Carbon dioxide emission from Katanuma volcanic lake, Japan

The lake Katanuma, located in northwestern Miyagi prefecture, Japan (38°44′N, 140°43′E; 306 m above sea level) belongs to the Narugo volcanic system (Fig. 1), which is one of the 44 Holocene volcanoes in the northeastern Japan arc. According to Sakaguchi and Yamada (1988) and Soda and Yagi (1991), the activity of the Narugo volcano began with pyroclastic activity; the Nizaka pyroclastic flow (73 ka) and the Yanagizawa pyroclastic flow (45 ka). A large caldera, of about 5 km in diameter was formed by these paroxysmal eruptions. Lake deposits found in the caldera indicate that water filled the caldera after 45 ka. Afterwards, lava flows and lava domes were formed in the inner part of the caldera. The youngest lavas, Toyagamori, have been ¹⁴C dated to be 11830±555 years BP (Omoto, 1993) (Fig. 2). Observed craters were formed in the lava surface by phreatic eruptions. The only known historical eruption at Narugo occurred in 837 AD.

Katanuma is a crater lake with strong acid water (pH: 2.0–2.2) influenced by the volcanic fluids. The lake forms an ellipse of 462 m E-W length and 326 m N-S breadth, with a 1380 m of shore line it has neither inflowing nor outflowing streams. The surface area is 1.26 × 10⁵ m² with a maximum depth of 21.3 m with an approximately volume of 7.48 × 10⁵ m³. In the bottom of the lake there are some high temperature zones where hot springs flow out (Sato, 1995). These zones are located at the south and northeast basins of the lake’s deepest area (Fig. 3). Volcanic gases emanating at the bottom and along the shore line of the lake dissolve in the shallow part and acidify the water by redox reactions of H2SO3, with heat supplied at the same time at arateof372 × 10⁻⁶ cal cm⁻² s⁻¹ (Sato, 1995).

Lake Katanuma is dimictic and covered by ice from January to mid-March. Inverse stratification occurs under the ice cover. After turnover in March, weak stratification develops with a thermocline at a depth of 3–5 m. Stratification is generally observed from April to late August. The circulation period is from late August to December (Takagi et al., 2005).

The CO₂ efflux survey was performed during August 2010 under stable weather conditions (no wind and sunny). 110 sampling sites were selected at the surface of Katanuma lake to obtain an even distribution of the sampling points over the area of 0.14 m² (Fig. 4). The GPS positioning of each measurement point was recorded with a resolution ±5 m. The accumulation chamber method (Parkinson, 1981; Chiodini et al., 1998) was modified in order to work on a lake by using a floating chamber (Fig. 5). Measurements of CO₂ were performed in-situ by means of a portable non dispersive infrared (NDIR) CO₂ analyzer (LICOR-800 system). The LICOR analyzer was interfaced to a hand size computer that runs data acquisition software. At each sampling site, water temperature, conductivity and pH were measured at a depth of 40 cm bellow the surface by means of a thermocouple and a portable pH meter. The pH meter was calibrated on site before the start of the survey.

Surface CO₂ efflux, water temperature and water pH maps for Katanuma lake were constructed using Conditional sequential Gaussian simulations (sGs) provided by the sgsim program (Deutsch and Journel, 1998; Cardellini et al., 2003). In order to characterize the chemical and isotopic composition of the bubbling gas from Katanuma lake, gas samples were collected and gas composition was analyzed later in the laboratory. Gas composition was analyzed by combining a Varian CP-3800 gas chromatograph with Ar as the carrier gas and a Quadrupole Mass Spectrometer (QMS) model Pfeiffer Omnistar 422. The ¹³C/¹²C ratio of CO₂ (expressed as δ¹³C-CO₂ ‰ vs. V-PDB) was determined with a Thermo-Finnigan MAT 253 mass spectrometer at the ITER laboratory. The analytical error for δ¹³C values is ±0.1‰. Elemental abundances of He and Ne, and He isotope composition of the gas samples collected in a 50 cm³ lead glass bottle with high vacuum stopcocks were analyzed in the laboratory of the Geochemical Research Center, The University of Tokyo. A magnetic-sector mass spectrometer VG 5400 was used for the analysis following the methodology described by Sumino et al. (2001).

