Environmental Geology

, Volume 47, Issue 4, pp 576–585

Geologic storage of carbon dioxide as a climate change mitigation strategy: performance requirements and the implications of surface seepage


  • R. P. Hepple
    • Lawrence Berkeley National LaboratoryEarth Sciences Division
    • Lawrence Berkeley National LaboratoryEarth Sciences Division
Original Article

DOI: 10.1007/s00254-004-1181-2

Cite this article as:
Hepple, R.P. & Benson, S.M. Environ Geol (2005) 47: 576. doi:10.1007/s00254-004-1181-2


The probability that storage of carbon dioxide (CO2) in deep geologic formations will become an important climate change mitigation strategy depends on a number of factors, namely (1) public acceptance, (2) the cost of geologic storage compared to other climate change mitigation options, and (3) the availability, capacity, and location of suitable sites. Whether or not a site is suitable will be determined by establishing that it can meet a set of performance requirements for safe and effective geologic storage. To date, no such performance requirements have been developed. Establishing effective requirements must start with an evaluation of how much CO2 might be stored and for how long the CO2 must remain underground to meet goals for controlling atmospheric CO2 concentrations. Answers to these questions provide a context for setting performance requirements for geologic storage projects.

According to the results presented here, geologic storage could be an effective method to ease the transition away from a fossil-fuel based economy over the next several centuries, even if large amounts of CO2 are stored and some small fraction seeps from storage reservoirs back into the atmosphere. An annual seepage rate of 0.01% or 10-4/year would ensure the effectiveness of geologic carbon storage for any of the projected sequestration scenarios explored herein, even those with the largest amounts of storage (1,000 s of gigatonnes of carbon-GtC), and still provide some safety margin. Storing smaller amounts of carbon (10 s to 100 s of GtC) may allow for a slightly higher seepage rate on the order of 0.1% or 10-3/year. Based on both the large capacity of geologic storage formation and the likelihood of achieving leakage rates much lower than the rates estimated here, geologic storage appears to be a promising mitigation strategy.


Carbon sequestrationCarbon storagePerformance requirementsMitigationLeakage and seepage


While the basic physics and principles of climate forcing are well established, forecasts of global climate evolution are highly uncertain. Atmospheric CO2 concentrations have increased from an estimated 180 parts per million by volume (ppmv) 25,000 years ago, during the most recent glacial maximum, to 280 ppmv 200 years ago, to the current concentration of over 370 ppmv (Houghton and others 2001). In parallel with current scientific research is a diverse and evolving body of policy options for dealing with climate change through preventive, mitigation, remediation, and adaptive measures. In the long-term, fuel switching to lower or noncarbon fuels, hydrogen as an energy carrier, renewable sources of energy, efficiency improvements, and energy conservation are the most promising alternatives. Even under the most aggressive projections of technology development and progressive policy regimes, the transition from the current dependence on fossil fuels would take many decades or longer. Predicting the shift in energy usage is also complicated by the uncertain factors of population and economic growth (Metz and others 2001; McCarthy and others 2001).

In the meantime, potential global climate impacts associated with carbon inputs to the atmosphere must be addressed. There is no known or expected panacea to negate the climate change issue, so any response must inevitably include a broad array of policy measures and technology options. The sequestration of carbon by various means is one class of mitigation strategies, which has great near-term promise as a way to hedge against the potential severity of consequences and to buy time to develop further currently uncertain options.

Sequestration is the effective removal of carbon from the atmosphere and its storage in forms such as dissolved carbon in the ocean or subsurface formation waters, in terrestrial biomass, as a fluid in deep geologic traps, or as mineral carbonates. Of the three basic types of sequestration – ocean, terrestrial, and geologic, the most mature technology at this time is the geologic storage of CO2 in underground formations such as depleted or nearly depleted oil and gas reservoirs, deep sedimentary brine-filled formations, and deep unmineable coal beds (Keith 2000; Holloway 2001). In geologic carbon storage, carbon dioxide would be captured from large point sources such as power plants, compressed, transported by pipeline, and injected into geologic reservoirs. Some pilot projects are already underway with Statoil’s Sleipner Saline Aquifer Carbon Dioxide Storage (SACS) Project in the North Sea, the Weyburn enhanced oil recovery project in Alberta, Canada, and several U.S. Department of Energy sponsored projects in the U.S. (NETL 2001; IEAGHG 2001).

