Using noble gas fingerprints at the Kerr Farm to assess CO2 leakage allegations linked to the Weyburn-Midale CO2 Monitoring and Storage Project

15 For carbon capture and storage technology to successfully contribute to climate mitigation 16 efforts, the stored CO2 must be securely isolated from the atmosphere and oceans. Hence, 17 there is a need to establish and verify monitoring techniques that can detect unplanned 18 migration of injected CO2 from a storage site to the near surface. Noble gases are sensitive 19 tracers of crustal fluid input in the subsurface due to their low concentrations and unreactive 20 nature. Several studies have identified their potential to act as tracers of deep fluid migration 21 to the shallow subsurface, but they have yet to be used in a contested situation. In January 22 2011 it was reported extensively in global media that high CO2 concentrations in soils and 23 related groundwater pollution had been identified on a farm property belonging to the Kerr 24 family, located near to the town of Weyburn in Saskatchewan, Canada. The origin of this CO2 25 pollution was cited to be the nearby Weyburn-Midale CO2 Monitoring and Storage Project. 26 Here, as part of an investigation funded independently of the Weyburn-Midale field operators, 27 we present δCDIC, He/He, He/Ne, Ne, Ar, Ar and Kr measured in waters obtained 28 from four groundwater wells located on and surrounding the Kerr property. We aim to establish 29 if stable carbon and noble gas natural tracers are effective at determining if migration of CO2 30 from the storage project was responsible for the alleged high CO2 concentrations and water 31 pollution measured on the Kerr farm. We compare the stable carbon isotope and noble gas 32 ‘fingerprints’ of the Kerr groundwaters to those expected in a water equilibrated with the 33 atmosphere under local recharge conditions, the produced CO2 obtained from production 34 wells, and the CO2 injected into the Weyburn and Midale oil fields. We find that the stable 35 carbon isotope data do not constrain the origin of the dissolved CO2 in the Kerr groundwaters. 36 Due to low noble gas concentrations in the captured CO2 we are unable to completely rule out 37 the presence of 20 to 34% contribution from injected CO2 to the groundwaters surrounding the 38 Kerr property. However, we find that all of the Kerr groundwater samples exhibit noble gas 39 fingerprints that would be expected in a shallow groundwater in contact with the atmosphere 40 and hence there is no evidence for the addition of a deep radiogenic component or dilution 41 from the addition of a gas phase low in atmospheric derived noble gases. Our findings 42 corroborate previous studies that indicate that elevated CO2 concentrations found on the Kerr 43 property are almost certainly of biological origin, and not migrated from the deep subsurface. 44 The comprehensive follow up to these CO2 leakage allegations outlined in this study provides 45 a robust framework for responses to any future leakage allegations at CO2 storage sites and 46 further highlights that no single technique can categorically identify the origin of CO2 in the 47 shallow subsurface. Hence, it is essential that the full range of geochemical tracers (stable 48 carbon and C isotopes, noble gases, water chemistry, process based gas ratios) are 49 integrated with a comprehensive understanding of geological and engineering data in 50 response to CO2 leakage allegations in the future. 51

nature. Several studies have identified their potential to act as tracers of deep fluid migration 21 to the shallow subsurface, but they have yet to be used in a contested situation. In January  Here, as part of an investigation funded independently of the Weyburn-Midale field operators, 27 we present δ 13 CDIC, 3 He/ 4 He, 4 He/ 20 Ne, 20 Ne, 36 Ar, 40 Ar and Kr measured in waters obtained 28 from four groundwater wells located on and surrounding the Kerr property. We aim to establish 29 if stable carbon and noble gas natural tracers are effective at determining if migration of CO2 30 from the storage project was responsible for the alleged high CO2 concentrations and water 31 pollution measured on the Kerr farm. We compare the stable carbon isotope and noble gas 32 Noble gas fingerprinting techniques have also recently been used to identify micro-seepage 173 of CO2 and CH4 above the Teapot Dome oil field in Wyoming (Mackintosh and Ballentine, 174 2012). This study found that 3 He/ 4 He ratios in the soils gas were considerably below the 175 atmospheric ratio, due to the addition of a radiogenic 4 He component, which was also reflected 176 by elevated 4 He concentrations and 4 He/ 20 Ne ratios relative to atmospheric values. Mackintosh 177 and  concluded that the detection of crustal-sourced helium micro-seepage 178 into water saturated systems will be enhanced by two orders of magnitude compared with soil 179 gases. This is due to the low solubility of helium in water which results in a baseline 180 concentration which is two orders of magnitude lower than the expected atmospheric 4 He 181 concentration in a soil gas (Mackintosh and Ballentine, 2012). an additional means to test for the presence of a deep subsurface sourced gas that is depleted 198 in atmospheric noble gases. 199

