorial responsibility: C. Oppenheimer * Mariana P. Jácome Paz [email protected]

1

Instituto de Geofísica, Universidad Nacional Autónoma de México, Av. Universidad 3000, Del Coyoacán CP 04510, DF, Mexico

2

Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Palermo, Via Ugo La Malfa 153, 90146 Palermo, Italy

structures (DDS after Chiodini et al. 2001) associated with high permeability zones such as faults and fractures. The transport of gases in a porous and homogeneous medium is the result of diffusion and advection (or convection) processes (Inguaggiato et al. 2013). Low flows are generally dominated by diffusion whereas high flows are dominated by advection (Chiodini et al. 1998). In general, diffuse degassing studies account for the combination of diffusion and advection fluxes through the volcanic edifice, excluding outputs from fumaroles, volcanic plume emissions, and shallow aquifers (López et al. 2004; Inguaggiato et al. 2011, 2012). Diverse techniques for measuring the CO2 efflux from soils have been developed. The accumulation chamber method has been most widely employed because of its portability and because the soil CO2 efflux can be measured in situ in a short period of time (~1–3 min) (e.g. Hernández et al. 2001; Pérez et al. 2012; Hernández et al. 2015). According to Chiodini et al. (1998), this technique is the best approach to measure soil CO2 efflux from volcanological-geothermal areas because it requires neither assumptions and nor corrections for soil characteristics. The physical mechanisms of CO2 degassing at volcanic lake surfaces (Bernard and Mazot 2004; Mazot and Taran 2009) are usually governed by bubbling with convectiveadvection processes (high CO2 fluxes) and by diffusion of CO2 through the water-air interface (lower CO2 fluxes). These enable measurement of CO2 efflux following the same principle of the accumulation chamber method applied to soils. A variation of the accumulation chamber method thereby enabled Bernard et al. (2004), Bernard and Mazot (2004), and Mazot (2005) to determine the CO2 emissions at the airwater interface in volcanic lakes. The purpose of this work is (i) to better understand the mechanism of CO2 degassing from Specchio di Venere lake area (ii) to carry out the first estimation of the total diffuse CO2 emission from Specchio di Venere lake surface with a floating

29

Bull Volcanol (2016) 78:29

Page 2 of 12

chamber system, and to identify the main CO2 degassing areas around the lake, and (iii) to estimate the entire CO2 flux of the system lake-shore-surrounding areas. Geological setting Pantelleria Island is a quiescent volcano located 110 km south of Sicily and 70 km north of Tunisia. The total emerged area of the Island is 84 km2 and its highest point is Montagna Grande (836 masl). The island is the emergent part of a larger submarine volcanic edifice in the Sicily Strait continental rift. This rift has a floor of 20 km thick continental-type crust that is characterized by horsts and graben formed by NW-SE tensional faults and NE-SW shear faults. Several faults around the island follow these regional trends (Dongarrá et al. 1983; Catalano et al. 2009). The main local tectonic structures on Pantelleria Island are represented by NNE trending dip-slip normal faults and by NW striking right-lateral strike-slip faults with normal component of motion; this mode of deformation has controlled the development of the eruptive fissures, dykes, and eruptive centers along NNE-SSW to NE-SW axes within the younger caldera on the island and a NW-SE trend in the northwestern sector of the island (Catalano et al. 2009). There are two main volcano-tectonic structures on the island: La Vecchia or Serra Ghirlanda caldera formed approximately 114 ka ago in the Fig. 1 Location of the Specchio di Venere Lake on Pantelleria Island

