Edinburgh Research Explorer Structural controls on the location and distribution of CO2 emission at a natural CO2 spring in Daylesford, Australia

9 Secure storage of CO 2 is imperative for carbon capture and storage technology, and relies on a 10 thorough understanding of the mechanisms of CO 2 retention and leakage. Observations at CO 2 11 seeps around the world find that geological structures at a local and regional scale control the 12 location, distribution and style of CO 2 emission. Bedrock-hosted natural CO 2 seepage is found in the 13 Daylesford region in Victoria, Australia, where many natural springs contain high concentrations of 14 dissolved CO 2 . Within a few meters of the natural Tipperary Mineral Spring, small CO 2 bubble 15 streams are emitted from bedrock into an ephemeral creek. We examine the relationship between 16 structures in the exposed adjacent outcropping rocks and characteristics of CO 2 gas leakage in the 17 stream, including CO 2 flux and the distribution of gas emissions. We find that degassing is clustered 18 within ~1 m of a shale-sandstone geological contact. CO 2 emission points are localised along bedding 19 and fracture planes, and concentrated where these features intersect. The bubble streams were 20 intermittent, which posed difficulties in quantifying total emitted CO 2 . Counterintuitively, the 21 number of bubble streams and CO 2 flux was greatest from shale dominated rather than the 22 sandstone dominated features, which forms the regional aquifer. Shallow processes must be 23 increasing the shale permeability, thus influencing the CO 2 flow pathway and emission locations. 24 CO 2 seepage is not limited to the pool; leakage was detected in subaerial rock exposures, at the 25 intersection of bedding and orthogonal fractures. These insights show the range of spatial scales of the geological features that control CO 2 flow. 27 Microscale features and near surface processes can have significant effect on the style and location 28 and rates of CO 2 leakage. The intermittency of the bubble streams highlights challenges around 29 characterising and monitoring CO 2 stores where seepage is spatially and temporally variable. CCS 30 monitoring programmes must therefore be informed by understanding of shallow crustal processes 31 and not simply the processes and pathways governing CO 2 fluid flow at depth. Understanding how 32 the CO 2 fluids leaked by deep pathways might be affected by shallow processes will inform the 33 design of appropriate monitoring tools and monitoring locations. 34

To date, there has been little focus on the influence of microscale features, such as bedding planes 55 and small fractures within foliated planes, on the surface expression of leaking CO2-though these 56 features are known to affect geofluid flow (Faulkner et al., 2010;Hippler, 1993;McCay et al., 2018).
Further, field experiments designed to mimic CO2 seepage by controlled CO2 release at shallow 58 depths have found CO2 flow pathways are influenced by a number of local factors, and thus the 59 creek bed. The bubbles are most apparent during the dry season when water level in the creek drops 120 (Shugg and Brumley, 2003). A hand pump on the east side of Sailors Creek draws water from a 121 borehole which was drilled in 2001. The bore encountered a significant flow of gassy mineral water 122 in a highly fractured horizon at 45 m depth, and the borehole casing was pressure cemented in this 123 portion (Shugg and Brumley, 2003). The mineral water is effervescent. Gases dissolved in the 124 mineral waters of the Daylesford region are reported to range from 88.6 -95.7 % CO2 with between 125 0.4 -0.8 % O2, 3.9 -10.2 % N2 and host trace quantities of He and Ne (Cartwright et al., 2000;126 Lawrence, 1969