Approach 200
We aimed to evaluate the effectiveness of δ 13 CDIC, 3 He/ 4 He, 4 He/ 20 Ne, Ne, Ar and Kr 201 fingerprints in determining the validity of the allegations of CO2 leakage made on the Kerr 202 property. To achieve this we undertook measurements of a suite of noble gases and C stable 203 isotope tracers from three different sources: (i) CO2 injected into the field (injected CO2); (ii) 204 fluids produced from the field (produced CO2); (iii) groundwaters at the Kerr property and 205 surrounding area (Kerr groundwaters). We aimed to determine if migration CO2 originating 206 from either the CO2 injected into, or CO2 contained in the fluids produced from the Weyburn 207 field was responsible for the alleged elevated CO2 concentrations on the Kerr property. To do 208 this we compare the noble gas and C isotope fingerprints between type (i), (ii) samples to 209 those of type (iii), the Kerr groundwaters. 210 Based on the studies outlined in the Scientific Background section we hypothesise that any 211 external CO2 addition to the Kerr groundwaters, bar those of shallow subsurface biologic 212 processes, would result in either the addition of crustal derived 4 He and/or a depletion in the 213 main atmospheric derived noble gases of 20 Ne, 36 Ar, 40 Ar and Kr. The addition of a crustal 214 radiogenic 4 He component can be identified by elevated 4 He/ 20 Ne ratios above those of the 215 atmosphere and/or a reduction in 3 He/ 4 He ratios below those of atmospheric values. We focus 216 on the Kerr groundwaters as opposed to soil gases based on the study of Mackintosh and 217 Ballentine (2012) which showed helium anomalies would be enhanced by two orders of 218 magnitude compared with soil gases as a result of the low solubility of helium in water. 219

