Main drivers of diffusive and advective processes of CO2-gas exchange between a shallow vadose zone and the atmosphere

International Journal of Greenhouse Gas Control 21 (2014) 113–129

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc

Main drivers of diffusive and advective processes of CO2 -gas exchange between a shallow vadose zone and the atmosphere E. Garcia-Anton a,∗ , S. Cuezva b , A. Fernandez-Cortes a , D. Benavente b , S. Sanchez-Moral a a b

Department of Geología, MNCN-CSIC, C/José Gutiérrez Abascal, n.2, 28006 Madrid, Spain Department of Ciencias Tierra y Medio Ambiente, Universidad de Alicante, 03080 Alicante, Spain

a r t i c l e

i n f o

Article history: Received 25 July 2013 Received in revised form 23 October 2013 Accepted 6 December 2013 Available online 28 December 2013 Keywords: Diffusion Advection CO2 mass balance Gaseous CO2 transport mechanisms Vadose zone

a b s t r a c t A multiparametric study of Altamira cave conditions was performed to identify mechanisms that affect CO2 . A daily survey was used to better understand the role of the shallow vadose system as a source/sink of this gas. Airborne particles were monitored to distinguish the air movement that was joined to ␦13 CO2 and were also used as a proxy of the origin of the CO2 . A gas transport model has been created based on the interaction of three air masses (soil–cave–exterior), which is driven by soil-derived CO2 diffusion to the cave and by the advective mixing of the cave with exterior air. The diffusive process increases cave CO2 and decreases ␦13 CO2 . The advective mixing induces a decrease in CO2 and an increase in the isotopic signal. The diffusive flux depends on soil CO2 production; the advective flux is driven by outer–inner density gradients, and both depend on the degree of exchange between air masses. Consequently, external conditions, such as temperature and humidity, regulate gas interchange. The created process-based model permits the quantification of CO2 fluxes. The consequence of the degassing stage is the release of light CO2 (␦13 C quantified in −24.82‰) into the exterior air (␦13 C measured in −11.34‰). The migration of gas in the vadose zone may influence many environmental processes, and therefore, the contribution of shallow underground systems to surface CO2 exchange and to the isotopic signal of troposphere should be accounted for. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction The region that is located between the soil surface and the groundwater table, which is called the vadose zone, has traditionally been studied within a hydrologic context (Hopmans and Van Genuchten, 2005). However, the vadose zone atmosphere may contain much CO2 that occupies pores, cracks and voids of soil, bedrock or unconsolidated sediment (Benavente et al., 2010; Kell, 2012). This CO2 source has traditionally been neglected or underestimated in studies regarding the net carbon balance in terrestrial ecosystems (Serrano-Ortiz et al., 2010). Furthermore, these shallow vadose environments show significant seasonal, and even daily, variations in CO2 concentration, which involve the exchange of much CO2 (g) with the lower troposphere and its role as a depot and/or emitter (Cuezva et al., 2011; Bourges et al., 2012). Within the complex set of biological, physical and chemical processes that are involved in the interstices of soil and bedrock (subsurface), the migration of gas movements in the vadose zone plays a crucial role in many environmental processes. The ongoing interest in below-ground CO2 capture and storage (CCS) as one

∗ Corresponding author. Tel.: +34 914111328 1178; fax: +34 915644740. E-mail address: [email protected] (E. Garcia-Anton). 1750-5836/$ – see front matter © 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijggc.2013.12.006

potential mitigation strategy to reduce human CO2 atmospheric emissions has emphasised the need for more knowledge regarding the geological storage capacity (Post et al., 2012; Nickerson and Risk, 2013). Assessments are needed to ensure that there is no CO2 leaking from the storage formation and seeping out of the subterranean environment (Cohen et al., 2013). In this regard, a complete understanding of the medium- and short-term gaseous CO2 transfer processes through the subterranean environment is key. Ultimately, the mechanisms that control the isolation, recharge and storage processes of gases in the subsurface environment must be identified, particularly, the ventilation/venting processes of these subterranean environments and the resulting release/loss of stored gaseous CO2 to the Earth’s atmosphere. Shallow caves are favourable sites to investigate the transport mechanisms that control gaseous exchange processes at the subsurface because these caves are easily accessible natural macropores that are close to the surface. Thus, recent attention has been given to the role of the double membrane system (host rock and soil), which envelops the air that is contained in caves in the uppermost vadose zone. The role of macro- and micro-fissure rock networks has been taken into account in cave ventilation processes (Baldini et al., 2006; Bourges et al., 2006; Nachson et al., 2012). Other studies have revealed a complex process of gas exchange within a karstic subterranean environment that is controlled by the blockage of airways, such as

114

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

thin fractures and connected pores, by water, following an increase in the water saturation of soil and rock by vapour condensation or infiltration (Fernandez-Cortes et al., 2011, 2013). In the short-term, the interaction pattern of the soil porous system with daily humidity fluctuations determines the daily cyclic operation of this double membrane (circadian pattern) and the CO2 subsurface storage or emission processes (Cuezva et al., 2011; Maier et al., 2010). However, CO2 transport mechanisms that act at a daily scale are not fully resolved. In addition, the quantification and daily balance of these subsurface CO2 flows and their relevance to regional CO2 remain unknown. The stable isotopic composition of gaseous carbon dioxide is a useful tool for understanding carbon cycling processes and has been widely used to determine the contributions of multiple carbon sources and the mechanisms of soil gas transport. The isotopic composition of soil CO2 is a window through which we may understand an array of chemical reactions, as well as physical and biological processes, in the soil environment (Amundson et al., 1998). ␦13 C isotopic analyses allow for the identification of carbon contributions to the soil CO2 efflux, as well as the relative contribution of soil carbon pools to the overall ecosystem respiration (Bowling et al., 2008). Specifically, these analyses allow us to distinguish and quantify the contributions of autotrophic and heterotrophic sources to soil respiration (Ekblad and Högberg, 2001; Formanek and Ambus, 2004; Vargas et al., 2011). Additionally, the prevailing mechanism of transport and of the soil CO2 efflux process can be ascertained, particularly providing information about the relative contributions of diffusion and advection (Cerling et al., 1991; Kayler et al., 2010; Bowling and Massman, 2011). Equivalent isotopic approaches are promising experimental techniques to obtain direct quantitative links to the sources of CO2 and to the mechanisms that drive gas transport and the transfer of C in soil–cave systems (Breecker et al., 2012) and in the atmosphere–soil–vadose subsurface. This study aims to model and quantify carbon/CO2 (g) transference processes in the vadose subsurface environment, which is crucial to assess the carbon sequestration or release in these terrestrial ecosystems. To accomplish this aim, we present a study that is based on a combination of the continuous multi-parameter monitoring of atmospheric characteristics (main climatic data and gas composition) and the stable carbon isotopic (␦13 CO2 ) data, in all three media that are involved (cave–soil–troposphere). Furthermore, a suspended aerosol study is used to discern air movement inside the cave at different times.

2. Site, materials and methodology 2.1. Field site description The study takes place in the Altamira cave (43◦ 22 40 N; Cantabria Province, Spain), which is a shallow vadose karst cavity that is characterised by remarkable stable environmental conditions (Quindos et al., 1987; Sanchez-Moral et al., 1999; Cuezva et al., 2009). This cave, due to its relatively small size, good accessibility, no tourist impact (Saiz-Jimenez et al., 2011), and sealed entrance is a suitable natural laboratory for gas and microclimatic monitoring under non-disturbed environmental conditions. The main entrance is closed by a metal gate with a highly insulating heat insulation core (slotted surface <4%), which acts as the initial barrier to stop the exchange of energy and matter with the outside. In addition, a second door isolates the Entrance Hall from the rest of the cavity (primarily Polychrome Hall and Walls Hall, Fig. 1). The cave is in the upper vadose area of the karstic system, under a hill 161 m.a.s.l. at a depth of 3–22 m (8 m on average) below the surface. The cavity has a single entrance in a topographically higher 4◦ 7 6 W,

position (152 m.a.s.l.) and includes several main rooms that have a downward trend from the outside access to the deepest part of the cave (Fig. 1). The main cave chamber, where the microenvironmental study was performed (Polychromes Hall), is situated 60 m from the cave’s entrance, which is at a lower topographic level (146.5 m.a.s.l.) to the surrounding chambers. The rock layer over the chamber averages 7.5–8 m thickness. The host rock in the Altamira Cave is a thin to medium, parallel bedded, Cenomanian (Upper Cretaceous) limestone succession from 13.5 to 15 m thick. The overlying soil above the cave is a heterogeneous and discontinuous artificial soil with little development (30–70 cm). The soil is silicate-based and poorly differentiated (a surface horizon “A” and, directly beneath this soil, a subsurface petrocalcic horizon). A developed plant cover (meadow vegetation, C3 plants) and high organic carbon were derived from this soil (10–15%). A GIS-based geological model, which was developed using detailed Digital Elevation Models (Elez et al., 2013), provides information regarding the primary distribution of discontinuity planes: a well-marked stratification system (N18◦ , 8.5◦ E) and a primarily vertical or sub-vertical strike system. In outcrop points at the surface, these discontinuities must be direct exchange channels with the outside atmosphere through which gaseous exchange fluxes occur. In this sense, we must take into account the adjacent set of sinkholes, which are located to the east of the cave, as another possible means of direct exchange with the outside atmosphere (Garcia-Anton et al., 2013). In this geographical area, the climate is moderately oceanic and humid, with an annual precipitation approximately 1400 mm and a mean annual temperature and relative humidity approximately 14 ◦ C and 85%, respectively. Cave air is characterised by a highly stable temperature and humidity throughout the year, with an indoor relative humidity permanently near saturation and mean annual temperature near 14 ◦ C, with 1.5 ◦ C annual thermal amplitude (Cuezva et al., 2009). Relatively high levels of air CO2 are registered during winter, which sometimes exceed 5000 ppm, and lowest values near 500 ppm during summer (from June to October), due to the most effective cave ventilation during this warmer and drier period (Sanchez-Moral et al., 1999; Kowalski et al., 2008), with an almost homogeneous spatial distribution of pCO2 -air along the cave (Garcia-Anton et al., 2013). 2.2. Micro-environmental monitoring Gas and microenvironmental monitoring were performed during the warm and dry season (September 2011). Inside the cave, a micro-environmental monitoring station recorded microclimatic data in the Polychrome Hall (Fig. 1). The monitoring station was composed of an 8-channel, 16-bit datalogger (COMBILOG TF 1020, Theodor Fiedrich & Co., Germany) with a suite of probes for the following parameters: air temperature and relative humidity (HygroClip S3, which combines a Pt100 1/10 DIN temperature sensor and a humidity sensor, Rotronic), atmospheric pressure (Vaisala BAROCAP PTB 100, silicon capacitive, Finland), and CO2 concentration (ITR 498 ADOS, Germany). The station scanned each sensor every 10 s and recorded the 15-min averages. A Radim 5WP Radon monitor (SSM&SISIE-Prague) was employed to measure the concentration of radon gas (222 Rn) in the air every half hour. Outside the cave, a weather station with two autonomous dataloggers stored 15-min means of the air temperature and relative humidity (HOBO U23 Pro v2, Onset, Bourne, MA, USA), as well as solar radiation and soil temperature (at 5 and 25 cm depth) (HOBO U12, Onset, Bourne, MA, USA, equipped with the following set of probes: pyranometer PYR-SA 2,5V, Apogee instruments, TMC20-HD of Onset and ECHO EC-10 of Decagon Devices, for radiation and soil temperature, respectively). Rainfalls were registered by an autonomous

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

115

Fig. 1. (A) Altamira cave location, (B) plan view with the air sampling points and the monitoring system location and (C) the cave main profile.