Methods 135
Fieldwork at Tipperary was conducted in March 2017, towards the end of the summer when the 136 creek level was low. The fieldwork aimed to collect geological and structural data at the site to 137 observe the style of CO2 degassing and measure gas fluxes. 138 High precision GPS measurements of bubble locations and outcrop/creek features were taken using 139 an Altus APS3G high precision GNSS survey system for Real Time Kinematic (RTK) position 140 measurements. A base station was set up at each locality and the Rover recorded the UTM 141 coordinates of the feature. The positional accuracy of the RTK equipment is < 1 mm, but human 142 error positioning the RTK will be on the order of < 1 cm. There were some time delays and 143 complications obtaining position measurements due to tree cover and the footbridge which, in 144 addition to the typically sporadic nature of the bubbles streams, meant that the location of the 145 bubble streams was recorded using a local reference grid rather than the RTK. The bubble location 146 error is therefore approximately ~10 cm. 147 CO2 flux measurements were obtained using a West Systems portable flux system with attached 148 accumulation chamber (type B) and LI-840A CO2/H2O gas analyser following the method established 149 by Chiodini et al. (2001). A hollow 50mm PVC pipe frame was attached to the base of the 150 accumulation chamber as a floatation device in order to facilitate flux sampling at the water surface. 151 The base of the accumulation chamber was therefore slightly submerged in water and this change 152 in volume was accounted for when applying the ACK (a conversion factor between ppm/sec 153 (instrument unit) and g/m 2 /day). ACK temperature and pressure corrections (see Annex A) were 154 made using meteorological measurements recorded at the nearby Ballarat Airport at 10 min 155 intervals (Weatherzone, 2017). Where required, the floating flux chamber was attached to a pole to 156 enable sampling of bubble streams without disturbing the creek sediments. 157 Flux measurements were made at bubbling points. Several readings were also taken at non-bubbling 158 points across the pool to account for background diffuse degassing. The measurement period 159 varied, but generally lasted for 90 seconds or longer, or until the accumulation in the chamber 160 reached a CO2 concentration of 20,000 ppm (at which point the accuracy of the gas analyser is 161 negatively impacted). Time restraints prevented the quantitative measurement of every mapped 162 bubble point, so our sampling focussed on the most vigorous and continuous bubbling points in the 163 interests of producing the most reliable upper bound estimate of the total CO2 emission rate. A more detailed discussion of the characteristics and style of the gas emissions at Tipperary is reported 165 in Roberts et al (2018). 166 The presence of 'dry' seeps (CO2 seepage from rock to atmosphere, not through water) was 167 investigated using a tube connected to a Li-COR 81000A soil gas flux system ('CO2 sniffer'), allowing 168 CO2 concentrations of the air to be continuously measured. The inflow tube was used to identify 169 structural features in the outcrop that hosted dry gas seeps. 170 Structural measurements of outcropping bedrock were collected digitally using FieldMove Clino. 171 The area of outcrop and the pool area were calculated from GPS measurements using ArcMap 10.2 172 ©ESRI 2013. Spatial statistical analyses were performed to quantitively examine seep distributions 173 with respect to geological structures. 174

Field observations 176
The low creek level in March 2017 meant that a series of isolated pools were found in the bed of 177 Sailors Creek rather than a flowing stream. CO2 degassing characterised by numerous bubble 178 streams was observed in a single pool of water close to the footbridge near to Tipperary Mineral 179 Spring ( Figure 2). The surface area of the pool was ~61.8 m 2 and water depth was greatest (40 cm) 180 in the centre of the creek. 181 Two units of Ordovician rock crop out in the creek bed. The majority is fine-grained buff coloured 182 massive sandstones that occur in 1 to 3 m thick beds, but a ~2.2 m thick blue-grey shale layer crops 183 out beneath the footbridge. Rock bedding is oriented NW-SE and dips 65 -80˚ to the NE ( Figure  184 3a,b). In contrast to the sandstone, the shale is thinly bedded (cm scale) with moderately well-185 developed bedding and parallel foliation. The shale is fissile, and more weathered than the 186 sandstone and is less well exposed at the creek edge. The sandstone is much more cohesive, and 187 while the fracture density is lower than in the shale, the fractures are longer. Three sets of fractures 188 and joints are observed at the outcrop (Figure 3a  and suggests that the dissolved CO2 content of the pool water is high, but also variable across the 237 pool, but no water samples were collected to verify this. The minimum daily emission rate for the 238 pool was estimated by applying the average of the background readings across the surface area of 239 the pool and neglecting the input of bubbling points giving a value of ~4900 g d -1 . Degassing rates at 240 bubbling points ranged from 11.4 to 374 g d -1 . These values will represent combined emission of 241 CO2 from bubbles and water surface degassing. Bubbling rates were greatest at the sandstone-shale 242 contact beneath the footbridge (figure 2). The maximum daily emission rate from the pool was 243 calculated by assuming that bubbling from the point sources was continuous and adding the sum of 244 the maximum measured bubble point emissions (2267 g d -1 ) to the maximum background degassing 245 rate giving a value of 7170 g d -1 . Yearly emissions from the pool can therefore be constrained within 246 the lower and upper estimates of 1.8 -2.6 t y -1 , which correspond to average flux rates across the 247 pool area of 79 -116 g m 2 d -1 . 248