central part of the island, and Cinque Denti or Monastero caldera (~45 ka ago), a subcircular structural feature about 6 km wide in the northwest part of the island (Cornette et al. 1983). Montagna Grande, the highest point of the island (836 masl) is a trachytic block uplifted and tilted about 35 ka ago in the center of Cinque Denti caldera floor (Cornette et al. 1983; Civetta et al. 1988). Other minor structures include the Zinedi fault to the northeast, the NNE Montagna Grande fault and minor NNE-SSW and subordinately NW-SE fractures in some cases associated to intense activity of fumaroles, especially in the Favara zone (Favara et al. 2001; Granieri et al. 2014) Volcanic activity Pantelleria’s past volcanism is characterized by calderaforming eruptions in the south and effusive activity in the north. The most recent activity involved several small submarine eruptions off the NW coast in 1831, 1845, 1846, 1863, and 1891 (Butler 1892). Activity today is limited to low temperature fumaroles (up to 100 °C) and thermal springs with temperatures up to 90 °C (Duchi et al. 1994; Dongarrá et al. 1983). Main gas emissions are located in Favara, Passo del Vento, Mt Gibele, and Cuddia di Mida areas (Fig. 1). Bubbling gases are evident in the Specchio di Venere (SV hereafter) Lake and the Gadir gulf, (northwest and north of


the island; Favara et al. 2001). Anomalous diffuse CO2 emissions along faults and fractures are observed in the area of Mt. Grande and in several areas close to the main fumarolic emissions and a mofette on the east shore of SV Lake and in the Fossa della Pernice area (Favara et al. 2001).

Specchio di Venere Lake The Specchio di Venere, also called Bagno dell’ Acqua, is an alkaline, saline endorheic lake located at the northern rim of Cinque Denti caldera. According to Madonia et al. (2013), its morphology is a result of the accumulation of meteoric water into an endorheic depression formed as a result of volcanism sea level changes and ground deformation. The SV Lake is approximately 450 m by 350 m; its floor is divided into two distinct sectors: the southwestern one formed by an apparently flat submerged platform, (depth 50 °C), lake water has higher temperatures ranging between 34 and 54 °C. Finally, water-rock interaction (dissolution and precipitation) and evaporation strongly contribute to the peculiar chemical composition of the lake water resulting in high contents of Cl and SO4, being classified as chloride-sulfatealkaline type with Cl as predominant species (Dongarrá et al. 1983). The geochemical model of SV Lake (Aiuppa et al. 2007; Granieri et al. 2014) highlights that the meteoric recharge zone of SV Lake is Mount Gelfiser (400 masl) located on the southwest side of the lake. The pH of the lake is high with minor variations (9.18 to 9.27) at the surface (Aiuppa et al. 2007). On the contrary, the shallow lake temperature varies from 20.6 to 25.4 °C in the northernmost side and from 35 to 55 °C in the southernmost side near the hot springs; the main factors contributing to these differences are (i) distance from the hot springs, (ii) changes in air temperature, (iii) solar radiation, and (iv) wind speed.

Page 3 of 12 29

CO2 manifestations and previous studies The main CO2 manifestations on Pantelleria Island include fumaroles, soil diffuse emissions, and bubbling and dissolved gases in the ground waters (Favara et al. 2001). The first soil CO2 study on Pantelleria Island was performed by Favara et al. (2001). They estimated the CO2 flux of the entire island (only 146 measurements) using the dynamic concentration method. A total CO2 output of 0.39 Mt year−1 (1068 t day−1) was estimated for the entire island. The main contribution to the total CO2 output was from diffuse soil degassing (877 t day−1), followed by dissolved CO2 in groundwater (93 t day − 1 ), focused soil degassing (77 t day−1), and bubbling CO2 in waters (36 t day−1). The fumaroles represented a negligible contribution of 3.8 kg day−1. They identified the western and southwestern shores of SV Lake with a CO2 emission of 41 t day−1 for an area of 0.16 km2 and the Favara area with a CO2 emission of 36 t day−1 for an area of 0.1 km2 as two zones of enhanced soil degassing. These areas have long been recognized as sites of anomalous degassing with the presence of thermal springs with abundant bubbling gases and a mofette in the former and active fumarolic vents with temperatures of about 100 °C in the latter (Favara et al. 2001). D’Alessandro (2007) carried out a soil CO2 survey with 807 measurements for the entire island and found the CO2 output concentrated in three main zones: SV Lake, Favara zone, and Mount Grande. They confirmed the sites of anomalous flux with the highest values for the Favara Grande with 5.87 t day−1 CO2 flux (0.096 km2), SV Lake with 30 t day−1 CO2 flux (0.103 km2), and Mount Grande with 14.5 t day−1 (0.028 km2). Granieri et al. (2014) estimated CO2 emission for the west shore of SV Lake close to the mofette of Fossa della Pernice (450 m2) and in Favara Grande and Favara Piccola areas. The study was performed in the cold season of 2004 using the accumulation chamber method (313 measurements in selected areas). They found the highest CO2 effluxes on the western shore of SV Lake, concentrated in the northern portion of the mofette zone with a mean value of 92 g m−2 day−1 and a range of values between 2.3 and 690 g m−2 day−1. They calculated CO2 fluxes of 7.1 t day−1 for the Favara area (0.09 km2) and 0.04 t day−1 for the SV lake shores covering an area of 450 m2. They estimated a total flux of 19.3 CO2 t day−1.