RG2-M pluviometer (Onset) with 0.2 mm resolution. All records are reported in Coordinated Universal Time (UTC).

2.3. vdMonitoring of atmospheric CO2 and ı13 C–CO2 isotopic signals A continuous detailed monitoring (>118,000 records with data every 2 s) of CO2 concentrations in the air and ␦13 C–CO2 variations in the air from Polychrome Hall was performed from 18:00 h of September 19th to 07:00 h of September 21st, 2011 using a laserbased gas analyser Picarro G2101-I (California, USA) that employs cavity ring-down spectroscopy (CRDS-WS) (Crosson, 2008). Monitoring included two night periods and one daytime period (a total period of 38 h) without raining episodes. Cave air was continuously pumped (3.1 l/min at atmospheric pressure) from Polychromes Hall using a 4 mm (inner diameter) polyurethane tube. To avoid problems related to overpressure effects, an air extraction tube was connected to the analyser using a T connection. The analyser performs in situ and real-time measurements of atmospheric levels of carbon dioxide (12 CO2 and 13 CO2 ) and automatically calculates the isotopic signal of ␦13 CO2 . Precisions of 200 and 10 ppb are guaranteed for 12 CO2 and 13 CO2 , respectively, with a resulting accuracy of 0.3‰ for ␦13 CO2 after 5 min of analysis. The device was calibrated prior to field deployment using synthetic gases with known specifications.

To characterise the spatial distributions and temporal variations of atmospheric [CO2 ] and its ␦13 C signal, several surveys for discrete samplings were also performed in the cave (14 samples), outer soil (11 samples) and atmosphere (5 samples) from September 19th to 21st, 2011 (Fig. 1). Air from the soil was sampled at a 30 cm depth using an iron tube that was nailed into the ground. All air samples were taken with an air pump and saved in gas-tight Tedlar bags (1 L of volume). 2.4. Suspended aerosol monitoring An airborne particle counter (TSI Aerotrack Model 9306), which is equipped with a long-life laser diode, was installed for aerosol monitoring in Polychromes Hall. This device has a 0.1 CFM (2.83 L/min) flow rate with ±5% accuracy and counts bin sizes from 0.3 to 25 ␮m, logging up to 6 particle sizes simultaneously (0.3–0.5, 0.5–1, 1–3, 3–5, 5–10 and >10 ␮m) every minute. The size resolution is <15% at 0.5 ␮m, and the counting efficiency is 50% at 0.3 ␮m and 100% for particles >0.45 ␮m. The airborne particle counter was installed 1 m above soil surface and 5 m away the monitoring station (Fig. 1). 2.5. Statistical data analyses A time series segmentation procedure of was performed to explore multiple singularities of the CO2 signal and to apply

116

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

further modelling on a daily scale. This procedure yielded an optimal partition, from a least squares point of view, of the original series into several subseries (segments) with a linear or constant trend (Hubert, 2000). The criterion to define the change point between segments was based on the segmentation cost, which was defined as the sum of the squared deviation of the data from the means of their respective linear segments. For the offline segmentation of the long CO2 time series, an algorithm called modified dynamic programming (mDP), which is based on the “branch and bound” type technique (Gedikli et al., 2010a,b) that allowed the location of multiple change points in a time series with several thousand observations was used. Here, a 1-min averaged time series of CO2 (2281 records) was used to achieve a cost-effective segmentation with this computer code. This algorithm produces accurate segmentations in less time than previously proposed segmentation algorithms (Gedikli et al., 2008; Aksoy et al., 2008). The algorithm determines the m-order segmentation by the introduction of a constraint according to Scheffe’s test of contrast, which ensures that all differences between two contiguous means remains simultaneously significant (Hubert, 2000). In addition, the optimal number of segments was limited to avoid any short segment connecting another two contiguous and longer segments, but with an opposite trend. This simple criterion ensured a smoother segmentation of the CO2 time series, without unrealistic short-term fluctuations during periods with prevailing downward and upward trends, which were due to the signal noise itself. 2.6. Air density calculation Variations in densities of air masses were calculated by using the ideal gases equation and by assuming that the air is a mixture of water vapour and dry-air (Eq. (1)). For cave air, the CO2 concentration was taken into account because higher levels of this gas could affect the absolute value of the air density that was calculated (Eq. (2)). ext =

Ma · (P − pw ) + Mw · pw R·T

cave =

Ma · (P − pw − pCO2 ) + Mw · pw + MCO2 · pCO2 R·T

(1) (2)

where ext and cave are the densities (g/L) of the exterior air and cave air, respectively; Ma , Mw , Ma and MCO2 are the molecular masses of dry air (28.97 g/mol), water vapour (18.02 g/mol), dry air without CO2 (28.95 g/mol), and CO2 (44.01 g/L), respectively (CRC, 2003). P is the total pressure (atm), and pi refers to the partial pressure (atm) of each gas (vapour and CO2 ). R is the universal gas constant (L atm/mol K), and T is the temperature of the air, which is measured in Kelvin degrees. The vapour partial pressure (atm) was obtained from relative humidity (%) and temperature (K) using the modified Magnus-Teten correlation (Buck, 1981): HR pw = · 6.1121 · e(17.368·(T +273.15))/(17.368+(T +273.15)) 100 · 1013.25 (3) The partial pressure of carbon dioxide (atm) was obtained from the concentration of this gas ([CO2 ] (ppm)), assuming the value as a part of the whole mixture of dry air: pCO2 = 10−6 · [CO2 ] · (P − pw )

in the Altamira cave, with an outgassing stage during summertime (from June to October), as well as a gas recharge phase during wet and rainy periods (Sanchez-Moral et al., 1999; Kowalski et al., 2008), were also distinguishable in 2011. Specifically, during September 2011 (outgassing stage), the CO2 concentration in the air inside the cave (Polychrome Hall) was relatively steady, usually ranging between 700 and 1200 ppm and not exceeding 2000 ppm in any case (Fig. 2). Significant CO2 increases were observed after rainfall events (days 4–7 and 18–21). These stages of CO2 accumulation in the cavity after rain events were only observed during the intervals in which the temperature of the outside atmosphere remained well below the external soil and became smaller than that of the cavity. The 222 Rn concentration generally ranged 500–1200 Bq m−3 , reaching even 2600 Bq m−3 at one point in time in relation to the rainfall that was produced before 19 September 2011. The parallelism between the CO2 and 222 Rn fluctuations indicated the existence of an alternation phenomenon that either insulated or ventilated the cave air (Cuezva et al., 2011). Fig. 3 displays the evolution of the external meteorological parameters CO2 concentration and ␦13 C–CO2 in the cave air over the continuous monitoring period on a daily scale, where the solar radiation signal delimited the day/night alternation. Cave air [CO2 ] showed a sinusoidal pattern, with a variation range of 400 ppm (inside Polychrome Hall). The maximum concentration was reached at approximately 11:00 h of September 20th, and the minimums were achieved at approximately 21:30 h and 00:30 h of September 19th and 21st, respectively. A similar variation curve was registered for the radon signal (222 Rn), with an amplitude of 700 Bq m−3 throughout the monitoring period. These daily variations in CO2 and 222 Rn concentrations had already been observed for this cave during summer (from June to October) in previous studies (Kowalski et al., 2008; Cuezva et al., 2011). ␦13 C–CO2 presented an opposite variation pattern relative to CO2 . The temperature inside the cave maintained a constant value of 13.9 ◦ C during the entire monitoring period. The air temperature at the exterior oscillated daily within a range of 11.78 ◦ C, with a smoother effect on the soil temperature as the depth increased. The relative humidity outside the cave followed an opposite cycle to that of temperature (ranging from 65 to 95%, Fig. 3), whereas the saturation state was maintained inside the cave (>99%). The air density values showed a strong temperature dependence. Inside the cave, the density varied slightly due to pressure variations and was maintained between 1.207 and 1.215 g/L. Outside the cave, the density varied between 1.175 and 1.230 g/L, which followed an opposite pattern relative to temperature. The results of the discrete samplings from September 19th to 21st that were simultaneously performed in the cave, outer soil and atmosphere are shown in Table 1. The results of the sampling registered mean values of 2199 ppm, 8047 ppm and 420 ppm for [CO2 ] and −23.7‰, −26.5‰ and −11.3‰ for ␦13 C–CO2 isotopic signals of the cave, soil and exterior air samples, respectively. The daily variation in ␦13 C–CO2 values for the outside atmosphere was observed, with lighter values in the early hours of the morning in contrast with heavier values throughout the day. There was spatial variation in the data for the cave air, with the highest concentration and lighter values in the inner parts of the cave (Grave Hall, see Fig. 1). 3.2. Daily variations in CO2 and ı13 C–CO2

(4)

3. Results 3.1. Prevailing and daily environmental conditions In terms of carrier (CO2 ) and trace (222 Rn) gas levels, the stairstep seasonal pattern that was observed in previous annual cycles

The segmentation results that were obtained from the execution of the mDP algorithm are displayed in Fig. 4. The segmentation procedure gave reliable results (modelled lineal segments fit original data with r2 = 1); therefore, this procedure is a useful and robust tool for the preliminary analysis of the time-boundaries on a daily scale of the functional correlation between [CO2 ] and its ␦13 C–CO2 . A ninth-order linear segmentation of the CO2 time series was found to

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

117

Fig. 2. Monthly period (September 2011) of the exterior temperature (Text ), temperature of soil at 25 and 5 cm depth (Tsoil [25 cm], Tsoil [5 cm]), cave temperature (Tcave ), gases levels (CO2 and 222 Rn) in the cave air, and other variables related to the external meteorological conditions (rainfall, relative humidity). The shaded area represents the period monitored in detail (Fig. 3).

Fig. 3. Parameters variation during monitored daily cycle inside and outside the cave: relative humidity of the exterior air (RH), solar radiation (Rad), difference between the exterior and cave air densities (ext –int ), exterior air pressure (P), radon activity inside the cave (222 Rn), CO2 concentration in the air (CO2 ) and ␦13 CO2 (␦13 C) inside the cave. Represented evolution curves of ␦13 C and 222 Rn have been smoothed from the original registered data. The shaded area represents the night-time period delimited by the null solar radiation.