Methods 220
Sample collection was undertaken over a period of three days in late June 2011, some 11 221 months following the soil gas sampling undertaken in the previous summer on which the 222 leakage allegations were based (Lafleur, 2010). Samples of injected CO2 were collected from 223 a the sampling port of a Cenovus injection well (Well ID -101/12-04-006-13 W2/0) located 224 approximately 10 km northwest of the Kerr quarter. A sample of CO2 separated from the 225 produced reservoir fluids (produced gas, water and oil) was collected from the sampling port 226 of the flowline emerging from a Cenovus satellite processing facility located at 16-30-05-13 227 ( Fig. 1). This flowline contained produced gas separated by the first stage separation system 228 at the satellite site from the oil, gas and water collected from 14 production wells that 229 surrounded the Kerr property. Gases were collected from both the pressurised injection well 230 and the satellite processing facility flow line using a high pressure to low pressure step down 231 regulator, allowing gas collection at slightly above atmospheric pressure in 70 cm long vacuum 232 tight copper tubes held in aluminium clamps. Shallow groundwaters were collected from the 233 domestic groundwater well on the Kerr farm, two domestic groundwater wells on the adjacent 234 Thackeray farm and the IPAC No. 1 monitoring well which was drilled during the sampling 235 program. This was located as close to the maximum CO2 anomaly reported by Petro-Find as 236 possible given the underlying ground conditions required for the drilling rig (Lafleur, 2010) (Fig.  237 1). All of the wells were of standard shallow groundwater bored type construction, drilled using 238 a rotary bucket auger and completed with PVC casing utilising a sand screen at the base. 239 Each well was 0.762 m in diameter, ranging in depth below ground surface from 3.09 m 240 (Thackery Farm Well) to 12.29m (Thackery House Well). Groundwater was encountered 241 between 1.76m (IPAC ~1 monitoring well) and 3.56m below ground level. Samples were collected from the base of the well water volumes using a peristatic pump, with each well being 243 pumped until water chemistry parameters stablished before sample collection. 244 δ 13 C (CO2) values of the gas samples and δ 13 CDIC values of the waters were determined at 245 the University of Rochester using a Delta S (Finningan) mass spectrometer and the analytical 246 error was ± 0.2 ‰. Standard extraction and purification procedures were used (Jenden et al., 247 1993) and the ratios are expressed as δ 13 C ‰ V-PDB. The dissolved gases were extracted 248 on a stainless steel and 1720 glass extraction line at the University of Rochester using 249 standard procedures (Poreda et al., 2004 (Poreda and Farley, 1992). Bulk gases were purified by consecutive exposure to 253 a Zr-Al getter (SAES ST-707) held at 450°C and a SAES SORB-AC cartridge held at 250°C 254 then cooled to 25°C. This was followed by the sequential trapping of Ar into an activated 255 charcoal finger at liquid N2 temperature (-178°C) and the He and Ne into an activated charcoal 256 finger at -261°C. He was released from the cryogenic finger at -242°C and expanded into the 257 spectrometer and measured, followed by Ne and Ar analyses. He, Ne, Ar, and Kr 258 concentrations were determined by comparison to an air standard of known volume (0.77 259 cm 3 ). Helium isotope ratios were normalized using a Rochester air standard. Neon isotope 260 ratios were corrected for interference by measurement of 40 Ar 2+ and CO2 2+ ( 40 Ar 2+ was typically 261 < 0.4 % of total 20 Ne signal on the faraday cup and CO2 2+ was below detection limits for 22 Ne). 