To address the question, “How much CO2 might be stored underground and for how long?” zeroth-order estimates for the annual amount of CO2 that would need to be sequestered to meet atmospheric stabilization targets of 350, 450, 550, 650, and 750 ppmv were developed. The difference was calculated between (1) the six IPCC SRES (Intergovernmental Panel on Climate Change, Special Report on Emissions Scenarios) marker emissions scenarios (Nakicenovic and others 2000) and (2) the economically optimized Wigley, Richels and Edmonds-(WRE) (Wigley and others 1996) allowable emissions for a range of long-term atmospheric CO2 stabilization targets. For each of the scenarios, the assumption was that geologic sequestration would be used as a bridging technology, allowing for the gradual phase out of fossil fuels over a period of up to 300 years. Because the SRES emissions scenarios made projections of emissions over only the first 100 years, 300-year emissions scenarios were created by extrapolating linearly from the year 2100 emissions of the SRES scenarios to the approximately steady-state emissions rates for stabilizing atmospheric CO2 at target concentrations in the year 2300. The large amount of carbon to sequester or the increasing trend of emissions for all scenarios, except A1T and B1 (Nakicenovic and others 2000), made this extrapolation necessary, as did the desire to explore the most conservative and far-reaching implications. The SRES emissions scenarios were used as the basis for this study because their express purpose was to act as a means of standardizing climate change modeling for comparison. An additional assumption was that geologic storage constitutes the only mitigation outside of the socioeconomic and demographic parameters included in the emissions scenarios, such as variable technology mixes, rates of development, and the strength of the movement toward global environmental and sustainability ethics. This assumption was made explicitly to avoid the modeling complexity associated with integrated assessment models that consider portfolios of mitigation options. This is a first order feasibility and risk assessment that explores potential amounts of carbon to sequester to estimate performance requirements for CO2 seepage.

To address a second important question, “What would be an acceptable surface seepage rate?” the authors calculated the rate at which CO2 might seep back to the surface, based on the simple conceptual model illustrated in Fig. 1. It was assumed that the amount of seepage would be proportional to the total amount of CO2 stored underground at any given time. To determine what would be an acceptable seepage rate, the calculated seepage was compared to the allowable emissions for stabilization of atmospheric CO2 at 350, 450, 550, 650, and 750 ppmv. When the estimated seepage was small compared to target allowable emissions, it was concluded that geologic storage would be an effective means for mitigating net greenhouse gas emissions to the atmosphere.
Fig. 1

Conceptual model of seepage for evaluating effectiveness

Note that surface seepage is not necessarily equal to the rate at which CO2 leaks from the primary storage reservoir, as the existence of stacked reservoirs (Johnson 1990; Allis and others 2001) and natural seeps (Hodgson 1980; Kvenvolden and Harbaugh 1983) demonstrates. Many subsurface processes such as solubility trapping, mineralization, diffusion, and residual gas trapping will attenuate the migration of CO2 as it moves towards the land surface. Therefore, performance requirements for surface seepage rates should not be construed as performance requirements for leakage from the primary storage reservoir. Setting performance requirements for the latter is substantially more complex and requires more careful consideration of the physical and chemical processes that occur as CO2 migrates through the subsurface, as well as of a number of legal and regulatory issues. In this study, only a single criterion was considered for assessing acceptable seepage rates, namely the effectiveness of geologic storage for mitigating greenhouse gas emissions into the atmosphere. Use of additional criteria, based on risks to human health or the environment, may change the acceptable seepage rates derived below, e.g., to avoid hazardous accumulations from localized seepage in low-lying or confined spaces such as basements (Benson and others 2002).