118

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

Table 1 CO2 concentrations and carbon isotopic signature ␦13 C of CO2 of air during the discrete samplings carried out from September 19 to 21. Origin

Point 1

Soil air Point 2

Subterranean atmosphere

Exterior atmosphere

Entrance Hall Crossing Hall of Walls Corridor Great Hall Grave Hall Well Hall Entrance Hall Crossing Hall of Walls Corridor Great Hall Grave Hall Well Hall

Sampling day

Sampling time (GMT)

19-9-11 19-9-11 19-9-11 21-9-11 21-9-11 20-9-11 20-9-11 20-9-11 21-9-11 21-9-11 19-9-11 19-9-11 19-9-11 19-9-11 19-9-11 19-9-11 19-9-11 19-9-11 21-9-11 21-9-11 21-9-11 21-9-11 21-9-11 21-9-11 21-9-11 20-9-11 20-9-11 20-9-11 21-9-11 21-9-11

15:25 15:20 16:13 7:42 9:16 12:00 8:19 17:50 7:50 9:11 16:18 15:36 15:40 15:44 15:47 15:51 15:54 16:00 8:09 8:13 8:16 8:26 8:24 8:29 8:39 7:09 11:53 17:40 7:19 9:12

CO2 (ppm) 4440.7 9846.4 9860.0 18,552.8 14,838.9 5845.3 5620.4 6378.1 4808.6 4043.2 4286.1 999.2 1547.1 1873.1 1946.4 2358.5 2658.6 2284.6 1434.3 1978.1 2171.9 2342.3 2675.1 3844.5 2677.1 419.2 393.7 399.5 491.3 399.0

12

CO2 (ppm)

4392.3 9739.2 9752.7 18,351.0 14,677.5 5781.5 5559.2 6308.6 4756.1 3999.1 4239.3 988.3 1530.1 1852.7 1925.1 2332.8 2629.6 2259.7 1418.6 1956.4 2148.2 2316.7 2645.9 3802.5 2647.9 414.6 389.4 395.0 485.8 394.5

13

CO2 (ppm)

48.4 107.2 107.4 201.7 161.4 63.7 61.3 69.5 52.4 44.1 46.8 11.0 16.9 20.5 21.3 25.8 29.0 24.9 15.7 21.6 23.7 25.6 29.2 42.0 29.2 4.6 4.4 4.4 5.4 4.4

␦13 C (‰) −25.5 −27.2 −27.3 −28.4 −28.2 −26.0 −26.0 −26.1 −25.8 −25.5 −25.5 −20.9 −22.7 −23.6 −23.9 −24.5 −25.1 −24.7 −22.1 −23.5 −23.7 −23.9 −24.2 −25.2 −24.2 −12.9 −9.3 −10.8 −12.8 −11.1

Points are referenced in Fig. 1.

match with the signal for this gas, which was registered in the study (38 h long). However, further data analyses were focused on the six segments (A → F) that defined a complete daily cycle because the segments that were located at the ends corresponded to incomplete repetitions of other daily segments, and the A1 -segment repeated at the beginning of a new circadian cycle. Each segment displayed a constant linear trend of gas concentration and, consequently, each change point could denote a modification of the degree of influence of a particular process on the increase or decrease of CO2 in the cave air. An almost identical segmentation was achieved for the isotopic signal, ␦13 C–CO2 , to that displayed in Fig. 4, except for some mismatches (time lags) that are described below, e.g., the segment-B that was subdivided in two sub-segments (B1 , B2 ) in the function of a breakpoint of the linear trend of ␦13 C–CO2 . Starting points for each segment of the CO2 and ␦13 C–CO2 signals are also listed in Table 2 and were grouped in two stages:

recharge and degassing, according to the function of the prevailing upward or downward trends. The recharge and degassing rates vary over time. The recharge stage entailed an increase in the CO2 concentration of +387 ppm and a coeval decrease in the ␦13 C–CO2 of −1.26 ‰. The highest rate of CO2 recharge into cave air matched with the time-segment B ([CO2 ] = +38 ppm/h), which primarily covered the night hours (from 01:46 h to 08:21 h). These data indicated that there was a slow and continuous supply of lighter CO2 to the atmosphere of the cave during the recharge stage. In contrast, the degassing stage produced a total depletion in the concentration of CO2 (−539 ppm), which was joined to an increase in ␦13 C in the cave air of +1.23‰. The highest gradient of gas decrease ([CO2 ] = −75 ppm/h) was detected a few hours after the beginning of the degassing stage, which matched segment E (afternoon hours between 14:00 h and 18:50 h). These data indicated that there was a faster supply of heavier CO2 to the atmosphere of the cave

Fig. 4. Daily segmentation of the CO2 concentration time series and the isotopic signal ␦13 C–CO2 , after executing the mDP algorithm (Gedikli et al., 2010a,b). Black-dashed lines correspond to each linear segment (A–F) fitting the CO2 time series and the vertical dotted lines define the time boundaries of each segment (see text).

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

119

Table 2 Variation rates of CO2 concentration and ␦13 C–CO2 during the continuous monitoring of the cave air (from 18:00 of September 19th to 07:00 September 21st of 2011). B1 and B2 are the stages of maximun recharge of CO2 . E is the stage of maximun degassing of CO2 . Segment

Start time (M/D/Y hh:mm)

[CO2 ] (ppm/h)

␦13 C (‰/h)

CO2 recharge

A B1 B2 C A1

09/19/2011 21:41 09/20/2011 01:46 9/20/2011 06:16 09/20/2011 08:21 9/21/2011 00:30

23 39 34 18 22

−0.04 −0.06 −0.23 −0.11 −0.03

CO2 degassing

D E F

09/20/2011 11:01 09/20/2011 14:00 09/20/2011 18:50

−23 −75 −19

0.00 0.21 0.01

during the degassing stage. The beginning of the degassing stage of cave coincided with the warmer hours of the outer atmosphere (maximums of air temperature and solar radiation, and the relative humidity drops below 75%) and with the highest density gradient between the cave air and outer atmosphere (ext < int , Fig. 3). Fig. 5A displays the correlation between the temporal variation of CO2 concentration in the air and its corresponding ␦13 C–CO2 signal in the Polychrome Hall during the recharge/degassing cycle, which was defined by the sequence of segments A → F. The inverse temporal correlation between the CO2 concentration in the air and ␦13 C–CO2 had two distinct and almost parallel pathways. The correlation of the daily cycle of the parameters resulted in the displacement of the points that registered at the end of the day to the left and slightly downwards compared with the initial values (Fig. 5A). Both pathways did not exactly define a daily cycle because the environmental conditions slightly changed on consecutive days. The A → C pathway corresponded to the night-time rechargestage of CO2 in the cave air. The CO2 concentration in the air progressively increased as ␦13 C–CO2 became lighter. First, (segment A) the data correlation displayed a negative trend, but with some scattering of the data. The distribution was well-defined linear, continuous and homogeneous when we focused on analysing the plenary recharge stage (segment B), which denoted a constant supply source of CO2 over time. However, the data correlation displayed an abrupt slope change (transition from sub-segment B1 –B2 , above described), which was denoted by a sudden reduction in ␦13 C–CO2 . Finally, the segment C section correlated in continuity with the B section; however, the variation ranges of the CO2 concentration and ␦13 C–CO2 were more narrow. The D–F pathway corresponded to the daytime degassing-stage of CO2 in the cave, i.e., the CO2 concentration in the air progressively decreased, whereas ␦13 C–CO2 became heavier. The CO2 concentration in the air and ␦13 C–CO2 temporal data correlation displayed a good fit to a negative trend line, but with a greater dispersion of the data regarding the gas-recharge stage, particularly within segment E. This result seemed to be indicative of both a changing source of CO2 and uneven ventilation pulses that caused the renewal and degassing of the cave air, which favoured the inlet of outer air with a lower CO2 concentration and heavier ␦13 C–CO2 signal. 3.3. Time delimitation of effective cave ventilation by aerosol monitoring A high resolution monitoring of aerosol concentrations in cave air was performed coeval to the gas monitoring to characterise in detail the alternation between the CO2 recharge and degassing by means of other ancillary variables. Fig. 6 displays the time series of size ranges for the aerosols in relation to the daily evolution of CO2 and ␦13 C. The coarsest particles, with diameters above 1 ␮m, were grouped to obtain a better distinction from the fine particles plots (0.3–0.5 and 0.5–1 ␮m). A quick experimental visit of 6 people for 10 min was completed during the gas monitoring period,

starting at midday of September 20th immediately when the beginning of the daily degassing or degassing stage was expected, in accordance with previous studies (Cuezva et al., 2011). Background levels of aerosols in the undisturbed cave corresponded with a pristine atmosphere with only a low concentration, 8 (±1) particles cm−3 , of the finest particles (0.3–0.5 ␮m) and non-significant levels of coarser aerosols. These levels were registered during the early morning hours, which matched the time segments that define the CO2 -recharge stage. Visitors walking into cave provoked the detachment of particles from the soil and their own clothes, and, therefore, a sharp rise in the concentration relative to natural backgrounds for each aerosols bin size; +84 part cm−3 (0.3–0.5 ␮m), +70 part cm−3 (0.3–0.5 ␮m) and +100 part cm−3 (>1 ␮m) was registered in only 8 min. The aerosol concentrations were reduced by up to 40% for each bin size in only 14 min, in spite of the experimental visit, which remained in the cave but was returning to the cave entrance. Once the visit left the cave, the coarsest particles (>1 ␮m) were removed from suspension, and initial background levels were restored 1 h and 48 min after the visit. The finest particles (<1 ␮m) remained suspended for a longer period than the monitoring survey. These particles still underwent two trend breaks during the recovery period, which were not detected in the coarser particles (>1 ␮m), as follows: (1) A brief resuspension of particles was registered once the experimental visit finished: +10 part cm−3 during 20 min for the coarser particles (0.5–1 ␮m) and +57 part cm−3 during 35 min for the finest particles (0.3–0.5 ␮m). The human presence in the cave reinforced the thermal stratification of the cave air due to the contribution of moist and warm air by breathing. Both a portion of the detachment particles and the humid and less-dense air parcel were displaced near the cave ceiling and, eventually, aerosols could act as nucleation points for vapour condensation. The gravitational sedimentation began to act on these remaining aerosols once the hygro-thermal disturbance disappeared and, consequently, these aerosols were removed from suspension with a certain delay, which broke the downward trend of the fine particle time series. (2) A natural breakpoint in the time series of the finest aerosols was registered 1 h and 50 min after finishing the experimental visit. This change in the downward trend allowed for the fine particles, which did not reach the initial background levels. The resuspension, or the influx, of an air parcel with a higher airborne particle concentration was denoted by an upward trend of the finest aerosols (Fig. 6), which was coeval with the degassing stage (segments E → F). 4. Data analysis and discussion 4.1. Evidence of cave-air CO2 sources Identifying the source of CO2 at every time point is essential for determining the dynamics of this gas in the cave air. In this regard,

120

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

Fig. 5. (A) CO2 concentration of the cave air and its ␦13 C–CO2 (5-min interval record) throughout a complete daily cycle defined by the sequence of segments A → F (see text). (B) Keeling plot for the discrete sampling in soil (black circles), exterior (black squares) and inside the whole cave (grey circles) carried out from September 19th from 14:47 up to 16:18 and 21st from 7:19 to 9:16. (C) Keeling plot obtained for the cave air data (grey circles) during the recharge period and the soil samples (black circles). (D) Keeling plot obtained for the cave air data (grey circles) during discharge period and the exterior samples (black circles).