CO2 seepage from outcropping rocks 249
The CO2 sniffer detected two locations where atmospheric CO2 concentrations were up to 6,000 250 ppm in the dry outcrops on the banks of the creek. In both cases the high CO2 concentrations were 251 extremely localised, and occurred in jogs or intersections in uncemented, bedding-orthogonal, 252 fractures in sandstones (Figure 4a,c). Further, these concentrations were consistently high. That is, 253 returning to the same location several minutes later, similarly high concentrations from between 254 2,000 to 6,000 ppm were recorded. Consistently high CO2 concentrations at these features suggests 255 that seepage was continuous during the survey period (several hours), and also rules out the 256 possibility that elevated CO2 concentrations were an artefact caused by density-driven pooling of 257 CO2 degassed from the pool at times of particularly low wind speed. Interestingly, when we poured 258 ~1L water over the seeping fracture the CO2 concentration returned to background atmospheric 259 levels and took 14 minutes for CO2 seepage to become re-established. Weak elevations in CO2 260 concentration (up to 500 ppm) were detected along a bedding-orthogonal fracture in the foliated 261 shale (Figure 4b,d). However, the area of shale outcrop was less than the sandstone (in part due to 262 bedding thickness, in part due to the morphology of the outcrop) and so we cannot compare 263 instances of CO2 detection per area of rock. 264  azimuth. For an ideal scenario with no finite size effects the correlation function will plot as a power 281 law, P ∝ r κ , where P is probability, r is radius, and the constant κ describes the spatial distribution of 282 points. For randomly distributed points, κ = 2. If points are clustered, κ < 2. For points that are 283 distributed on a line, κ = 1. Other arrangements, such as points distributed on multiple lines, will 284 give κ values between 1 and 2. 285 The study area is spatially limited; the pond is asymmetric and approximates a 11 x 5 m rectangle. 286 As such, synthetic data was created to act as a 'control' for comparison with the CO2 bubble stream 287 distributions. The total numbers of measured and synthetic points are the same (60). Synthetic data 288 were generated from multiple random (Poisson) point distributions for different spatial scenarios, 289 including an 11 x 11 m exposure and an 11 x 5 m exposure with long axis orientated NNE, mimicking 290 Tipperary pool which is only 4.4 m at its narrowest point and longer in the NE-SW orientation (see 291 Figure 2). Additional synthetic data were generated for different spatial scenarios, outlined in Table  292 inset Points randomly distributed within 11 x 1 m rectangle, long axis orientated ~NW (simulating points located within a bed, or around a fault or geological contact)

308
(iii) greater than or equal to 1m but less than 2 m; and (iv) greater than or equal to 2 m but less than 10 m. Roll-off in 309 the TPCF occurs at ~4 m separation distance due to the outcrop extent.  While bubbling points were observed in the field to be located along bedding, foliation and fracture 330 planes, the point azimuths < 2 m do not exhibit very clear spatial trends other than the bedding and 331 foliation. At Tipperary, a range of fracture orientations were measured (see Fig 3c) with dominant 332 sets trending NE-SE and NW-SE. Bubble pairs do show peaks in these orientations at < 2 m 333 separation, but it is difficult to distinguish these from noise. 334