Methodology Sampling and analytical methods The survey was carried out in June 2014. More than 400 CO2 efflux measurements were performed for soil surfaces and the lake surface. Measurements were made in situ with a portable

29


Page 4 of 12

West System® flowmeter equipped with an infrared spectrometer LI-COR 1800® under dry stable atmospheric conditions (June 2014). Soil CO2 efflux measurements were made with the Btraditional^ accumulation chamber method while CO2 efflux measurements on the water lake were performed with a modification of the accumulation chamber method including a floating system described in Bernard et al. (2004), Bernard and Mazot (2004), and Mazot et al. (2011). Were performed 226 points over SV lake surface and 230 points over soil CO2 efflux measurements around the lake. Together with the CO2 efflux measurements for the surface lake, water temperature was measured at 10 cm depth. Soil gas samples at different depths (25, 50, 75, and 100 cm) were collected from the mofette area to analyze chemical and isotopic composition. The Liuzza well was drilled inside the high CO2 flux area near the mofette and contains sparkling water while the Polla 3, located on the rim of Crater Lake, is the thermal water flowing inside the lake. Water samples were stored in polyethylene bottles to analyze major components and stable isotopes compositions. At each site, outlet temperature, electrical conductivity, and water pH were measured using an ORION 250A+ conductometer and thermocouple and an ORION 250A+ pH-meter (±0.5). Water samples for cations were acidified with Suprapur® HNO3. Alkalinity was analyzed in situ by titration with HCl 0.1 N. Major anions and cations were analyzed at the INGV lab (Palermo) using a Dionex 1100 IC in suppressed mode and equipped with a AS14A anion column and a AG14A pre column working with continuous flow of carbonatebicarbonate eluent and CS12A cation column and CG12A pre column that works under continuous flow of MSA (methane sulfonic acid) with ER eluent regeneration. Analysis of dissolved gas content in natural waters samples was carried out following Capasso and Inguaggiato (1998) and based on the partitioning equilibrium of gases between liquid and host gas phase (Ar). Water samples for dissolved gas analysis were collected in totally filled and sealed flasks with known volume. The equilibrated gases were analyzed by double detector (TCD, methaniser-FID configuration) Perkin Elmer Clarus 500 gas chromatograph using Ar as carrier gas and a Restek Shincarbon ST 3 m packed column. Analyses of the dissolved He isotopic composition were performed following Inguaggiato and Rizzo (2004). Analysis of CO2 fluxes Graphical Statistical Approach method (GSA) One way to estimate the total soil CO2 flux is based on a statistical graphic methodology which consists on the separation of data in log-normal families or populations with different ranges of fluxes highlighted by the inflection in the

cumulative probability plot (Sinclair 1974; Chiodini et al. 1998; Cardellini et al. 2003). The cumulative probability, the data histograms, and logarithmic probability plots show main log-normally distributed populations reflecting the different source contributions of CO2 degassing for the study area. The estimated mean flows are used to calculate the total flux associated to each population (David 1977). An estimate of the area covered by each population (Si) is calculated by multiplying the total area S by the proportion of the family in question (fi): Si ¼ f iS

ð1Þ

The total flux of carbon dioxide (CO2)T is estimated multiplying the area of each family by the mean flux of the family (Mi). X ðCO2 ÞT ¼ SM ð2Þ i i i Although this method is a good approximation to understand the contribution of different sources to the diffuse CO2 emission, the results can be affected by errors in choosing the inflexion points. The multi-modal log-normal distribution of CO2 fluxes is a suitable model, but does not necessarily correspond to the natural system investigated and the partition on families is not unique (Chiodini et al. 1998). Therefore, we just used this methodology in order to separate families before carrying out the total flux calculation and analyzing the spatial distribution.