it is useful to apply the “Keeling plot” method for the analysis of the ␦13 C–CO2 isotopic signal (Keeling, 1958) to better understand the processes that control the ecosystem isotope discrimination (Pataki et al., 2003). This two-ended mixing model determines, by a linear

regression approach, the carbon isotopic signal of the CO2 “sources” of a specific ecosystem that contributed to increases in atmospheric CO2 . The CO2 concentration of the cave air should be the result of mixing the background atmospheric CO2 with soil-produced carbon dioxide, according to the Keeling model (Garcia-Anton et al., 2012): cave-mixed = exterior-background + soil-source. The role of carbonate host rock dissolution as a contributing source of CO2 (Serrano-Ortiz et al., 2010) has been considered negligible here due to the almost null rates of dripping water inside the cave during the season in which the study was developed. To determine the carbon isotopic signal of the CO2 “source” of the Altamira cave ecosystem, we have employed data results from the discrete sampling of September 19th (from 14:47 h to 16:18 h) and 21st (from 7:19 h to 9:16 h), which were performed in the cave, outer soil and exterior atmosphere (Table 1). Exterior (as the background), soil (as the source) and cave air samples (as the mixed component) fit the Keeling model with a high degree of correlation (r2 = 0.99, see Fig. 5B). Within this linear correlation, the data for the cave air are extremely close to soil air values but farther from the exterior air data. The yintercept value (−27.32‰) indicates a dominant component of CO2 that originated from soil organic respiration, which is in accordance with averaged ␦13 C–CO2 data of −27‰, which are the net results of the combination of equilibrium and kinetic fractionations that occur during the photosynthesis of C3 plants and of CO2 that is derived from the decomposition of C3 biomass (Cerling et al., 1991; Amundson et al., 1998). Accordingly, soil-produced CO2 reaches the cavity, which increases the CO2 concentration and reduces the isotopic signal ␦13 C–CO2 . The lowest CO2 concentration and heaviest isotopic signal are registered in the shallower and more exteriorinfluenced zones inside the cave (i.e., the Entrance Hall in Fig. 5B and 1), whereas the opposite pattern characterises the zones with a greater degree of isolation (i.e., the Grave Hall in Fig. 5B and 1). The Keeling model has been used to explore the correlation between air masses that are involved in each daily-exchange process (recharge and degassing). For this propose, the Keeling model has been applied to the ␦13 C–CO2 isotopic signal of the cave-air and compared with the external air mass that dominates the process in each instance (soil or exterior atmosphere). For the recharge period, the Keeling regression has been obtained for each group of samples, which were differentiated depending on origin (soil and cave, Fig. 5C). The regression functions for each group of samples tend to converge at the same y-intercept (␦13 C–CO2 ); however, the estimated characteristic isotopic signal for the CO2 -source is slightly lower for cave samples (−29.57‰) than for soil samples (−28.96‰). The above data point directly to the fractionation process due to diffusion as the dominating mix process during this recharge stage. The Keeling plot has also been obtained for the set of measurements that registered inside the cave during the degassing stage and were compared with the points representing the exterior air (Fig. 5D). Exterior values are aligned with the regression line for cave air data. Thus, for the mean value of the exterior concentration of CO2 (420.5 ppm), there is an absolute error of −0.63 ‰ (relative error of 5.6%) between the mean actual ␦13 C (−11.34‰) and the estimated value using the Keeling regression for cave measures (−11.98‰). The correlation that was observed supports the model in which the exterior air is one of the two masses that affects the characteristics of the cave air (background and exterior). ␦13 C–CO2 of the y-intercept is heavier for cave air samples during the degassing stage (−28.37‰) than that of the recharge period (−29.57‰), which points to a heavier source for the degassing stage than for the recharge period. 4.2. Suspended aerosols and air movement in the cave Aerosol is a preferable environmental monitoring indicator that provides equal or greater sensitivity over more commonly used

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

121

Fig. 6. Time series of the airborne particles (aerosols) for each size range relative to the CO2 concentration of cave air and the ␦13 C–CO2 . The contribution and particles re-suspension by a short experimental visit and the time segments are displayed (see text).

indicators, such as cave air CO2 and temperature. Cave ventilation has been confirmed as one of the key processes that control cave aerosol introduction, transport, size distribution and deposition (Dredge et al., 2013). Christoforou et al. (1996) determined that cave ventilation and its air flow patterns act as the primary mode of transport for pollutant particles into caves from outdoors. In this study, the aerosol monitoring results (Fig. 6) allowed us to perform an accurate definition of the time-boundaries for the daily air flow that is responsible for the CO2 -degassing of cave air. The lowest values for suspended particles register from the beginning of the survey to the experimental visit disruption. No variations in measured aerosols register, whereas CO2 concentration progressively increase and ␦13 C–CO2 decrease. A lighter CO2 input by diffusive transport should prevail during the gas-recharge stage because no volumetric air movement has been detected by the monitoring aerosols. Thereafter, a natural increase in the finest particles levels (0.3–0.5 ␮m) registers in the cave air immediately after the experimental visit and coincides with the CO2 -degassing stage. From midday (September 20th, 12:00–13:00 h GMT), the concentration of the finest particles displays an opposite trend to that of CO2 concentration and to a coeval evolution with its carbon isotopic signal (␦13 C–CO2 ). The averaged ratio increment of the finest particles during the degassing stage is +36 particles cm−3 h−1 , and this ration continues to increase until midnight, at which the minimum CO2 and maximum ␦13 C values register (Fig. 6). Particles between 0.5 and 1 ␮m display a softer rise but always remain above the background levels. Therefore, degassing entails an air influx from the outer atmosphere, which causes a noteworthy increase in levels of the finest particles (<1 ␮m), particularly in those particles with a diameter size below 0.5 ␮m. Likewise, the breakpoint in the recovery trend of the finest particles after the experimental visits accurately indicates the influence of the air flow, which does not allow the cave atmosphere to return to the background levels, which were registered before the anthropogenic disturbance. 4.3. Modelling of CO2 -recharge and degassing processes The CO2 concentration of the cave air is determined by the mixing pattern of the three air masses that are involved in the cave–soil–exterior system at each time point. The recharge stage would involve the input of soil-derived CO2 .The gas that results

from organic respiration in the soil continuously migrates to lower concentrated areas. Consequently, the gases that fill the entire freespaces system of the host rock beneath the soil must tend to reach a near-equilibrium state. The cave has a higher air interchange rate with the exterior atmosphere than that of the smaller host rock spaces. Thus, the gas concentration in the air of the cave tends to balance with the gas concentration in the air that fills the free-space system of the host rock. Therefore, light CO2 that is generated in the soil finally reaches the cavity and modifies the atmosphere that initially existed (background). The degassing stage would involve air-input from the outer atmosphere and mixing with cave air (background). Thus, when the ventilation of the cave air occurs, the imbalance in the CO2 concentration of the cave air with the immediately adjacent host rock spaces increases. Then, the diffusive flux of CO2 through the host rock spaces becomes more efficient. With the aim to quantify and characterise CO2 variations inside the cave, and considering the evidence described above, models that are based on the mass balance (Frisia et al., 2011) have been separately developed for recharge and degassing, assuming that diffusion dominates the first stage and that advection dominates the second stage. Other biological factors that control the CO2 flux into the cave air have been considered, but not included, in the mass balances because these factors do not exert a high degree of control compared with gas diffusion or advection. These air mixing processes essentially differ in that advection is a volumetric flux (including air-mass flux), whereas diffusion exclusively implies a mass flux (Corey et al., 2010). A volumetric flux, then, will depend on several physical parameters, primarily including the temperature, density, and pressure differences between air masses, whereas a mass flux depends on chemical parameters, i.e., the concentration and/or partial pressure differences of a specific gas between air masses. Cave air properties will primarily result from the interaction of both processes, and these processes can be modelled from the characteristics of the air masses that are involved. The basic equations for both processes are then formulated; however, more extensive derivations of diffusion and advection equations are shown in the Appendix. The application of segmentation procedures to the CO2 time series has enabled the discrimination between different time intervals (segments) within which the CO2 trend is constant. Segments with slow rates of

122

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

CO2 variation could denote the simultaneous influence of both mechanisms: diffusion and advection. Therefore, for a better understanding of each individual mechanism, we have focused the analysis exclusively on the segments (B and E) with the highest variation rates, considering the diffusion mechanism as dominant in segment B and advection as dominant in segment E. The variation (expressed as standard deviation) of the detrended time series of the CO2 signal is ±4.72 ppm during segment E, with prevailing advection and ±3.77 ppm during segment B (diffusion). This diverse CO2 -variability between segments denotes the different operation of each process, i.e., a laminar flux for the CO2 -diffusion of soilderived gas air to the cave air and, in contrast, a certain trend to a turbulent flux during air mass movements between the exterior and the cave. 4.3.1. Diffusive recharge stage The CO2 concentration inside the cave increases due to a continuous gas-inflow of soil-produced CO2 . According the diffusion definition, the model assumes no volumetric variations during the process (Appendix A). Therefore, the CO2 concentration of the cave air is a time-dependent parameter, and its variations can be expressed as follows: [CO2 ]c (t) = [CO2 ]b + [CO2 ]s (t)

(5)

where, [CO2 ]c (t): concentration of CO2 inside the cave (ppm) for each time t; [CO2 ]b : background concentration of CO2 inside the cave at the beginning of the process (ppm); [CO2 ]s (t): input of soilproduced CO2 to the cave air, which is dependent on the time (ppm) that resulted from the diffusive flux. [CO2 ]s represents the mass of incoming soil-produced CO2 to the cave. Therefore, the isotopic signal of CO2 -inflow does not represent the isotopic signal of the soil-produced CO2 but the isotopic signal of the incoming component, i.e., the isotopic signal of CO2 inflow is fractionated with respect to the original value according to the diffusion definition. Regarding the correlation of isotopes, the value for the resulting air-mixture that exists in the cave follows the proposed model (see Appendix A): ı13 Cc (t) =

([13 CO2 ]b + [13 CO2 ]s (t))/([12 CO2 ]b + [12 CO2 ]s (t)) − std std × 1000

(6)

␦13 Cc (t):

where isotopic signal of CO2 inside the cave, which is dependent on time (‰); [13 CO2 ]b : background concentration of 13 CO inside the cave at the beginning of the process (ppm); 2 [12 CO2 ]b : background concentration of 12 CO2 inside the cave at the beginning of the process (ppm); [13 CO2 ]s (t): input of soil-produced 13 CO to cave air, which is dependent on time (ppm); [12 CO ] (t): 2 2 s input of soil-produced 12 CO2 to cave air, which is dependent on time (ppm); std: international reference standard PDB equal to 0.0112372 (Craig, 1957). Two sub-segments were distinguished within segment B as a function of the ␦13 C–CO2 time series: B1 and B2 (Fig. 6). Although the CO2 flux remains nearly constant within segment B, the change that is observed in the slope of ␦13 C indicates a variation in the ratio of isotopologue flows (transition B1 → B2 , Fig. 4). The increase

in ␦13 C observed in B2 (0.23‰/h) compared with B1 (0.06‰/h) indicates an increase in the 12 CO2 flux compared with 13 CO2 (Table 2).). The isotopic signal of the soil-produced CO2 component (␦13 Cs ) can be obtained from the application of Eq. (5) for the two isotopologues: ı13 Cs (t) =