CO₂ efflux values ranged between bellow the detection limit of 0.5 and 322 g m⁻² d⁻¹. Water temperature and pH ranged from 28.1 to 30.6 and 2.00 to 2.19, respectively. Figure 6(a) shows the histogram of log CO₂ efflux values (in g m⁻² d⁻¹) versus their frequency. The distribution of CO₂ efflux differs from a log-normal distribution indicating that there are at least two different mechanisms of degassing through the lake surface. Following the methodology described by Sinclair (1974), a Log-normal probability graph was constructed with the CO₂ efflux data (cumulative percentile frequencies against class intervals) to distinguish the presence of overlapped Log-normal geochemical populations that are recognizable by the inflection point on the curve (Fig. 6(b)). The inflection point of the curve allows us to distinguish the threshold value between the populations. The resulting probability graphs allowed to separate three geochemical populations: population I showing a mean of 21.1 g m⁻² d⁻¹ and represented 24.3% of the total data; population II, showing a mean of 98.2 g m⁻² d⁻¹ and represented 29.63% of the total data, and population III, showing a mean of 292.5 g m⁻² d⁻¹ and represented 3.3% of the total data. Rest of the cumulative probability corresponds to the mixing between the three identified populations. Population I represents the background values due to CO₂ degassing by diffusion through the water-air interface as has been proposed by Mazot and Taran (2009) for El Chichón lake, México, and checked using the thin boundary layer model (Liss and Slater, 1974). Population II represents the CO₂ efflux values from a deep source of a magmatic origin and Population III may represent gases resulting from mixing of diffuse and bubbling degassing. The existence of three different geochemical populations indicates that there are at least two different mechanisms of degassing through the lake surface, and supports the hypothesis of a deep contribution at Katanuma lake.

A total of 200 simulations were performed over a grid of 5696 squared cells (5 m × 5 m) following the variogram model to construct a spatial distribution map for CO₂ efflux (Fig. 7), water temperature (Fig. 8(a)) and pH (Fig. 8(b)). The experimental variograms were fitted with a spherical model with the following parameters: nugget of 0.15 and range of 45 m. An average map was then constructed using the average of the different values simulated at each cell. Since quantification of the uncertainty of the total CO₂ is an important task, the mean and standard deviation of the 200 simulated values of total CO₂ output were assumed to be characteristics of the CO₂ efflux and its uncertainty (Cardellini et al., 2003). Figure 7 shows the CO₂ efflux distribution map at Katanuma lake based on the mean simulated total CO₂ output values.

Inspection of the CO₂ efflux contour map for Katanuma lake (Fig. 7) shows the existence of three anomalous areas of CO₂ degassing through the lake surface without a clear emission trend. High CO₂ efflux values were always measured nearby visible gas bubbling emanations and located above the deepest areas of the lake, south and northeast basins of the lake, where hot springs are thought to issue. Figure 8(a) shows water temperature contour map for Katanuma lake together with bathymetric contour lines. Even when there is a homogeneous distribution of temperature at the surface of the lake, some slight increase in temperature was observed at the southern most part of the lake. To estimate the total output of diffuse CO₂ released from Katanuma lake, we considered the contribution of each cell obtained after SGS and the average of the 200 simulations to estimate the total output with one standard deviation as the uncertainty. The result was 17±0.6td⁻¹ for diffuse CO₂. This value is smaller than the emission rates reported for other volcanic lakes (Pérez et al., 2011). However, in order to compare individual CO₂ emission of various volcanic lakes, it is important to consider the lake-type and the area. Figure 8(b) shows water pH contour map for Katanuma lake together with bathymetric contour lines. Most of the surface water showed an almost homogeneous acidic pH (∼2.0) distribution. Katanuma, as an acid lake with an area of 1.26 × 10⁵ m², would have a normalized CO₂ emission of 136 t d⁻¹ km⁻², a value close to that from other acid volcanic lakes in the world (Pérez et al., 2011).

As reported by Takagi et al. (2005), Katanuma lake circulates from late August to December and stratifies from April to late August. Our survey was carried out at the beginning of the monomictic circulation period (August), which significantly reduces accumulation of CO₂ in the deepest part of the lake, and maximizes CO₂ emission to the atmosphere. During the overturn period, when the present study was carried out, the emission of CO₂ is mainly diffusive, advective and convective, showing the maximum CO₂ emission in the year. In order to evaluate the CO₂ emission from the bubbling spots, we measured the CO₂ flux from 10 spots using the same method. We visually counted approximately 100 bubbling spots and assumed an average CO₂ emission rate per bubbling spot of 14 kg d⁻¹. Under the above assumption we estimated a total CO₂ emission of 1.4 t d⁻¹ from bubbling degassing. As a whole, adding diffuse and visible bubbling degassing, the total CO₂ emission from Katanuma is estimated to be 18.4 td⁻¹.