As in any forecasting, there is uncertainty in the present study. Nevertheless, this work stands as an attempt to establish a starting point in the discussion over performance requirements for geologic storage of CO2. To explore the implications of uncertainty for these two basic questions, “how much carbon may need to be stored underground and what seepage rate of CO2 back to the atmosphere would be acceptable?” The authors examined the sensitivity of these conclusions to different assumptions about storage amounts and “acceptable” seepage rates for injection periods of 50, 100, 200, and 300 years.

Model and methods

Model scenarios were generated from the SRES anthropogenic emissions scenarios shown in Fig. 2a, with linear extrapolations from projected anthropogenic emissions levels in the year 2100 to the long-term, decaying tail of allowable emissions levels for a given target stabilization concentration in the year 2300. These emissions levels are roughly 0.9 gigatonnes of carbon per year (GtC/year) for 350 ppmv, 1.9 GtC/year for 450 ppmv, 2.7 GtC/year for 550 ppmv, 3.5 GtC/year for 650 ppmv, and 4.3 GtC/year for 750 ppmv (9). (1 GtC = 3.667 GtCO2). The WRE allowable emissions were taken from an impulse-response function of a simple carbon cycle model, and the shape of that curve was economically optimized under simple assumptions. Long-term allowable emissions were zeroth-order estimates because in principle, they decay to zero over many hundreds of years, and because processes included in the carbon cycle model contain significant uncertainties (e.g., the projected decrease in the North American terrestrial carbon sink). Curves for a range of atmospheric CO2 stabilization targets are shown in Fig. 2b. The annual amount to store in a given year (S) was equal to annual anthropogenic emissions (E) for a given scenario, i (A1B, A1F1, A1T, A2, B1, or B2), according to (8), minus allowable emissions (T) for a given stabilization target, j (350, 450, 550, 650, or 750 ppmv), according to Eq. (1):
Fig. 2a–f

(a) IPCC SRES illustrative and marker emissions scenarios (Nakicenovic and others 2000); (b) WRE stabilization curves, which are allowable annual emissions for target concentrations of atmospheric CO2 (Wigley and others 1996); (c) Annual sequestration for A1B/550 ppmv; (d) Cumulative amount of carbon stored for A1B/550 ppmv; (e) Annual seepage from stored carbon for A1B/550 across a range of simple linear seepage rates and compared to allowable emissions for stabilization at 550 ppmv; (f) Effective time frame for simple linear seepage rates as the percentage of cumulative carbon stored for A1B/550 ppmv

$$ S{\left( t \right)}_{{ij}} = E{\left( t \right)}_{i} - T{\left( t \right)}_{j} ;{\text{ if }}E < T,S = 0 $$
where t is time in years.
The model for seepage was in the form of Eq. (2), where seepage in a given year, L(t), equaled a rate constant, r, times the cumulative amount of carbon remaining underground at the end of the previous year, C(t-1). The rate constant, r, was explored over three orders of magnitude and set to either 1% (10-2), 0.1% (10-3), 0.01% (10-4), or 0.001% (10-5) per year.
$$ L{\left( t \right)}_{{ij}} = r{\left[ {C{\left( {t - 1} \right)}_{{ij}} } \right]} $$
The cumulative amount of carbon stored at the end of a given year was calculated using Eq. (3), where I(t) was the amount injected during the year in question. The authors used an iterative arithmetic model instead of a first order rate equation because of the substantial effects of considering leakage and seepage over the period of CO2 injection (e.g. 300 years).
$$ C{\left( t \right)}_{{ij}} = C{\left( {t - 1} \right)}_{{ij}} + I{\left( t \right)}_{{ij}} - L{\left( t \right)}_{{ij}} $$