([13 CO2 ]c (t) − [13 CO2 ]b )/([12 CO2 ]c (t) − [12 CO2 ]b ) − std std × 1000

(7)

where [13 CO2 ]c (t): concentration of 13 CO inside the cave, which is dependent on time (ppm); [12 CO2 ]c (t): concentration of 12 CO inside the cave, which is dependent on time (ppm). From Eqs. (5) and (7) a CO2 -input of 173.14 ppm can be calculated from the soil-source, with an isotopic signal of −27.71‰ in sub-segment B1 , whereas, during sub-segment B2 , the CO2 -input reached 70.27 ppm from a source with ␦13 C of −35.96 ‰ (Table 3). The light isotopic signal confirms the organic origin of the source from soil, and the difference between the ␦13 C–CO2 -source for B1 and B2 suggests a change in the isotopic signal of the CO2 coming into the cave. Observed differences may be due to a non-ideal steady-state diffusive gas transport in which kinetic effects (Risk and Kellman, 2008) or “enhanced diffusion” processes (Bowling and Massman, 2011) can be affecting, enhancing or decreasing (respectively) diffusive isotope fractionation. In steady-state systems, the molecular diffusion of CO2 is characterised by isotopic fractionation due to the slightly greater movement capability of the lighter isotopologue, i.e., the diffusion favours 12 CO2 particle movements compared with 13 CO2 . The ratio between the two diffusion coefficients (12 D/13 D), which is normally used in the literature, is obtained from Graham’s equation, and this ratio is quantified as 1.0044 (Craig, 1953; Cerling et al., 1991). The ratio of diffusion coefficients of the isotopologues can be obtained from the registered data using Fick’s law of diffusion (see Appendix D for details) as follows: 12 D 13 D

cs

cs

=

12 C

12 C t − t−1 12 C − 12 C s m

·

13 C 13 C

t

− 13 Cm − 13 Ct−1

s

(8)

where Ct : CO2 concentration inside the cave at time t (ppm); Ct − 1 : CO2 concentration inside the cave at previous time t − 1 (ppm); Cs : soil CO2 concentration (ppm); Dcs : effective diffusion coefficient Dcs from the soil to the cave (cm2 s−1 ); Cm : mean concentration of CO2 inside the cave during the studied interval (ppm). Table 3 summarises the parameters that have been used to calculate the ratio of diffusion coefficients of the isotopologues to simulate the CO2 -gas inflow from the soil to the cave by a dominant diffusive mechanism during the prevailing recharge stage (segment B). Calculations have been performed by using concentrations in the cave air during the studied period (segment B) and the average values that have been obtained for the soil air during the entire campaign (Table 1). The ratio of effective-diffusion coefficients for isotopologues has been obtained using parameters for sub-segments B1 and B2 (Table 4). The values that have been obtained for 12 Dcs /13 Dcs ratio agree with the theoretical diffusion principle that flux, to the extent that flux depends on the capability of particle movements, is greater

Table 3 Parameters used to calculate the ratio of CO2 -diffusion coefficients of the isotopologues during the prevailing recharge stage (segment B). Sub-segment

Point

Time

CO2 (ppm)

␦13 C (‰)

12

B1

Bo B1 f B2 o Bf

9/20/11 1:51 9/20/11 6:16 9/20/11 6:21 9/20/11 8:21

1654.45 1827.59 1828.48 1898.75

−23.5 −23.9 −23.9 −24.4

1636.49 1807.76 1808.65 1878.16

B2

CO2 (ppm)

13

CO2 (ppm)

17.96 19.83 19.84 20.59

␦13 C–CO2 -source −27.71 (‰) −35.96 (‰)

Suffixes “o” and “f” represent the start and end points of each sub-segment (B1 and B2 ) and their corresponding CO2 values represent the Ct-1 and Ct values of Eq. (8) (see text), respectively.

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

123

Table 4 Ratios of effective-diffusion coefficients for the isotopologues 1 2 CO2 and 13 CO2 (12 Dcs /13 Dcs ) after applying Eq. (8) (see text) with parameters from Table 3 Cm : mean concentration of CO2 inside the cave during the studied interval (ppm), and Cs : soil concentration (ppm). Location

Segment

CO2 (ppm)

␦13 C (‰)

12 12

Cave Cave Cave Soil

B (average) B1 (average) B2 (average) Campaign average

1776.60 1741.02 1863.62 8047.32

−23.9 −23.7 −24.2 −26.5

for the lighter isotopologue. Segment B shows 12 Dcs /13 Dcs ratios that are greater than the unit, i.e., 12 Dcs > 13 Dcs (Corey et al., 2010). To know the goodness-of-fit of the created model, an evolution curve has been simulated from Eqs. (5) and (6), according to the results that have been obtained for segment B for the daily cycle, and then compared with actual data (Fig. 7A and B). A Least square regression has been used to obtain the time-depending expressions of [12 CO2 ]c , [13 CO2 ]c and [CO2 ]c , from which parameters have been obtained to develop the model. The two dataset results highly correlated (r2 = 0.91 comparing actual and simulated ␦13 C, Fig. 7B). Actual and simulated values have been represented in a Keeling plot (Fig. 7A). Notably, in contrast to the Keeling approach, the data representation does not adjust to a straight line, but its left extreme tends to curve to the more negatives values of ␦13 C. As the model demonstrates, that fact is due to the variation in the isotopicsignal of the soil component contribution, which is linked to the influences of the kinetic effect or “enhanced diffusion” process. Of note, estimations of the isotopic signal of the source using the Keeling model may lead to certain mistakes in diffusive systems with a non-stable source of CO2 because (1) the calculated value refers to a fractionated value of ␦13 C and not to the original isotopic signal from soil; (2) kinetic effects may cause an underestimation of the actual value (Risk and Kellman, 2008); and (3) enhanced diffusion may cause an overestimation of the actual value (Bowling and Massman, 2011). 4.3.2. Advective degassing stage CO2 -degassing of the cave air is controlled by the air movement that produces the inflow of the exterior air that is joined to an output of the cave air (Appendix B). Volumetric mixes of air parcels with different characteristics (including CO2 concentration and its ␦13 C) are assumed: [CO2 ]c (t) = [CO2 ]b + Ve (t) · ([CO2 ]e − [CO2 ]b )

(9)

where [CO2 ]c (t): concentration of CO2 inside the cave (ppm) for each time t; [CO2 ]b : background concentration of CO2 inside the cave at the beginning of the process (ppm); [CO2 ]e : concentration of CO2 in the exterior atmosphere (ppm); Ve (t): total volume of exterior air input into the cave, which is dependent on the time (relative volume) that results from the advective flux (or ventilation rate) between the exterior and the cave. Accordingly: ı13 Cc (t) =

Cm (ppm) – cave Cs (ppm) – soil

1757.32 1722.13 1843.40 7960.24

13 13

12

Cm (ppm) – cave Cs (ppm) – soil

19.27 18.89 20.21 87.08

1.0034 1.0004 1.0091

is registered for the cave air before advection is triggered (␦13 C background). Segment E is considered the period that is more affected by a net dominant advective flux, according to the highest variation rates of CO2 and ␦13 C–CO2 during the degassing stage (Fig. 4) and to the revealing natural resuspension of the finest airborne particles that are observed during aerosol monitoring (Fig. 6). Therefore, the degassing stage for segment E has been simulated by applying the advective model that has been proposed (Eqs. (8) and (9), Appendix B). The exterior concentration of CO2 and its isotopic signal have been obtained as the average of the samples that have been collected in the exterior during the entire campaign (Table 1). Background parameters have been taken as cave air characteristics at the beginning of the segment E. The time evolution of ␦13 C–CO2 of the cave air has been estimated according to Eq. (10), and the function of Ve has been calculated from the CO2 concentration data. Actual and simulated data have been plotted using the Keeling model (Fig. 7C). Modelled values correlate well with the observed situation (r2 = 0.94 for actual and simulated ␦13 C), which confirms that advection prevails during segment E (Fig. 7D). The mass balance assumes that the cave air is the result of a mix of air masses. The mass balance model has been separately applied for CO2 , 13 CO2 and 12 CO2 concentrations within segment E (Table 5). The results that have been obtained for the rate and percentages of the exterior air that come into the cave environment could be considered approximately equivalent, except for a small difference between isotopologues. Nevertheless, the differences that have been obtained between inflow volumes for the two isotopologues have been enhanced using a relative parameter, Rde (expressed as ‰). For the degassing stage by advection, Rde is calculated as follows: Rde =

Ve (12 C) − Ve (13 C) × 1000 Vet (12 C)

(11)

Ve (12 C): relative volume of the exterior air inlet that has been calculated for 12 CO2 ; Ve (13 C): relative volume of the exterior air inlet that has been calculated for 13 CO2 ; Vet (12 C): value of relative volume of the exterior air inlet that has been calculated for 12 CO2 at the end of interval E. The evolution of Rde shows a slight decrease with time within segment E (Fig. 8), which indicates a greater effective ventilation for

([13 CO2 ]b (t) + Ve (t)([13 CO2 ]e − [13 CO2 ]b ))/([12 CO2 ]b (t) + Ve (t)([12 CO2 ]e − [12 CO2 ]b )) − std × 1000 std

where, [13 CO2 ]e : concentration of 13 CO2 in the exterior air (ppm); [12 CO2 ]e : concentration of 12 CO2 in the exterior air (ppm). This model for the degassing stage implies several assumptions: (1) the value of Ve should be equal to that calculated for each gas species that composes the air mass if only advection between the exterior air and the cave environment is occurring; (2) the volume of the cave-air output that results from advection is equal to the input air from the exterior (Ve = Vout ); and (3) the CO2 concentration of the output air is equal to the concentration inside the cave at the beginning of the advection process (concentration background), and, therefore, the isotopic signal corresponds to the value that

Dcs /13 Dcs

(10)

heavier isotopologue than expected (Ve (13 C) > Ve (12 C)). Therefore, during the predominantly advective daytime (segment E), the CO2 Table 5 Percentages (Ve (t)) and rates (Ve (t)/dt) of incoming air from exterior to cave during the prevailing degassing stage by advection (segment E). Results are obtained from concentrations of CO2 and its isotopologues (see text). Parameters used

[CO2 ]

[12 CO2 ]

[13 CO2 ]

Total exterior air incoming (%) Increase of exterior air per hour (%/h)

24.77 5.214

24.77 5.214

24.76 5.212

124

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

Fig. 7. Comparison of measured and simulated data using models created for diffusion and advection processes for segment B (figure sections (A) and (B)) and segment E (figure sections (C) and (D)). A and (C) are the Keeling plots of the data sets. (B) and (D) correlate the measured with simulated ␦13 C from models.

composition of the cave air is not exactly the result of the advective mix process between the exterior and inner air; i.e., the CO2 concentration of the cave air is not only controlled by a bulk mix process between two air masses (cave background and outer atmosphere). This fact indicates that during the cave degassing stage, the soil-produced CO2 diffusion continues to contribute to the cave air, which enhances 12 CO2 compared with 13 CO2 . Consequently, 13

ı Cc (t) =

and (8) and by considering [CO2 ]b and [CO2 ]e as constant values (see Appendix C for details): [CO2 ]c (t) = [CO2 ]b + [CO2 ]s (t) + Ve (t) · ([CO2 ]e − [CO2 ]b − [CO2 ]s (t)) (12) where [CO2 ]b represents the concentration of CO2 inside the cave before the two processes interact in the same time interval. Accordingly, the time evolution of the isotopic signal inside the cave (␦13 Cc (t)) is calculated as follows:





([13 CO2 ]b + [13 CO2 ]s (t) + Ve (t) · ([13 CO2 ]e − [13 CO2 ]b − [13 CO2 ]s (t)))/([12 CO2 ]b + [12 CO2 ]s (t) + Ve (t) · [12 CO2 ]e − [12 CO2 ]b − [12 CO2 ]s (t) ) − std std

a dynamic model that integrates both processes (soil-produced CO2 diffusion and advective interchange with the exterior air) is required and is defined as a function of time by combining Eqs. (5)

Fig. 8. Evolution of CO2 concentration of the cave air and the ␦13 CO2 , compared with the estimated percentage of exterior-air volume inflow (Ve , %) and relative difference between estimated volumes for the isotopologues (Rde ;‰).