Discussion 335
The role of geological structures and CO2 seepage at Tipperary 336 Central Victorian mineral water springs are commonly channelled by regional thrust faults and can 337 emerge close to anticline crests (Shugg, 2009). Although obscured in the field area, the regional 338 geological map shows an inferred NNW-SSE trending anticline to the West of Sailors Creek, its 339 projected axis passing less than 20 meters from the creek. An inferred NW-SE fault runs in the same 340 orientation as the creek but is mapped as terminating before intersecting the creek (Figure 1) However, the mineral waters probably degas CO2 during their ascent to surface; we observed dry 353 seepage from outcropping rocks, while at nearby springs down-hole camera surveys found bubbles 354 starting to form around 20-30 m below the water table (Shugg 2009). When two-phase flow 355 establishes, the CO2 may migrate to surface via different pathways to its 'parent' water, depending 356 on the hydraulic properties of the available flow pathways and water table depth. There is no 357 appreciable thermal anomaly between the mineral waters and the surrounding groundwater 358 (Weaver et al., 2006), so they are not ascending due to thermal buoyancy drive. Instead they may 359 be migrating towards the surface due to a combination of hydraulic head and the fluid flow 360 pathways offered by nearby structures, enhanced by buoyancy from gas lift due to CO2 ebullition as 361 the waters depressurise during ascent. 362 CO2 seepage at Tipperary spring concentrates near the western sandstone-shale contact. Some 81% 363 of total measured bubble stream emissions emerge from the shale dominated features in the river 364 bed and we detected extremely localised dry seepage from open fractures within outcropping 365 sandstone and shale. The bedding and foliation orientation of rocks exposed in Sailors Creek follows 366 the regional trend from NNW-SSE Devonian compression. Our spatial statistical analyses find that bubble point data exhibit this NNW-SSE (150-170°) trend at all point separation distances. Bubble 368 points located within <2 m of each other show other preferred alignments (NE-SW and SE-NW, NNE-369 SSW, ENE-WSW) but these trends are weak compared to the NNW orientations. Seepage mostly 370 occurred in a narrow region, ~1 m width. In the field, we noted that bubble locations appeared to 371 be primarily controlled by bedding and foliation planes, but also by fractures and joints across both 372 the sandstone and shale members. Thus, while regional structures may govern mineral water and 373 CO2 flow (Shugg, 2009), it seems that primary features (the sandstone-shale contact) may control 374 fluid flow in the deep and shallow subsurface, and at very shallow depths the small secondary 375 structures (fractures, foliation) offer pathways to surface. 376 What is unusual at Tipperary is that gas primarily discharges from shales. Mudrocks and shales 377 typically have low permeabilities, and high capillary entry pressures for two-phase flow, which 378 makes them good seals for conventional hydrocarbon traps. At other sites around Daylesford, such 379 as Sutton Spring, mineral water and gas discharges from the joints and fracture faces in sandstone 380 beds (Shugg, 2009). These sandstones form the regional aquifer. Within these sandstone units 381 intergranular porosity is limited to certain horizons. Therefore, groundwater flow is predominantly 382 hosted by fractures and joints. Observations from exposed bedrock at Tipperary suggest that bulk 383 rock permeability is most likely offered by the primary fracture set (NE-SW trend) together with the 384 bedding. However, at Tipperary, our observations, corroborated by spatial statistical analyses, find 385 that CO2 bubbles preferentially emerge from foliation and fracture intersections in the shale. This 386 indicates that in the shallow subsurface the high density of subvertical foliation and bedding-387 orthogonal fractures in the shale must be more transmissive than bedding and fracture planes in 388 the sandstone. At outcrop, the fractured, folded and uplifted shales of the Ordovician succession 389 clearly are not sealing. This could be due to unloading and weathering, and so these pathways have 390 opened only close to the surface. Conversely, the bulk permeability of the shale units may be greater 391 than the sandstone for these units. Figure 6 schematically summarises the proposed mechanism for 392 CO2 delivery to the creek bed at Tipperary. 393

402
Other seeps worldwide emerge from clays. For example, in the Cheb Basin, CO2 degassing close to 403 a fault zone in clay dominated rocks occurs as highly localised emissions from fine fractures, 404 facilitated by "micro-channels" in the clays which originated from shear (Bankwitz et al., 2003). 405 Where low permeability rocks outcrop in Italy, CO2 degassing occurs as vent like emissions rather 406 than as springs or spring associated emissions, which more commonly occur from high permeability 407 rocks (Roberts et al., 2014). However, at Tipperary, what is surprising is that CO2 preferentially 408 emerges from the shales rather than the sandstones. 409 That said, CO2 seepage is not confined to the shale. Seepage occurs from sandstones submerged in 410 Tipperary pool, and the CO2 sniffer detected high CO2 emissions from isolated points in the 411 outcropping sandstone. Therefore there are gas flow pathways in the sandstone, but there are 412 fewer pathways in the sandstone than in the shales (where gas emission is greater), and these 413 pathways are extremely localised; occurring at jogs and intersections along low-dip bedding-414 orthogonal fractures, where they intersected the bedding plane (Figure 5e). 415 Interestingly, the sniffer results indicate that it is likely that dry CO2 seepage is greater from the 416 outcropping sandstone than from outcropping shale. This is in contrast to CO2 fluxes measured in 417 Tipperary pool where more numerous and distributed bubble streams occur in the shale, and 75% 418 of the total CO2 flux via bubbles from Tipperary Pool is occurring from the shales. It is possible that 419 this is a sampling artefact; the seep area is limited, the shale bed is thinner than the sandstones, 420 and the area of outcrop is much smaller for the shale because it has preferentially eroded on the 421 creek banks. If this is a real signal, there could be several explanations for the contrasting behaviour 422 of the seeps through water and into air. Firstly, the rate of CO2 seepage through the two lithologies 423 could be the same but is occurring via distributed pathways (lots of small fractures and bedding 424 partings) in the shale, with fewer localised high flux fracture-bedding intersections in the sandstone. 425 As the shale was more foliated and thinly bedded, there were many more bedding parallel features 426 to permit flow and so facilitate distributed seepage than in the massive bedded sandstone. 427 Secondly, the outcrop style varies between the two lithologies: the sandstone stands proud of the 428 surface and some individual bedding planes are exposed, whereas the shale is more eroded and the 429 bedding is only viewed end-on. This means that the fractures that are low-dip (which are the ones 430 that host the high dry seepage from the sandstone) are exposed in the sandstone, but are unlikely 431 to be exposed in the shale at this site. Thirdly, flow pathways in the sandstone might be more likely 432 to become obstructed by river sediments than the smaller aperture features in the shale. Regardless 433 of the reason, our observations suggest that flow pathways in the very shallowest subsurface, and 434 therefore how CO2 seeps present, are highly sensitive to local conditions. 435