To illustrate this methodology, the SRES scenario A1B and a 550 ppmv target was used. The combination of scenario A1B with a 550 ppmv stabilization target was not chosen on the basis of probability or desirability and is not endorsed in any way. All scenarios were considered equally probable, and no determination of safe or reasonable stabilization targets has yet been made. Figure 2c–f provide graphs for this scenario of the annual storage rate (S), the cumulative amount of carbon stored (C), the annual seepage (L), and the amount of carbon remaining underground over a 1,000-year period. These graphs show that the maximum annual sequestration rate would be about 7 GtC per year and that the cumulative amount of carbon stored would be about 1,000 GtC. In addition, Fig. 2e compares annual seepage to allowable emissions for the 550 ppmv target, which demonstrates that for seepage rates of 1% per year, seepage would be higher than the allowable emissions for the period from 2150 to 2280. Thus, in this scenario, geologic storage would not be effective if 1% seeped annually. On the other hand, for all of the other seepage rates (0.1 to 0.001% per year), seepage would be well below the allowable emissions, indicating that storage could be effective. For these lower seepage rates, the maximum annual amount of seepage is roughly equal to steady-state seepage for centuries after injection stops. Figure 2f shows the amount of carbon remaining underground over a 1,000-year period. For two of the cases (seepage rates of 0.01 and 0.001% per year), 90% and 99%, respectively, of the carbon would remain underground after 1,000 years. For a seepage rate of 1% per year, most of the carbon would return to the atmosphere after 400 years, again demonstrating that geologic storage would not be effective if seepage rates were this high.


Calculations such as those illustrated in Fig. 2 were made for all permutations of scenarios and stabilization targets discussed earlier. The results are detailed in Figs. 3 and 4 and discussed below.
Fig. 3a,b

(a) Target amounts to sequester (horizontal bars indicate averages), (b) Capacity estimates based upon diverse criteria (Metz and others 2001; Beecy and others 2001; U.S. DOE 2001)

Fig. 4a–d

Maximum annual surface seepage for four different seepage rates (horizontal bars indicate tails of allowable emissions as shown in Fig. 2b; Wigley and others, 1996), (a) 0.001% per year, (b) 0.01% per year, (c) 0.1% per year, (d) 1% per year. Stabilization targets in ppmv

Total amount of carbon to sequester

Figure 3 shows that target amounts of carbon to sequester for 300-year scenarios varied from 0 to 4,500 GtC, and averages for the five different stabilization targets ranged from 900 to 2,500 GtC. Most scenarios required some storage, including all scenarios for stabilization at 350, 450, or 550 ppmv, and the moderate to heavily fossil fuel-intensive scenarios for stabilization at 650 and 750 ppmv. As seen in Fig. 3a, the total amounts of carbon stored vary roughly by a factor of two across both the emissions scenarios (e.g., 810 to 4,500 GtC at 350 ppmv) and stabilization targets (e.g., 500 to 2,400 GtC for A1B). Figure 3b shows a range of geologic storage capacity estimates for the purpose of comparison. The low capacity estimate includes projections of relative costs; the mid-range is based upon enclosed trapping structures or lower bound estimates; the high capacity estimate is based upon upper bounds and total pore volume available in the subsurface in the desired depth range, roughly 800 m to 2 km. As shown by the comparison between Fig. 3a and b, the storage capacity of geologic formations is of a similar order of magnitude to the potential emissions of the scenarios examined here.

Surface seepage of sequestered CO2

The amount of surface seepage of CO2 depends on the scenario selected and the assumed seepage rate. In Fig. 4, the WRE allowable emissions levels are indicated by red horizontal bars. As shown, with few exceptions, seepage rates of 1% per year are unacceptably high. For stabilization at 350, 450, and 550 ppmv, seepage rates must be less than 0.01% per year to be acceptable for all of the scenarios. At 650 and 750 ppmv, seepage rates less than 0.1% per year could meet the criterion of acceptable seepage. Under the simple seepage model used herein, seepage is a linear function of the amount stored. Therefore, the more carbon stored, the more important the seepage rate becomes.