× 1000

(13)

4.3.3. Daily CO2 balance and net fluxes The CO2 balance in the cave air, and specifically in the Polycrome Hall, during the study period has resulted in a daily loss of 152.21 ppm and in a decrease of 0.15‰ for ␦13 C–CO2 , which has been counted within the sequence of segments A → F (roughly 27 h). These results indicate a daily carbon imbalance in both CO2 concentration and in isotopic signal. According to the mass balance that has previously been described, CO2 -exchange fluxes can be quantified. An increase in CO2 inside the cave during the recharge period has been assumed as an inflow of soil-derived CO2 . The total amount of CO2 contribution from the soil can be calculated based on the ideal gas equation (PV = nRT) and using the mean values of pressure and temperature that have been registered during the recharge period (1004.89 mbar and 13.93 ◦ C, for cave air) and an average high of 1.13 m for Polychrome Hall (Elez et al., 2013) to estimate the volume of this subterranean atmosphere. The characteristic ␦13 C of the inlet CO2 has been calculated using the correlation between increases in isotopologues concentrations. Calculations have quantified the total inflow of CO2 as 717.34 milligrams per unit of cave volume in 13 h and 19 min (duration of gas-recharge stage, i.e., the sequence of segments A → C). This inflow results in a flux of 0.38 ␮mol of CO2 per unit of cave surface (m2 ) and time (s) during the recharge stage. The average degassing flux (ventilation rate by advection) represents the quantity of CO2 that evacuates from the cave to the exterior atmosphere per unit of cave surface and per second. In this instance, the average pressure and temperature that have been used are those values that have been registered during the

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

degassing stage, i.e., 1001.76 mbar and 13.95 ◦ C for the cave air. Characteristics of the output air (CO2 and ␦13 C–CO2 ) are those values that have been registered at the beginning of the process (segment D). According to results, the total amount of CO2 that has been incorporated into the atmosphere is 1299.83 milligrams per unit of cave volume in 13 h and 24 min (duration of the degassing stage, i.e., the sequence of segments D → F), which results in a flux of 0.69 ␮mol of CO2 per unit of cave surface (m2 ) and time (s).

5. Conceptual model and concluding remarks The Altamira cave can be considered a significant example to model gas transfer processes. This cavity acts as a vadose macropore, which is separated from the atmosphere by the soil layer and host rock. The insulating door in the sole cave entrance is permanently closed, and, therefore, the gas transfer between the subterranean and the external atmosphere primarily occurs through these two media, soil and rock. Soil could be considered an isotropic medium in which gas transport is determined by the pore size distribution, inter-particle porosity, and water content. In these sense, as we ascertained in a previous study (Cuezva et al., 2011), the Altamira overlying soil has the ability to limit gas exchange between the cavity and outside atmosphere, which depends on water content not only in the sealing of the pores through water saturation from rainfalls but also through adsorption and condensation of vapour in the porous system. This soil layer acts as a permeable/impermeable membrane, which is open to gas exchange during dry summer days with low atmospheric relative humidity, but is closed at night (and during winter) due to the presence of water within soil pores. The loss of air-filled porosity (by water saturation) is primarily produced in the surface zone of the soil, which is in direct contact with daily variations in humidity and temperature. In contrast, the gas transport through the host rock is primarily determined by the presence of bedding planes, fractures, fissures, joints and karstic voids. The free movement of air through these spaces is only limited when are filled by water. During summer, this situation only occurs in moments when rainfall occurs. Therefore, the host rock acts as an anisotropic permeable membrane or as an interface for gas transfer during the dry summer period (Cuezva et al., 2011). The pathway and direction of gas transfer are conditioned by the location and spatial distribution of all these host rock spaces. The mechanisms that control the gas transfer between a vadose environment (e.g., cave) and the external atmosphere vary throughout the day, to the extent that the climate-driven forces do (i.e., primarily the meteorological and soil conditions). Diffusion is a molecular motion flow that is driven by concentration gradients. The diffusive flux of soil-derived CO2 to the cave directly depends on soil production and diffusivity, which are controlled over time by the atmospheric conditions that regulate organic activity in soil and pore-space water saturation. Advection is a bulk motion flow that is primarily driven by an imbalance of pressure forces (air pressure gradients between cave and atmosphere). An advective flux, which injects the outer air to the cave air and extracts subterranean air, is strictly due to cave-exterior barometric pressure gradients and can be triggered either by local temperature differences (thermal buoyancy by different densities in warm and cold air), by wind (Venturi effect) or by a combination of both. Exterior atmospheric conditions are directly and indirectly driving both processes because diffusion depends on organic activity, advection depends on air density (or pressure) gradients, and both of them depend on the degree of water saturation of the host rock system of pores and fissures. Moreover, the cave morphology and its location relative to exokarst, as well as host rock discontinuity

125

configurations, determine the gas flow directions (Ogretim et al., 2012). As a result, the CO2 gas movement occurs by the combination of advection and diffusion processes; however, a particular mechanism of gas transfer prevails for each daily stage (Fig. 9). 5.1. Recharge night-time process The primary trigger mechanism for the daily closure of the soil membrane is the gradual increase in RH during nightfall. This increase induces the adsorption and condensation of atmospheric water vapour in the porous system of the overlying soil surface, which hinders the gas transfer that becomes trapped in cavities. The loss of air-filled porosity gradually increases with the increase in RH. The pore system is then finally closed, which stops cavity ventilation and blocks the gas exchange between the cave and exterior. Therefore, the soil acts, in this instance, as an impermeable membrane to gaseous exchange (Cuezva et al., 2011). A CO2 concentration gradient between deeper soil zones and karst–epikarst usually exists and, consequently, a downward CO2 diffusive flow is established between both environments. The absence of ventilation results in a constant and progressive increase in CO2 concentration in the cavity. Accordingly, this night-time CO2 recharge entails a decrease in the vadose ␦13 CO2 . Therefore, for the daily cycle that was analysed in this study, an increase of 25% for [CO2 ], which was accompanied by a decrease of 5.4% for isotopic signal are registered during this recharge stage. 5.2. Degassing daytime process A progressive depletion of the concentration of CO2 inside the cavity occurs. A daytime loss of water by evapotranspiration opens the soil pore space to gas exchange and allows gas outflow to the atmosphere. This event usually occurs when the outside relative humidity drops below approximately 75%, which is also in agreement with previous results in this cave (Cuezva et al., 2011). The outside air temperature is higher than the air temperature inside and, subsequently, the cave air density is higher than the outside air density (Text > Tint, ext < int ), which favours advective air movement through the host rock discontinuities that are located at lower levels and that are associated with existing sinkholes. This daytime venting process entails a decrease in the vadose ␦13 CO2 ; thus, for the daily cycle that was analysed, a decrease of 28% for [CO2 ], which was accompanied by an increase of at least 5% for the isotopic signal, was registered during this degassing stage. In addition, an unmistakable sign of the outside air influx by advection is the natural rise of airborne finest particles (particles with diameter <0.5 ␮m increase with a rate of +36 particles cm−3 h−1 ) from background conditions of cave air with an approximately null concentration of aerosols. On windy days, the process can be intensified by the vertical ventilation due to a Venturi effect (Roland et al., 2013). Simultaneously, the diffusion process continues even during this stage, to the extent that the soil–cave concentration gradient also exists. This mechanism is, however, less efficient than advection, and, therefore, degassing occurs. Finally, a daily carbon imbalance was observed in both CO2 molar fractions and the ␦13 CO2 isotopic signal. In global daily computing, the CO2 balance into cave air resulted in a loss of 152 ppm and a decrease of 0.15‰ for the ␦13 CO2 isotopic signal. This result was achieved after a cave-recharge of 387 ppm of isotopically light CO2 (␦13 C characteristic of −29.17‰) by diffusion from the soil, and one subsequent cave-degassing resulted in an output of at least 703 ppm of CO2 (with ␦13 C characteristic of −24.82‰) as a result of an input of exterior air with a total contribution of CO2 , which was estimated to be 164 ppm (assuming a constant value of the isotopic signal in −11.34‰). This result is ultimately a short-term

126

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

Fig. 9. Graphic representation of the proposed model with a day–night alternation of the processes affecting CO2 concentrations inside Altamira related to changes on the external conditions. The environmental values correspond to: minimum (night) and maximum (day) temperature at exterior and soil (5 cm depth); reached levels of CO2 and its corresponding ␦13 C inside the cave during recharge (night) and degassing (day); mean values of CO2 and ␦13 C of collected samples in soil and exterior air (Table 1) during ante meridiem (night) and post meridiem (day-time) hours.

significant contribution of isotopically light CO2 from the vadose environment to the troposphere, which reinforces the need to consider these particular terrestrial ecosystems in the carbon balance at different spatial and temporal scales.

[CO2 ]c = [CO2 ]b + [CO2 ]s

(A2)

Regarding to the isotopic signal, by the application of Eq. (A2) to the isotopologues concentrations of the cave air ([12 CO2 ]c [13 CO2 ]c ) and considering the definition of the isotopic signal (␦13 C):

Acknowledgements This research was supported by the Spanish Ministry of Science and Innovation, project CGL2010-17108 and CGL2011-25162. E.G.A. was supported by a CSIC JAE-Predoctoral grant. S.C. was funded by a postdoctoral fellowship from the Spanish Ministry of Science and Innovation, research programme Juan de la Cierva. All from the Altamira Cave Research Centre and Museum staff are acknowledged for their collaboration throughout the research period. Dr. Gedikli of Istanbul Technical University, for sharing his software of automatic segmentation algorithm online.