262 The two sigma analytical error for the 3 He/ 4 He ratio is approximately 0.5% and those for both  We report the concentrations of noble gases dissolved in water, rather than the concentrations 271 of noble gases in the headspace gas degassed from the waters. This is because the amount 272 of headspace (ie non noble gas) gases exsolved from the Kerr groundwaters was insufficient 273 to obtain high quality ratio concentrations of noble gases relative to the total exsolved gas. 274 However, the concentration of the individual noble gases degassed from the water samples 275 was sufficient for high quality analysis to be performed; hence these concentrations are 276 presented relative to the amount of water degassed. Reporting the dissolved noble gas 277 concentration in groundwaters in this manner is standard practice in shallow groundwaters 278 where small quantities of dissolved gases are present (Kipfer et al., 2002). 279 To allow direct comparison between the different sample types, and to detect any external 280 input to the Kerr groundwaters from the Weyburn EOR operations, we calculate the noble gas 281 concentration in water that would arise from equilibrium of the noble gases within the injected 282 and produced CO2 with a shallow groundwater in the area surrounding the Kerr property.  (Crovetto et al., 1982;Smith, 1985). Under these conditions the calculated 287 Henry's constants for He, Ne, Ar, Kr and Xe are 14.12, 11.69, 3.41, 1.80 and 1.19 GPa, 288

respectively. 289
We also calculate the expected concentration and isotope ratio ranges of atmosphere-derived 290 noble gases dissolved in the groundwater, known as air-saturated water (ASW). These  values ranging from -13.4 ± 0.2 ‰ to -19.0 ± 0.2 ‰ (Table 1).

Noble Gas Concentrations 313
4 He concentrations exhibit marked distinctions depending on sample type (Fig. 2). The lowest 314 concentration of 41.63 ± 0.5 µcm 3 kg -1 is that calculated for a shallow groundwater which has 315 equilibrated with the injected CO2 (see Methods). The groundwaters sampled from the wells 316 on and around the Kerr property (the Kerr groundwaters) exhibit a range of 39.2 ± 0.6 to 86.9 317 ± 1.3 µcm 3 kg -1 which is almost identical to the air saturated water (ASW) concentration range 318 of 42.1 ± 0.6 to 85.8 ± 1.2 µcm 3 kg -1 . This indicates that there is no presence of 4 He in excess water in equilibrium with produced CO2 from the Weyburn field (Table 2). The highest noble 329 gas concentrations are those measured in the Kerr groundwaters, which overlap with the 330 calculated ASW range (Table 1 and Fig. 2 and 3). overlapping with the ASW range of 0.288 ± 0.007 -0.325 ± 0.01 (Fig. 4). The injected CO2 335 exhibits a higher ratio of 12.6 ± 0.3, with the duplicate produced CO2 samples ranging from 336 1000 ± 21 to 1488 ± 31, significantly above the calculated air saturated water (ASW) range 337 (Fig. 4). The above ASW ratios of the produced CO2 indicate an excess of 4 He above 338 atmospheric levels. in the CO2 produced from the Weyburn field (Fig. 5). The CO2 injected into the Weyburn field 342 has a slightly higher ratio of 0.193 ± 0.001 Ra. The range observed in the Kerr groundwaters 343 of 0.880 ± 0.004 to 1.103 ± 0.006 Ra is significantly above that of the other samples (Fig. 5). The produced CO2 δ 13 C (CO2) duplicate values obtained in this study were -14.8 and -12.4 ± 365 0.2 ‰ (Table 1) throughout their study, as more of the injected CO2 reaches the production wells, and a similar 374 evolution would be expected in the phase 1C region of the field. 375 Whilst it would obviously have been beneficial to undertake multiple measurements of the CO2 376 injected into and CO2 produced from the Weyburn CO2-EOR field, the above comparison 377 shows that the samples we have collected are representative of the range of CO2 injected and 378