Sensitivity of “acceptable” seepage rates to varying time frames and amounts of carbon to sequester

Defining acceptable seepage rates for shorter periods of active injection lasting only 50, 100, or 200 years results in a slightly different conclusion because of the smaller total amounts of carbon being stored, but reinforces the order-of-magnitude estimates established during the preliminary analysis for 300-year scenarios in the preceding section. Under scenarios with medium to large amounts of sequestration (B2, A1B, A2, A1F1), the first 50 years account for approximately one-tenth of the 300-year total amount of carbon to sequester. For the scenarios with lower amounts of required carbon storage (B1, A1T), the first 50 years account for roughly one-quarter to one-half of the 300-year totals. The breakdown of amounts to sequester by period is shown in Fig. 5, and the consequences on the seepage rate of varying the amount to sequester are shown in Fig. 6. Seepage rates deemed acceptable range from 1% to 0.01% per year, with sequestration scenarios requiring no storage denoted as NS. The trend in seepage rates as a function of time or amount of carbon to sequester is clear: the less carbon stored, the higher the acceptable seepage rate may be, as long as the criterion is a fixed number, e.g., 0.5 or 0.1 GtC/year. The case where the cutoff for “acceptable” seepage was altered from the values of the WRE tails to 0.5 or 0.1 GtC/year is shown in Fig. 7 for IPCC emissions scenario A1B. The more stringent the cap on total seepage and the greater the amount of carbon stored, the lower the acceptable annual seepage rate must be, in this case as low as 0.001% or 10-5/year when more than 1,000 GtC are stored.
Fig. 5

Total target sequestration for all combinations of emissions scenarios and stabilization targets, plus the average (AVG) for each target, broken down by time periods as indicated. Stabilization targets given in ppmv

Fig. 6.

Seepage rates in percent per year to meet cutoffs at 300 years from (Wigley and others 1996). NS = no sequestration required. Stabilization targets given in ppmv

Fig. 7a–c

“Acceptable” seepage rates or amounts for scenario A1B using three different definitions: (a) WRE curves at 300 years as an approximate steady-state (Wigley and others 1996), (b) 0.5 GtC/year, and (c) 0.1 GtC/year. “Unacceptable” rates are white text, highlighted in gray, and “acceptable” rates are shown in black text

For the sake of simplicity, one of the most significant assumptions in this thought experiment was that geologic CO2 storage would be the only explicit climate change mitigation policy. Because this is so unlikely, the more realistic case is also considered where geologic storage accounts for only 30% of excess CO2, which is the proportion of CO2 emissions that currently comes from large point sources such as power plants and refineries in the U.S. (This proportion varies from country to country, and in absence of technology breakthroughs or large-scale conversion to non-fossil energy sources, it is likely to increase in the future.) This is shown in Fig. 8, which includes the amounts and cutoffs for acceptable seepage according to the three criteria discussed previously. With smaller amounts of CO2 being stored, the acceptable seepage rate is correspondingly larger at 0.1% or 10-3. In this case, the authors assume that the other mitigation options are 100% efficient; if their efficiency were the same as geologic CO2 storage, then there would be no difference between the scenarios depicted in Figs. 7 and 8.
Fig. 8a–c

“Acceptable” seepage rates or amounts for scenario A1B using three different definitions with the assumption that geologic CO2 storage accounts for 30% of excess emissions: (a) WRE curves at 300 years as an approximate steady-state (Wigley and others 1996), (b) 0.5 GtC/year, and (c) 0.1 GtC/year. “Unacceptable” rates are white text, highlighted in gray, and “acceptable” rates are shown in black text