Appendix A. Appendix A. Diffusive model Assuming that the cave air is a two mixed component, the relationship between an initial component (background) and a source component (soil air) gives the concentration of CO2 inside cave. From a simple volumetric mass balance: Vc · [CO2 ]c = Vc · [CO2 ]b + Vc · [CO2 ]s

measurement unit). Assuming no volumetric variations during the process:

(A1)

where [CO2 ]c is the concentration of CO2 inside the cave (ppm), [CO2 ]b is the background concentration of CO2 inside the cave at the beginning of the process (ppm), [CO2 ]s is the input of soil-produced CO2 (ppm) and Vc refers to the cave air volume (in any volumetric

ı13 Cc =

([13 CO2 ]b + [13 CO2 ]s )/([12 CO2 ]b + [12 CO2 ]s ) − std × 1000 std (A3)

where ␦13 Cc represents the isotopic signal of the CO2 inside the cave (‰), [13 CO2 ]b and [12 CO2 ]b are respectively the background concentration of 13 CO2 and 12 CO2 of the cave air at the beginning of the process in ppm, while [13 CO2 ]s and [12 CO2 ]s are the input of soil-produced 13 CO2 and 12 CO2 (ppm) respectively. Std is the international reference standard PDB equal to 0.0112372 (Craig, 1957). According to the model, the isotopic signal of the soil component can be obtained to know the contribution of the soil component to the isotopic signal of the cave air, as follows: (A2) and (A3) can be rewritten as time dependent expressions [CO2 ]c (t) = [CO2 ]b + [CO2 ]s (t) ı13 Cc (t) =

(5)

([13 CO2 ]b + [13 CO2 ]s (t))/([12 CO2 ]b + [12 CO2 ]s (t)) − std std × 1000

(6)

[CO2 ]s (t), [13 CO2 ]s (t), and [12 CO2 ]s (t) are the total input of soilderived CO2 , 13 CO2 and 12 CO2 to cave, which are dependent on

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

time (ppm) as the result from diffusive fluxes.

Appendix C. Appendix C. Integrated model

([13 CO2 ]c (t) − [13 CO2 ]b )/([12 CO2 ]c (t) − [12 CO2 ]b ) − std

ı13 Cs (t) =

std × 1000

127

(7)

where [13 CO2 ]c (t): concentration of 13 CO inside the cave, which is dependent on time (ppm); [12 CO2 ]c (t): concentration of 12 CO inside the cave, which is dependent on time (ppm).

A model in which the diffusion and the advection processes are simultaneously affecting cave air can be obtained by combining Eqs. (A2) and (B3), assuming that advection induces the movement of the cave air affected simultaneously by diffusion of soil-produced CO2 . Accordingly, the cave background component ([CO2 ]b ) in Eq. (B3) is replaced by the expression defining the cave concentration component ([CO2 ]c ) in Eq. (A2):

Appendix B. Appendix B. Advective model

[CO2 ]c = [CO2 ]b + [CO2 ]s + Ve · ([CO2 ]e − [CO2 ]b − [CO2 ]s )

During advection the cave air is the resulting air-mixture of the initial subterranean atmosphere (CO2 -background before the

Here [CO2 ]b represents the concentration of CO2 inside the cave before both processes start. Consequently:

ı13 Cc =

([13 CO2 ]b + [13 CO2 ]s + Ve · ([13 CO2 ]e − [13 CO2 ]b − [13 CO2 ]s ))/([12 CO2 ]b + [12 CO2 ]s + Ve · ([12 CO2 ]e − [12 CO2 ]b − [12 CO2 ]s )) − std × 1000 std

(C1)

(C2)

the time-dependent equations derived from C1 and C2 are: [CO2 ]c (t) = [CO2 ]b + [CO2 ]s (t) + Ve (t) · ([CO2 ]e − [CO2 ]b − [CO2 ]s (t))

 12

ı13 Cc (t) =

([13 CO2 ]b + [13 CO2 ]s (t) + Ve (t) · ([13 CO2 ]e − [13 CO2 ]b − [13 CO2 ]s (t)))/([12 CO2 ]b + [12 CO2 ]s (t) + Ve (t) · [ CO2 ]e − [12 CO2 ]b − [12 CO2 ]s (t) ) − std std

process) and certain quantity of the exterior air input due to a volumetric movement. Considering a mass balance, CO2 concentration of the cave air can be written as a relation between the background component and the exterior component: [CO2 ]c · Vc = [CO2 ]b · Vb + [CO2 ]e · Ve

Vc = Vb + Ve = 1

(B2)

The model for the advection can be written: [CO2 ]c = [CO2 ]b + Ve · ([CO2 ]e − [CO2 ]b )

(B3)

As consequence the resulting isotopic signal can be also modelled:

where [13 CO2 ]e and [12 CO2 ]e are the concentration of 13 CO2 and in the exterior air (ppm) This model for the volumetric flux into the cave allows quantify the ventilation rate (Ve ) that affects the subterranean atmosphere and causes the degassing, thus from Eq. (B3):

Ve =

[CO2 ]c − [CO2 ]b [CO2 ]e − [CO2 ]b

(B5)

Finally, Eqs. (B3) and (B4) can be rewritten as time dependent expressions as follows: [CO2 ]c (t) = [CO2 ]b + Ve (t) · ([CO2 ]e − [CO2 ]b ) ı13 Cc (t) =

(9)

Ct = Ct−1 +

F(t) ·A V

(D1)

where Ct : CO2 concentration at time t (ppm); Ct − 1 : CO2 concentration at previous time t − 1 (ppm); F/V: Particles surface flux per volume of the air mass established (ppm cm−2 ); A: Contact surface between the air masses (cm2 ). The particles flux of a gas in an air mass (i) due to its contact with other air mass (j) with different concentration is given by Fick’s law of diffusion (Nickerson and Risk, 2009), as follows: Fij =

−Dij · dCij

(D2)

zij

where Fij : particles flux (particle cm−2 s−1 ); Dij : effective diffusion coefficient (cm2 s−1 ); ␦Cij : difference between air masses (B4)

concentrations (ppm); zij : distance between air masses (cm) Combining Eqs. (D1) and (D2): Ct = Ct−1 − Dij ·

Cij zij

(t) ·

A V

(D3)

In studied case, CO2 concentration of the cave air (Cc ) increases due to the gas exchange with the overlying soil that is the source of CO2 . The effective diffusion coefficient (Dcs ) represents a factor dependent on air-filled pore space and gas diffusivity, e.g. modified Millington–Quirk pore-space relationship (Nickerson and Risk,

([13 CO2 ]b + Ve (t) · ([13 CO2 ]e − [13 CO2 ]b ))/([12 CO2 ]b + Ve (t)([12 CO2 ]e − [12 CO2 ]b )) − std × 1000 std

where Ve (t): total volume of exterior air input into the cave, which is dependent on the time (relative volume) that results from the advective flux (or ventilation rate) between the exterior and the cave.

(13)

An air mass in contact to other air mass varies its concentration by molecular diffusion following the equation (Nickerson and Risk, 2009):

([13 CO2 ]b + Ve ([13 CO2 ]e − [13 CO2 ]b ))/([12 CO2 ]b + Ve ([12 CO2 ]e − [12 CO2 ]b )) − std × 1000 std

12 CO 2

× 1000

(12)

Appendix D. Appendix D. Calculation of isotopic fractionation from Fick’s law

(B1)

where [CO2 ]e is the concentration of CO2 in the exterior atmosphere (ppm), [CO2 ]b is the background concentration of CO2 inside the cave at the beginning of the process (ppm) and Vc refers to volume of the cave air (expressed in volumetric units). Vb and Ve refer to the total volume of background component and the total volume of exterior component, respectively, both expressed in any volumetric measure unit. Vb and Vs can be expressed as relative volumes of Vc as follows:

ı13 Cc =



(10)

2009) assuming the diffusion of gas throw soil with water saturation as negligible. Supposing soil concentration (Cs ), distance between air masses (zcs ) and effective diffusion coefficient (Dcs )

128

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129

as constant parameters during the whole process, diffusion surface flux (Ft ) can be described as: Dcs Ft = − · zcs

tt (Cc (t) − Cs ) · dt

(D4)

tt−1

Therefore, for this studied case (see Section 4.3.1) the function defining the CO2 variation (Cc (t)) with time can be approximated to a line defined by the extreme values (Ct , tt ), (Ct − 1 , tt − 1 ), as follows: Dcs Ft = − · zcs

tt   Ct − Ct−1 tt − tt−1

· Ct −

Ct − Ct−1 · tt tt − tt−1



− Cs



· dt

(D5)

tt−1

and after solving the integral function is: Ft = −





Dcs 1 · · (Ct + Ct−1 ) − Cs .(tt − tt−1 ) 2 zcs

(D6)

Defining the mean CO2 concentration of the cave air for the studied interval as: Cm =

1 · (Ct + Ct−1 ) 2

(D7)

Finally, the simplified equation for the studied case can be expressed as follows: Ct = Ct−1 + Dcs ·

Cs − Cm A · t · V zcs

(D8)

where t is the elapsed time between t − 1 and t (s). The correlation between isotopologues effective diffusion coefficients can be obtained from the application of D8 as we assumed the same parameters A, z and V for the two air masses studied (soil and cave air): 12 D 13 D

cs cs

=

12 C

t 12 C

− 12 Ct−1 · 12 C s− m

13 C s 13 C t

− 13 Cm − 13 Ct−1

(8)

References Aksoy, H., Gedikli, A., Unal, N.E., Kehagias, A., 2008. Fast segmentation algorithms for long hydrometeorological time series. Hydrological Processes 22, 4600–4608. Amundson, R., Stern, L., Baisden, T., Wang, Y., 1998. The isotopic composition of soil and soil-respired CO2 . Geoderma 82, 83–114. Baldini, J.U.L., Baldini, L.M., McDermott, F., Clipson, N., 2006. Carbon dioxide sources, sinks, and spatial variability in shallow temperate zone caves: evidence from Ballynamintra Cave, Ireland. Journal of Cave and Karst Studies 68, 4–11. Benavente, J., Vadillo, I., Carrasco, F., Soler, A., Linan, C., Moral, F., 2010. Air carbon dioxide contents in the vadose zone of a Mediterranean karst. Vadose Zone Journal 9, 126–136. Bourges, F., Genthon, P., Mangin, A., D’Hulst, D., 2006. Microclimates of l’Aven d’Orgnac and other French limestone caves (Chauvet, Esparros, Marsoulas). International Journal of Climatology 26, 1651–1670. Bourges, F., Genthon, P., Genty, D., Mangin, A., D’Hulst, D., 2012. Comment on Carbon uptake by karsts in the Houzhai Basin, southwest China by Junhua Yan et al. Journal of Geophysical Research: Biogeosciences 117, G03006. Bowling, D.R., Massman, W.J., 2011. Persistent wind-induced enhancement of diffusive CO2 transport in a mountain forest snowpack. Journal of Geophysical Research: Biogeosciences 116, G04006. Bowling, D.R., Pataki, D.E., Randerson, J.T., 2008. Carbon isotopes in terrestrial ecosystem pools and CO2 fluxes. New Phytologist 178, 24–40. Breecker, D.O., Payne, A.E., Quade, J., Banner, J.L., Ball, C.E., Meyer, K.W., Cowan, B.D., 2012. The sources and sinks of CO2 in caves under mixed woodland and grassland vegetation. Geochimica et Cosmochimica Acta 96, 230–246. Buck, A.L., 1981. New equations for computing vapor-pressure and enhancement factor. Journal of Applied Meteorology 20, 1527–1532. Cerling, T.E., Solomon, D.K., Quade, J., Bowman, J.R., 1991. On the isotopic composition of carbon in soil carbon-dioxide. Geochimica et Cosmochimica Acta 55, 3403–3405. Cohen, G., Loisy, C., Laveuf, C., Le Roux, O., Delaplace, P., Magnier, C., Rouchon, V., Garcia, B., Cerepi, A., 2013. The CO2 -vadose project: experimental study and modelling of CO2 induced leakage and tracers associated in the carbonate vadose zone. International Journal of Greenhouse Gas Control 14, 128–140. Christoforou, C.S., Salmon, L.G., Cass, G.R., 1996. Air exchange within the Buddhist cave temples at Yungang, China. Atmospheric Environment 30, 3995–4006.