Gaussian simulation algorithm (sGs) The sequential Gaussian simulation methodology is used to estimate the total CO2 output, spatial distribution of CO2 fluxes and to recognize anomalous or high permeability zones (Cardellini et al. 2003). This methodology is based on the sequential sGs and described in Deutsch and Journel (1998). According to Cardellini et al. (2003), the simulation is conditional and sequential, i.e., the variable is simulated at each unsampled site by random sampling of a Gaussian conditional cumulative distribution function defined on the basis of the main statistical values of the original data set and of previously simulated data within its neighborhood. Following the sGs procedure, CO2 efflux values from SV Lake were simulated in a distribution map using a grid of 10 × 10 m, for the south of the studied area, a grid of 2 × 2 m and for the mofette zone, a grid of 1 × 1 m. Since processed data set for each studied zone did not show a normal distribution, an n-score transformation was applied to the CO2 efflux values. After obtaining the variogram and spherical model with nugget effect less than 0.35, 100 simulations were done

Bull Volcanol (2016) 78:29 Table 1 text

Page 5 of 12 29

Water chemical composition of the SV Lake (meq/l). TDS is expressed in gram per liter. Historical values were taken from references cited in

Date 1881 1963 1966 June 1980 Sept. 1980 Dec. 1980 June 1987 April 1989 May 1990 1990 Jan. 1994 Oct. 1994 Sept. 1995 May 1996 May 2008 Dec. 2008 July 2014

T

pH

Na

K

Ca

21.4

9.1

173 264.2 210 272

5.29 9.87 5.91 8.6

0.5 0.2 0.3

4.03 5.6 6.91 9.3

159 197.5 160 227.5

26.2 15.3

9 9 9.1 9.1 9.2 9

281 278 310 357 313 384.8 325.5 280 358 273, 3 324.3 323.2 351.86

10.2 14.6 10.3 10.6 13.7 20.5 10 5.17 14.2 9. 40 9.2 9.2 11.56

0.3 0.2 0.1 0.3 0.5 0.5 0.54 14.2 17.9 1 0.5 0.5 1.3

10.3 9.7 10.8 12.9 12.1 7.98 13.1 7.6 9.95 10.3 12.7 12.8 15.99

232.9 229.2 255 301 260 340 271.3 238.5 327 234.7 286.7 313.9 297

23.5 24.5 26

19.7 26, 2 14, 5 27.2

9.25 9. 3 9. 1 8.97

Mg

Cl

Br

0.43 0.36

0.53 0.4

0.43

Alk

TDS (g/l)

Reference

7.1 14.4 7.1 13.8

58 37.8 72.8 62.1

10.2 14.2 11.2 15.5

Foerstner (1881) Bencini et al. (1966) Bencini et al. (1966) Dongarrá et al. (1983); Azzaro et al. (1983)

14.9 20.3 17.9 19.9 22.9 20.6 18.4 16.3 20.8 15.1 17.0 16.9 25.22

63.2 63.4 61.5 67.2 58.7 79.8 59.9 48.9 54.2 52.6

16.0 16.2 17.6 20.4 18.2 22.8 18.6 16.0 21.6 15.9 19.0 19.9 20.6

Dongarrá et al. (1983)); Azzaro et al. (1983) Dongarrá et al. (1983)); Azzaro et al. (1983) Bocchi et al. (1988) D’Alessandro et al. (1994) D’Alessandro et al. (1994) D’Alessandro et al. (1994) D’Alessandro et al. (1996) D’Alessandro et al. (1996) D’Alessandro et al. (1996) Parello et al. (2000) Cangemi et al. (2010) Cangemi et al. (2010) This work

57

where kCO2 is the gas transfer velocity (in cm h−1) for CO2, Cw and Cw/a refer to the concentration of CO2 in water and in the water film at the water-air interface, respectively, and 240 is the conversion factor to transform milligram per square centimeter per hour in gram per square meter per day unit (Liss and Slater 1974). kCO2 is a coefficient dependent on wind speed (measurable in field) and can be estimated according to the following equation (Liss and Slater 1974):

for each studied area followed by the estimation of the mean E-type value for each cell using the post processing tools of Wingslib® software.