Since water from Katanuma lake is acidic (pH ∼2.0), almost all CO₂ is present as dissolved CO₂ and will diffuse out of the water, allowing volcanic gas fluxes to be channeled through the acid lake. Therefore, 18.4 td⁻¹ of CO₂ can be considered as a deep-seated CO₂ emission from Katanuma lake, because relatively high CO₂ efflux values were measured during this survey including data from all geochemical populations (16.6–322gm⁻² d⁻¹).

Chemical and isotopic compositions of bubbling gases were analyzed to constrain the origin of the gases emitted through the water of Katanuma lake. Bubbling gases are clearly CO₂-dominated (72.6%V) and contain He (10 ppmV), O₂ (4.6%V), N₂ (22.5%V), Ar (0.2%V) and CH₄ (477 ppmV) in minor amounts. N₂/Ar ratio at the Katanuma bubbling gases (112.5) is notably higher than the air ratio (83.6) and air-saturated water (ASW; ∼36.8), indicating an additional non-atmospheric N₂ source, typical for arc volcanoes. Measured δ¹³C-CO₂ value for Katanuma bubbling gases (−1.59±0.03‰) is heavier than MORB values of −6.5±2.5‰ (Pineau and Javoy, 1983; Taylor, 1986). The measured value indicates a typical isotopic composition for gases from volcanic arcs (e.g. Taylor, 1986; Marty et al., 1989; Sano and Marty, 1995; Sano and Williams, 1996; Pedroni et al.1999; Shaw et al.2003).

The ³He/⁴He ratio was also measured at 3.46±0.06 Ra, where Ra denotes the atmospheric ³He/⁴He ratio (1.4 × 10⁻⁶, Mamyrin et al.1970). The relatively low ⁴He/²⁰Ne ratio of the sample (3.73) is higher than that of the atmosphere (0.318), but still suggests atmospheric contamination of the sample. The measured ³He/⁴He ratio is similar to the value of 4.86±0.17 Ra reported by Asamori et al (2010) from the hot spring nearest to Naruko volcano. The value supports the existence of a hydrothermal system which supplies materials from underlying magma. In general, He in natural gases is composed of three components: atmospheric He, crustal or radiogenic He, and primordial He derived from the upper mantle. In order to estimate the mixing ratio of these components using the observed ³He/⁴He and ²⁰Ne/⁴He ratios, we followed the methodology proposed by Sano et al. (1982). The calculated contribution for each component was 35.9%, 55.6% and 8.5% for MOR-type (upper mantle), crustal and air, respectively. The large contribution of the upper mantle He in Katanuma gas sample confirms active magmatic activity beneath the Naruko-Katanuma volcanic system.

Bubbling springs with lower gas/water ratios generally show higher ³He/⁴He ratios and consequently lower CO₂/³He ratios (Rouwet et al, 2009). As He is less soluble in water than CO₂ (Stephen and Stephen, 1963), and ³He is lighter than ⁴He, gases liberated from a weakly bubbling spring tend to be enriched in ³He during near-surface degassing processes and vapour separation in an aqueous hydrothermal system. This has been observed in the other volcanic areas, e.g. Tacaná volcano in Guatelama (Rouwet et al, 2009).

In order to detail deep magmatic and near-surface physical degassing processes at Katanuma lake, we have calculated the CO₂/³ He ratio of the bubbling gas. The CO₂/³He ratio is 1.54 × 10¹⁰, a value that fits well with the average value established for worldwide arcs (1.5 ± 1.1 × 10¹⁰; Sano and Williams, 1996). The three-component mixing model reported by Sano and Marty (1995) to calculate relative proportions of carbon sources contributing to arc magmas; MORB (M), marine limestones (L) and organic sediments (S) was used to estimate the fraction of each C source for Katanuma gas sample. L source contributes 53.6%, S source 3.6% and the M source 42.8%. These values are in the same order than the reported by Sano and Marty (1995) for bubbling gas and fumarolic gas samples from Japanese volcanic areas, with limestones as the main C component followed by the MORB-type C. Organic sedimentary C fraction is relatively low (3.6%), suggesting that the fumarolic feeding system beneath Katanuma lake corresponds to high temperature fumaroles.