CO2 flux at Tipperary 436
Estimating total CO2 flux at Tipperary Spring is challenging given the intermittency of the CO2 bubble 437 streams. Bubbling, and therefore CO2 flux, was not continuous in the pool. The flux of CO2 from 438 depth is probably continuous, but the bubbles are intermittent either due to high connectivity of 439 the flow pathways (with flow paths turning 'on' and 'off') and/or due to water saturation of the flow 440 pathways; CO2 gas pressure must build enough to overcome capillary flow pressure, and the 441 pressure of the water in the fracture. Such temporal and spatial variability has been observed at other CO2 seeps including Laacher See (Germany -CO2 bubbling is observed from the floor of a crater 443 lake), Panarea (Italy -submarine geothermal region) and at the QICS project (Scotland -simulated 444 CO2 leak to the marine environment) Blackford  Natural non-volcanic seeps are the most appropriate analogues for seeps that might potentially 471 develop above engineered carbon stores, if injected CO2 seeps to the surface by natural CO2 472 pathways. However there are limitations to their comparability. Storage sites will be selected for 473 specific sealing characteristics. In contrast, surface seepage at natural CO2 seeps occur because the 474 geology is not ideal for long term CO2 trapping; the geology around Daylesford is composed of highly 475 heterogeneous and tectonised sediments crossed by faults. Faulted reservoirs with heterogeneous 476 or poorly permeable overburdens will probably not be selected for long term geological storage. In 477 addition, natural CO2 seeps will be migrating by natural fluid pathways whereas the greatest risks of 478 CO2 leakage from engineered CO2 stores are man-made pathways, such as improperly sealed 479 boreholes (IPCC, 2005) or geomechanical effects from the pressure response to CO2 injection 480 (Verdon et al., 2013). That said, natural CO2 seeps may still be comparable to leakage through the 481 overburden, independent of the leakage pathways from the reservoir, and seepage through clay 482 formations, can serve analogues of CO2 migration through cap rocks. 483