418
These studies highlight that δ 13 C measurements alone cannot be used as a distinctive means 419 to determine the origin of CO2 measured on the Kerr property as also outlined by a recent 420 signal to noise analysis (Risk et al., 2015).

3 He/ 4 He ratios and 4 He concentrations 447
In Fig. 5  (1 Ra) and the average ASW 4 He concentration. Tick marks indicate the portion of 4 He 451 originating from either the CO2 injected or CO2 produced from the Weyburn field on the mixing 452 lines presented. Three of the measured groundwaters contain an excess of 3 He relative to the 453 predicted concentration in ASW, resulting in 3 He/ 4 He ratios above the ASW ratio of 1 Ra. This 454 can be explained by the presence of 3 He originating from the decay of tritium which was 455 emitted to the atmosphere as a result of nuclear weapons testing from the 1950s to 1980s 456 (Happell, 2004). This additional 3 He component is variable and creates some uncertainty 457 around the baseline 3 He/ 4 He ratio of the groundwater in the region. Hence, to account for this 458 variation we also use the higher value of 1.1 Ra as a worst case scenario for assessing the 459 portion of noble gases originating from either the CO2 produced from, or injected into, the 460 Weyburn field. 461 The three Kerr groundwaters with below ASW 3 He/ 4 He ratios lie close to the two mixing lines 462 plotted. The concentration of 4 He measured in the fluids produced from the Weyburn field is 463 two orders of magnitude higher than the atmospheric value, as a result of crustal radiogenic 464 contributions from interaction with the crustal fluids present in the EOR field (Fig. 5). This is 465 reflected in the higher than ASW 4 He/ 20 Ne and lower than ASW 3 He/ 4 He ratios exhibited by 466 the produced CO2 gas sample (Fig. 4). Using the mixing model presented allows us to resolve 467 that the Kerr groundwater with the lowest 3 He/ 4 He could conceivably contain a maximum 0.14 468 % contribution to 4 He from the produced fluids using the best case 3 He/ 4 He endmember (1 Ra) 469 or a 0.25 % contribution using the worst case 3 He/ 4 He endmember (1.1 Ra). 470 of the calculated ASW range (Fig. 4). 484 Using the mixing model presented allows determination that the Kerr groundwater with the 485 lowest 3 He/ 4 He and highest 4 He/ 20 Ne ratios indicates a maximum of a 0.14 % contribution, in 486 the best case scenario, to 0.25 % in a worst case scenario, to the 4 He concentration from the 487 produced CO2. Figure 5 also shows that using the best case (1 Ra) and worst case (1.1 Ra) 488 3 He/ 4 He end member, a 20 % to 32 % contribution to the measured 3 He/ 4 He and 4 He/ 20 Ne of 489 the Kerr groundwaters could originate from the CO2 injected into the Weyburn field could 490 account for the lowest 3 He/ 4 He and highest 4 He/ 20 Ne ratio observed. it also has an above ASW 3 He/ 4 He ratio, and shows a 4 He/ 20 Ne ratio which is within the ASW 501 range. Additionally this groundwater sample does not have an elevated 4 He or a depleted 36 Ar 502 concentration compared to the predicted ASW range. 503 Recent work using noble gases to investigate the contamination of groundwaters by natural 504 gas from unconventional gas production in the USA has shown that well waters with high 505 methane concentrations, located close to gas production wells, have below ASW 506 concentrations of 20 Ne and 36 Ar (Darrah et al., 2014). This is a result of the fugitive methane 507 containing insignificant concentrations of groundwater derived 20 Ne and 36 Ar concentrations 508 and consequently the migrating methane 'strips' out these noble gases from the groundwaters. 509 This occurs as the noble gases are much more soluble in CH4 than in water and is identical to However, we are unable to rule out a best case possibility of a 0.14 %, and a worst case 535 possibility of a 0.25 %, contribution to the groundwater sample with the lowest 3 He/ 4 He ratio 536 from the produced fluids, or a 20 % to 32 % contribution from the injected CO2 to the sample 537 with the lowest 3 He/ 4 He ratio. The inability to firmly rule out a significant contribution to the 538 Kerr groundwaters from the injected CO2 is a key limitation of our study and is due to the low 539 helium concentrations measured in the injected CO2. This is most likely to be the result of the 540 solubility based capture method used to extract the CO2 from the gasification process, which 541 results in the majority of the insoluble 4 He being lost as it is not captured by the capture 542 technique and hence is vented with the non-captured flue gas (Flude et al., 2016). 543 As we find no evidence of a 4 He component above that of ASW we conclude that the below 544 ASW 3 He/ 4 He ratios are most probably the result of measured low 3 He concentrations, a 545 potential reflection of increased analytical error in measuring such small amounts of 3 He in the 546 waters. This is due to the extremely low concentration of 3 He in ASW of 60 to 120 parts per 547 trillion and highlights that 3 He/ 4 He ratios are not a robust means to assess the presence of, or 548 lack of presence of a radiogenic component in this study. We therefore recommend that future 549 investigations of this type focus on 4 He/ 20 Ne ratios, which is a more sensitive and robust 550 measure of the presence of a non-atmospheric radiogenic component and also avoids the 551 complication of elevated baseline 3 He/ 4 He ratios due to the presence of tritium derived 3 He. 552 Whilst we have included the injected CO2 end member to make a robust assessment of all of 553 the possible sources of CO2 near to the Kerr Farm it is not necessarily valid. The nearest CO2 554 injection well to the Kerr quarter is located some 1.4 km away and CO2 injection at this location 555 ceased in 2005 (Cenovus Energy Inc., 2011). The section of the Weyburn EOR field located 556 directly beneath the Kerr property has remained under water injection throughout extraction 557 operations of the oil field since the 1960's (Sherk et al., 2011). Hence, we believe it to be 558 unlikely that injected CO2 could migrate over 1.5 km laterally and through ~1.5 km of 559 overburden without encountering formation water containing an excess 4 He fingerprint 560 inherited from the radiogenic decay process. Given that CO2 is an excellent solvent (Warr et 561 al., 2015) and has been shown to strip out radiogenic noble gases from formation waters 562 In light of our findings, we recommend that further investigation into the composition of 568 captured CO2 is needed to quantify how useful noble gases will be in tracking injected CO2 569 within CO2 storage reservoirs and identifying how quickly the radiogenic fingerprint of the 570 storage reservoir is inherited by injected CO2. We also recommend that operators of CO2 571 injection sites establish both the geochemical baseline of their reservoir prior to CO2 injection, 572 and routinely monitor the geochemical fingerprint of the CO2 injected, including both stable 573 carbon and noble gas isotopic measurements. This comprehensive geochemical database 574 could then be used as a robust reference basis for geochemical investigations should 575 allegations of leakage be made. 576