Corey, A.T., Kemper, W.D., Dane, J.H., 2010. Revised model for molecular diffusion and advection. Vadose Zone Journal 9, 85–94. Craig, H., 1953. The geochemistry of the stable carbon isotopes. Geochimica et Cosmochimica Acta 3, 53–92. Craig, H., 1957. Isotopic standards for carbon and oxygen and correction factors for mass-spectrometric analysis of carbon dioxide. Geochimica et Cosmochimica Acta 12, 133–149. CRC, 2003. Handbook of Chemistry and Physics – CRC, 84th ed. CRC Press, LLC, USA, 2369 pp. Crosson, E.R., 2008. A cavity ring-down analyzer for measuring atmospheric levels of methane, carbon dioxide, and water vapor. Applied Physics B – Lasers and Optics 92, 403–408. ˜ Cuezva, S., Sanchez-Moral, S., Saiz-Jimenez, C., Canaveras, J.C., 2009. Microbial communities and associated mineral fabrics in Altamira Cave, Spain. International Journal of Speleology 38 (1), 83–92. Cuezva, S., Fernandez-Cortes, A., Benavente, D., Serrano-Ortiz, R., Kowalski, A.S., Sanchez-Moral, S., 2011. Short-term CO2 (g) exchange between a shallow karstic cavity and the external atmosphere during summer: role of the surface soil layer. Atmospheric Environment 45, 1418–1427. Dredge, J., Fairchild, I.J., Harrison, R.M., Fernandez-Cortes, A., Sanchez-Moral, S., Jurado, V., Gunn, J., Smith, A., Spötl, C., Mattey, D., Wynn, P.M., Grassineau, N., 2013. Cave aerosols: distribution and contribution to speleothem geochemistry. Quaternary Science Reviews 63, 23–41. Ekblad, A., Högberg, P., 2001. Natural abundance of 13C in CO2 respired from forest soils reveals speed of link between tree photosynthesis and root respiration. Oecologia 127, 305–308. ˜ Elez, J., Cuezva, S., Fernandez-Cortes, A., Garcia-Anton, E., Benavente, D., Canaveras, J.C., Sanchez-Moral, S., 2013. A GIS-based methodology to quantitatively define an Adjacent Protected Area in a shallow karst cavity: the case of Altamira cave. Journal of Environmental Management 118, 122–134. Fernandez-Cortes, A., Sanchez-Moral, S., Cuezva, S., Benavente, D., Abella, R., 2011. ˜ Characterization of trace gases’ fluctuations on a ‘low energy’ cave (Castanar de Ibor, Spain) using techniques of entropy of curves. International Journal of Climatology 31, 127–143. ˜ Fernandez-Cortes, A., Benavente, D., Cuezva, S., Canaveras, J.C., Alvarez-Gallego, M., Garcia-Anton, E., Soler, V., Sanchez-Moral, S., 2013. Effect of water vapour condensation on the radon content in subsurface air in a hypogeal inactive-volcanic environment in Galdar cave, Spain. Atmospheric Environment 75, 15–23. Formanek, P., Ambus, P., 2004. Assessing the use of delta C-13 natural abundance in separation of root and microbial respiration in a Danish beech (Fagus sylvatica L.) forest. Rapid Communications in Mass Spectrometry 18, 897–902. Frisia, S., Fairchild, I.J., Fohlmeister, J., Miorandi, R., Spoetl, C., Borsato, A., 2011. Carbon mass-balance modelling and carbon isotope exchange processes in dynamic caves. Geochimica et Cosmochimica Acta 75, 380–400. Garcia-Anton, E., Cuezva, S., Fernandez-Cortes, A., Sanchez-Moral, S., Benavente, D., 2012. Daily variations of CO2 , ␦13 CO2 and CH4 of cave air controlled by external weather conditions: example of rapid survey in Altamira cave (north of Spain). Geophysical Research Abstracts 14 (eISSN: 1607-7962), EGU2012-4859-2. Garcia-Anton, E., Cuezva, S., Jurado, V., Porca, E., Miller, A.Z., Fernandez-Cortes, A., Saiz-Jimenez, C., Sanchez-Moral, S., 2013. Combining stable isotope (␦13 C) of trace gases and aerobiological data to monitor the entry and dispersion of microorganisms in caves. Environmental Sciences and Pollution Research, 1–12. Gedikli, A., Aksoy, H., Unal, N.E., 2008. Segmentation algorithm for long time series analysis. Stochastic Environmental Research and Risk Assessment 22, 291–302. Gedikli, A., Aksoy, H., Unal, N.E., 2010a. AUG-Segmenter: a user-friendly tool for segmentation of long time series. Journal of Hydroinformatics 12 (3), 318–328. Gedikli, A., Aksoy, H., Unal, N.E., Kehagias, A., 2010b. Modified dynamic programming approach for offline segmentation of long hydrometeorological time series. Stochastic Environmental Research and Risk Assessment 24, 547–557. Hopmans, J., Van Genuchten, M., 2005. Vadose zone: hydrological process. In: Encyclopedia of Soils in the Environment. Hillel, D, pp. 209–216. Hubert, P., 2000. The segmentation procedure as a tool for discrete modeling of hydrometeorological regimes. Stochastic Environmental Research and Risk Assessment 14, 297–304. Kayler, Z.E., Sulzman, E.W., Rugh, W.D., Mix, A.C., Bond, B.J., 2010. Characterizing the impact of diffusive and advective soil gas transport on the measurement and interpretation of the isotopic signal of soil respiration. Soil Biology and Biochemistry 42, 435–444. Keeling, C.D., 1958. The concentration and isotopic abundances of atmospheric carbon dioxide in rural areas. Geochimica et Cosmochimica Acta 13, 322–334. Kell, D.B., 2012. Large-scale sequestration of atmospheric carbon via plant roots in natural and agricultural ecosystems: why and how. Philosophical Transactions of the Royal Society B: Biological Sciences 367, 1589–1597. Kowalski, A.S., Serrano-Ortiz, P., Janssens, I.A., Sánchez-Moral, S., Cuezva, S., Domingo, F., Were, A., Alados-Arboledas, L., 2008. Can flux tower research neglect geochemical CO2 exchange? Agricultural and Forest Meteorology 148, 1045–1054. Maier, M., Schack-Kirchner, H., Hildebrand, E.E., Holst, J., 2010. Pore-space CO2 dynamics in a deep, well-aerated soil. European Journal of Soil Science 61, 877–887. Nachson, U., Dragila, M., Weisbrod, N., 2012. From atmospheric winds to fracture ventilation: cause and effect. Journal of Geophysical Research 117, G02016, http://dx.doi.org/10.1029/2011JG001898. Nickerson, N., Risk, D., 2009. Keeling plots are non-linear in non-steady state diffusive environments. Geophysical Research Letters 36, L08401.

E. Garcia-Anton et al. / International Journal of Greenhouse Gas Control 21 (2014) 113–129 Nickerson, N., Risk, D., 2013. Using subsurface CO2 concentrations and isotopologues to identify CO2 seepage from CCS/CO2 –EOR sites: a signal-to-noise based analysis. International Journal of Greenhouse Gas Control 14, 239–246. Ogretim, E., Mulkeen, E., Gray, D.D., Bromhal, G.S., 2012. A parametric study of the transport of CO2 in the near-surface. International Journal of Greenhouse Gas Control 9, 294–302. Pataki, D.E., Ehleringer, J.R., Flanagan, L.B., Yakir, D., Bowling, D.R., Still, C.J., Buchmann, N., Kaplan, J.O., Berry, J.A., 2003. The application and interpretation of Keeling plots in terrestrial carbon cycle research. Global Biogeochemical Cycles 17, 22. Post, W.M., Izaurralde, R.C., West, T.O., Liebig, M.A., King, A.W., 2012. Management opportunities for enhancing terrestrial carbon dioxide sinks. Frontiers in Ecology and the Environment 10, 554–561. Quindos, L.S., Bonet, A., Diazcaneja, N., Fernandez, P.L., Gutierrez, I., Solana, J.R., Soto, J., Villar, E., 1987. Study of the environmental variables affecting the natural preservation of the Altamira cave paintings located at Santillana-Del-Mar, Spain. Atmospheric Environment 21, 551–560. Risk, D., Kellman, L., 2008. Isotopic fractionation in non-equilibrium diffusive environments. Geophysical Research Letters, 35.

129

˜ Roland, M., Serrano-Ortiz, P., Kowalski, A.S., Goddéris, Y., Sánchez-Canete, E.P., Ciais, P., Domingo, F., Cuezva, S., Sanchez-Moral, S., Longdoz, B., Yakir, D., Van Grieken, R., Schott, J., Cardell, C., Janssens, I.A., 2013. Atmospheric turbulence triggers pronounced diel pattern in karst carbonate geochemistry. Biogeosciences Discussions 10, 1207–1227. Saiz-Jimenez, C., Cuezva, S., Jurado, V., Fernandez-Cortes, A., Porca, E., Benavente, D., Canaveras, J.C., Sanchez-Moral, S., 2011. Paleolithic art in peril: policy and science collide at Altamira Cave. Science 333, 42–43. Sanchez-Moral, S., Soler, V., Canaveras, J.C., Sanz-Rubio, E., Van Griekend, R., Gyselsd, K., 1999. Inorganic deterioration affecting the Altamira Cave, N Spain: quantitative approach to wall-corrosion (solutional etching) processes induced by visitors. Science of the Total Environment 243/244, 67–84. Serrano-Ortiz, P., Roland, M., Sanchez-Moral, S., Janssens, I.A., Domingo, F., Godderis, Y., Kowalski, A.S., 2010. Hidden, abiotic CO2 flows and gaseous reservoirs in the terrestrial carbon cycle: review and perspectives. Agricultural and Forest Meteorology 151, 529. Vargas, R., Carbone, M.S., Reichstein, M., Baldocchi, D.D., 2011. Frontiers and challenges in soil respiration research: from measurements to model-data integration. Biogeochemistry 102, 1–13.