Implications for CCS 484
It is important to assure regulatory bodies and the public of CO2 storage integrity. This includes 485 demonstrating capability to (i) select storage sites that will successfully retain CO2 in the subsurface, 486 and (ii) identify potential CO2 leakage and (iii) quantify any leaked CO2. 487 In the case of CO2 migration from onshore engineered storage sites, if the leaked CO2 migrates to 488 the near surface it could dissolve into groundwaters (and perhaps emerge as a dissolved constituent 489 of groundwaters at natural springs), seep to atmosphere as a dry gas, or seep into water bodies such 490 as lakes or rivers. Where migrating CO2 dissolves into groundwaters, the groundwater flow systems 491 will then govern its flow path, and only at shallow depth will decreasing pressures cause gas 492 ebullition and facilitate the ascent of a separate gas phase. Indeed, studies of onshore natural 493 analogues and field sites find that CO2 seeps are more likely to emerge in topographic low points 494 where there may be rivers or lakes, though there are examples of seeps that buck this trend (e.g. if 495 flow is fault controlled) (Roberts et al., 2014). 496 In this work, we examine the surface expression of CO2 seepage originating from transport of CO2-497 rich regional groundwaters. We find that, while regional features may govern CO2 delivery, in the 498 shallow subsurface CO2 pathways are localised to small scale geological features, and that fluxes are 499 intermittent and consequently difficult to quantify due to the intermittency of bubbling pathways. 500 To date, most research has focussed on predicting the large-scale geological features that may 501 enable CO2 to migrate from the storage reservoir such as large faults, boreholes or gas chimneys 502 (IEAGHG, 2017). Such macroscale (seismically resolvable) features are likely to be known about at 503 the site characterisation phase of a project. However, shallow crustal processes change the rock properties that affect CO2 spread and delivery to surface. Different, smaller scale geological 505 features, that are not likely to be seismically resolvable, may become important controls on CO2 506 flow in the shallow subsurface. At Tipperary, CO2 seep distribution is controlled by microscale 507 features such as foliation and bedding planes, joints and fractures in outcropping rock, probably 508 dilated by uplift and weathering, which leads to degassing from a shale formation that is typically 509 sealing. These observations support previous research investigating the role of topography and 510 lithology in CO2 seep location and characteristics (Roberts et al., 2014), and has important 511 consequences for the design of CCS monitoring approaches. Surface monitoring programmes must 512 focus on more than the processes and pathways governing leakage at depth; they must also 513 consider how the CO2 fluids leaked by natural or man-made pathways might disperse in the near 514 surface and be expressed at the surface. These shallow processes will inform the design of the right 515 monitoring tools and monitoring locations. 516 Our work thus provides insight into the scale of which geological features control CO2 flow and the 517 spatial and temporal variability of CO2 leakage. Essentially, site characterisation during site selection 518 and monitoring design must assess the geology and hydrogeology at a range of spatial scales. 519 Surface processes, often overlooked, will govern the style and location of leakage, and so should 520 inform the design of appropriate monitoring strategies. 521

Conclusions 522
We have studied the location and characteristics of CO2 emission at Tipperary natural CO2 seep in 523 Daylesford, Victoria (Australia) as an analogue for leakage from engineered CO2 stores. Seepage 524 largely occurs as bubble streams in a pool in Sailors Creek, close to the Tipperary mineral springs 525 which have high dissolved CO2-content. We also observed CO2 degassing from subaerial rock 526 outcrop. Observation and spatial statistical analyses find that at a meso-scale (multiple meters) the 527 location of CO2 bubble streams are controlled by the sandstone-shale geological contact. At a 528 smaller (meter to centimetre) scale, gas emission is controlled by structural features, primarily 529 fractures intersecting the foliation or bedding planes. The intermittency of the bubble streams, and 530 their distribution, makes CO2 flux challenging to quantify. Unusually, CO2 emission is greatest from 531 the shale, rather than the sandstone that forms the regional aquifer. Surface processes are likely to 532 be affecting rock transmissivity, which governs CO2 flow at the near surface. Our work has important 533 implications for characterising and monitoring of CO2 stores: microscale features and near surface 534 processes can have significant effect on CO2 leak locations and rates. Flow pathways through the very shallowest part of the subsurface are highly dependent on local conditions, and may produce 536 the highest flux in counter-intuitive locations (e.g. hosted by the 'low permeability' shales at 537 Tipperary). Understanding of shallow crustal processes and specific site conditions are essential to 538 inform the design of effective surface monitoring tools and approaches. Secondly, should leakage 539 from the storage reservoir occur, the surface leak identification and quantitation approaches must 540 be extended to consider intermittent or variable CO2 emission rates. 541

Acknowledgements 542
We are grateful for contributions, discussions, and assistance from Jade Anderson, Ivan Schroder To convert the rate of change in CO2 concentration measured in the accumulation change (e.g. in 559 ppm/s) to a flux (mole/m 2 /day), the rate needs to be multiplied by a correction factor which 560 considers the volume of the chamber, temperature and pressure: 561 where 563  P is the barometric pressure expressed in mBar  Tk is the air temperature expressed in degrees Kelvin 566  V is the chamber net volume in cubic meters (less the portion of the chamber submerged to 567 create a seal on the water surface) 568  A is the chamber inlet area in square meters 569 570 Data Availability Statement 571 All data (precise bubble locations and distributions, fluxes, and whether bubbling occurred at a 572 fracture or foliation/bedding at Tipperary pool), are available from the UKCCSRC Data and 573 Information Archive, under the DOI: (to be added). 574