Comparison with findings from other studies completed on the Kerr site 577
Our interpretation that there is no evidence of migration of the CO2 injected into or produced 578 from the Weyburn EOR field into the Kerr groundwaters is further corroborated by comparison 579 with the results of separate investigations into the allegations of CO2 contamination on the 580 Kerr property. 14 C measurements were a key component of the investigation instigated by the 581 field operators, Cenovus, by TRIUM Environmental Inc. This study analysed radiocarbon 582 within 78 samples of soil gases over the entire of the Kerr quarter and found that these 583 contained high levels of 14 C, indicating a recent carbon source. The 14 C values measured were 584 identical to those measured on a control site well outside of the Cenovus CO2-EOR operations, 585 whilst 14 C measurements from CO2 originating from the Dakota gasification plant and the 586 recycled gas injected into the Weyburn field showed that these contained no measurable 14 C 587 (Cenovus Energy Inc., 2011). This showed that the CO2 contained in the soil gases overlying the Kerr property had to have a recent, 'living' high 14 C source, rather than a 'dead' non 14 C 589 containing fossil fuel origin. 590 Further support to the lack of evidence of CO2 migration from depth is provided by the 591 relationship between the concentration of O2 and N2 with CO2 in the soil gas samples in both 592 the investigation performed on behalf of the field operators (Cenovus Energy Inc., 2011) and 593 that performed by IPAC-CO2 (Romanak et al., 2014). The Cenovus funded study found that 594 soil gas CO2 measurements for both the Kerr Quarter and two off site control localities were in 595 natural equilibrium with N2 and O2, providing an indication of the origin of the CO2. If the CO2 596 was from a natural biogenic soil respiration process, O2 is consumed to yield CO2 within the 597 soil due to plant and microbial respiration activity. As a result of this consumption of O2 its 598 concentration decreases, whilst CO2 is produced and its concentration increases. N2 599 concentration is unaffected in that natural process. In contrast, if the injected, industrial source 600

Conclusions 608
We conclude that the carbon isotope data do not constrain the origin of elevated dissolved 609 CO2 concentrations in the Kerr groundwaters, due to the lack of a distinct fingerprint between 610 the injected and produced CO2 relative to that of baseline values in the shallow subsurface in 611 the region. Our combined noble gas fingerprints show no evidence of the presence of noble 612 gases from the injected CO2, or from the CO2 produced from the Weyburn CO2 Enhanced Oil 613 Recovery field, within the groundwaters surrounding the Kerr property. All of the Kerr 614 groundwater samples exhibit noble gas fingerprints which would be expected in shallow 615 groundwaters and show no evidence for the addition of a deep radiogenic component or 616 dilution from the addition of a gas phase low in atmospheric derived noble gases. 617 However, we are unable to categorically rule out a best case possibility of a 0.14 %, and a 618 worst case possibility of a 0.25 %, contribution to the groundwater sample with the lowest 619 probably due to the solubility based capture method used to capture the CO2. However, we 624 believe it is unlikely that injected CO2 could migrate from the nearest CO2 injection well (some 625 1.5 km away, and at a depth of 1.5 km) without encountering formation water rich in radiogenic 626 noble gases and inheriting a radiogenic fingerprint high in 4 He and with a low 3 He/ 4 He ratio, 627 similar to that of the produced CO2. 628 We also determine that 4 He/ 20 Ne ratios are a more robust indicator of the lack of a deep 629 radiogenic component in the Kerr groundwaters than 3 He/ 4 He ratios. This is due to a Hence, it is clear that the integration of the full range of geochemical tracers (stable carbon 645 and 14 C isotopes, noble gases, water chemistry, process based gas ratios) is the most 646 effective means to understand the CO2 source and refute the leakage allegations made at the 647 Kerr Farm. Therefore, future investigations into allegations of CCS related CO2 leakage should 648 use a similar comprehensive range of geochemical tools and integrate them with a good 649 understanding of geological and engineering data at the site. 650

Acknowledgements 651
We acknowledge the helpful and constructive reviews from five anonymous reviewers that 652 commented on a previous related manuscript and two anonymous reviewers who provide 653 constructive comment on this version. We thank Thomas Ogilvie, Dr Janis Dale and Karen 654 Collins for help in field collection of the water samples. We further acknowledge the 655 cooperation of Cameron and Jane Kerr along with Ian and Sheila Thackeray, property owners, 656 Goodwater, SK, for providing access to their property and allowing us to collect water samples.   The 4 He/ 20 Ne ratio of ASW is well constrained at 0.15 to 0.17 under the recharge conditions 865 experienced in the summer in Saskatchewan. High 20 Ne content relative to 4 He is a strong 866 indicator of atmospheric input. The produced CO2 and injected CO2 exhibit 4 He/ 20 Ne well 867 above those of ASW. All analytical error bars are smaller than printed symbols. 868 Mixing lines on the plot depict the trend which would result from mixing a groundwater with 872 the best case ASW 3 He/ 4 He ratio of 1 Ra and the average 4 He concentration measured in the 873 Kerr groundwaters, with the CO2 injected into and produced from the Weyburn field. All 874 analytical error bars are smaller than printed symbols. 875