CO2 degassing from the water surface In order to compare the statistical techniques and physical models for CO2 degassing flux, the flux (F) between water and air can be calculated using the thin boundary layer model (Liss and Slater; 1974; Bernard et al. 2004). These authors describe the use of a two-layer model to estimate the flux of various gases across the air-sea interface. The diffusive flux is calculated by an empirical equation similar to Fick’s first law 0 1 F g m−2 day−1 ¼ k CO2 @C w −C . A 240 ð4Þ w

SO4

0:063 k CO2 ¼ 6:81 0:0538 u31

ScCO2 600

−12

ð5Þ

Where u1 is the wind speed at 1 m height and ScCO2 is the Schmidt number, i.e., the ratio between water viscosity and diffusivity of CO2 at the sampled T, defined as: ScCO2 ¼ 1911:1−118:11t þ 3:4527t 2 −0:04132t 3

a

ð6Þ

Table 2 Water chemical and isotopic composition of Polla 3 spring and Liuzza well (meq/l). Polla 3 was named as Bsorgente sud^ by Parello et al. (2000). Liuzza well is nearest to the mofette zone Sample name

T

pH

Na

K

Mg

Ca

F

Cl

Br

SO4 Alk R/Ra He/Ne δ13C (TDIC) Reference

Polla 3 July 2014 50.6 6.41 132.2 4.15 8.43 3.66 0.5 117.9 0.19 8.51 24 4.02 1.208 Polla 3 May 1996 58. 0 6.3 136.9 4.5 3.8 8.7 118.3 7.2 25.2 2.39 Liuzza well July 2014 26.9 6.05 21.63 1.2 2.84 1.14 0.44 9.58 0.02 1.41 14.7 4.09 0.933

−1.02 +3.47

This work Parello et al. (2000) This work

29


Page 6 of 12

Fig. 2 Langelier diagram and Cl-SO4-Alk diagram for SV Lake, Polla 3 spring, and Liuzza well water samples

Where t is the water temperature in degree Celsius.

The equilibrium coefficient depends of the activities of each component: . K ¼ ½HCO3 ‐ ½Hþ ½H2 CO3 ¼ 10‐6:34 ðAt 25 CÞ

Cw calculation and analytical laboratory technique For the calculation of dissolved CO2 in water (Cw), we used the field-measured parameters (pH and HCO3) and assuming chemical equilibrium between the carbonated species in solution. This estimation is based in the following equations: CO2 ðgÞ þ H2 O ðwÞ ¼ H2 CO3 ðwÞ −

H2 CO3 ðwÞ ¼ HCO3 þ H

Table 3 Place

Resulting in [H2CO3] = CO2 (w) = 106.34[HCO3−] [H+], as pH = −log [H+], then log CO2 (w) = 6.34 − pH + log [HCO3−]. Cw/a is taken from the mean value of CO2 concentration in the atmosphere during measurements. With the aim of correcting SV Lake CO2 flux values, the environmental parameters, mean temperature, and mean wind

ð7Þ

þ

ð9Þ

ð8Þ

Gas chemical and isotopic composition of mofette zone Sampling date He (ppm) H2 (ppm) O2 (%) N2 (%) CO (ppm) CH4 CO2 (%) δ13C

R/Ra (R/Ra)c Reference

Mofette 13/09/1990 Mofette 17/10/1994 Mofette 16/05/1996 Mofette 25 cm 03/07/2014 Mofette 50 cm 03/07/2014

10 8.1 16 15 14