First, the scenarios generated here provide order-of-magnitude estimates of the amounts of carbon that may need to be stored underground. The actual quantities of CO2 eventually stored may be smaller because of a shorter period of intentional carbon storage or because geologic storage does not compete well with other mitigation options. Yet even the large projections – 100 s to 1,000 s of GtC – are in the range of estimated global geologic storage capacity (Metz and others 2001; Beecy and others 2001; U.S. DOE 2001). Well-characterized oil and gas reservoirs are likely to achieve these target seepage rates and could accommodate much of the volume of emissions over the coming decades. However, total storage requirements will eventually exceed the capacity of oil and gas reservoirs for most scenarios of 100% geologic carbon storage, and due to the worldwide distribution of oil and gas reservoirs, many countries do not have this option even now. Consequently, large amounts of CO2 may need to be stored in deep saline formations, which are currently poorly characterized compared to oil and gas reservoirs. Significant effort will be needed to assess and utilize the apparently large storage capacity of deep saline formations. At the same time, the learning curve associated with experience should improve the ability to locate and assess high-quality storage sites. The tradeoffs in terms of time, storage space, experience, and the moving target of “effectiveness” clearly require further research and in the coming decades will evolve along with mitigation technologies and policies.

For seepage rates of less than 0.01% per year, geologic sequestration would be effective for mitigating the buildup of atmospheric CO2 for all of the scenarios evaluated. At these low seepage rates, the maximum annual seepage never exceeds 0.5 GtC/year for any of the projected sequestration scenarios. For comparison, the total estimated worldwide volcanic and magmatic degassing is estimated to be 0.07 to 0.13 GtC/year (Benson and others 2002), and the estimated long-term, steady-state level for stabilization at 350 ppmv is less than 1 GtC/year. In addition, a 0.01% per year seepage rate would ensure that at least 90% remained effectively sequestered after 1,000 years. Because seepage rates less than 0.01% per year meet several criteria for all scenarios, this may be a reasonable long-term global performance requirement for surface seepage, especially in light of the precautionary principle integral to public health and environmental risk assessment. Constrained only by potential health, safety, and environmental considerations, larger seepage rates could be tolerated under shorter periods of intentional storage, shorter goals for the target time of sequestration from the atmosphere, smaller amounts of carbon storage, and in all cases a learning period of up to several decades.

The requirement of 0.01% per year seepage rates from CO2 storage reservoirs also appears to be achievable when compared to natural hydrocarbon seepage rates from oil and gas fields. For example, worldwide hydrocarbon seepage is estimated to be 0.25 Million tonnes per year (Mt/year) plus or minus one order of magnitude, and this amount is three orders of magnitude less than the potential cutoff of 0.5 GtC/year (500 MtC/year). Also, the mean residence time for natural hydrocarbons is estimated to be at least 50 million years, indicating natural seepage rates far below the rate of 10-4 contemplated here (Kvenvolden and Harbaugh 1983). Although the ability of saline formations to retain buoyant gases for long periods of time has yet to be demonstrated, it is expected that detailed site-specific studies and careful selection protocols can be used to identify sites with seepage rates that meet the performance requirements identified above. In the end, the safety and effectiveness of engineered materials and systems such as injection wells and unknown plugged and abandoned wells may be the determining factors of actual seepage rates (Celia and Bachu 2002). However, regardless of whether performance requirements are being set for natural, man-made or both, defining such requirements will be needed to guide site selection and regulatory permitting.

According to the results presented here, geologic storage could be an effective method to ease the transition away from a fossil fuel-based economy over the next several decades to centuries, even if large amounts of CO2 are stored and some small fraction seeps from storage reservoirs back into the atmosphere.


The authors would like to acknowledge helpful comments from and discussions with colleagues Curt Oldenburg, Larry Myer, Chin-Fu Tsang, David Keith, and Peter Cooke. This work was supported by the Laboratory Directed Research and Development Program at Lawrence Berkeley National Laboratory under DOE Contract No.AC03–76SF00098.

Copyright information

© Springer-Verlag 2004