The purpose of this workshop, held May 30-31, 2013, was to take a broad look at the risks, actualized or potential, associated with the development of shale gas resources.1 To identify the risks to be examined, the committee organizing the workshop considered both existing accounts of these risks and the public discourse and controversy about shale gas development. It also conducted its own elicitation of concerns about such risks by inviting input from a variety of individuals and groups that may have an interest in or a concern about shale gas development. Because it was not feasible to go into detail on all the types of risks that have raised concern, the organizing committee had to be selective and had to group some types of concerns into broader categories for the purpose of inviting presentations for the workshop.
The National Research Council (NRC) project director for the two workshops, Paul C. Stern, opened the workshop by describing its purpose and thanking the sponsors. Mitchell Small, chair of the steering committee for the workshops, distinguished the purposes of the two workshops. He described this initial workshop as being about characterizing the risks and the second as being about options for risk management and governance. He quoted the charge to the committee as follows: “This project will point the way to a risk-analytic approach aimed at more adequately informing
1The agenda for this workshop, the speakers’ abstracts and slide presentations (in PDF format), and video archives of the presentations and discussions are available at: http://sites.nationalacademies.org/DBASSE/BECS/CurrentProjects/DBASSE_069201 [July 2014].
public choices, will suggest governance models that hold promise for meeting the challenges of environmental protection in an era of declining regulatory capacity, and will direct attention of the energy policy community to the need to include fundamental social challenges—not just technological ones—in the development of policies and best practices.”
Small spoke briefly about the widespread presence of shale gas globally, noting that the U.S. Energy Information Administration now projects that by 2040, shale gas will account for about half of U.S. natural gas production—a considerably higher projection than was made 15 years ago. The expectation also is for well densities of four-eight wells or more per square mile in areas where development occurs. The workshop will be looking at the implications of this density of development for various kinds of risks, he said.
He summarized the agenda for the workshop and said that the presentations would review the literature on the risk topic being covered; characterize the risks, including who is exposed, the routes of exposure, and the endpoints; discuss the methods used for estimating the risk; summarize evidence on the magnitude and attribution of the risk; and identify the key uncertainties and the kinds of studies needed to reduce them. He outlined the procedures for the workshop and encouraged all participants to maintain a tone of shared search for understanding, even though the topic is increasingly controversial.
Webler, a researcher who specializes in collaborative processes for environmental decision making that engage experts with interested and affected parties, presented work he conducted at the request of the organizing committee. His collaborators on this work were Andrei L. Israel of Pennsylvania State University, Gabrielle Wong-Parodi of Carnegie Mellon University, and Paul C. Stern of the NRC. Webler began by characterizing shale gas extraction, with its combination of hydraulic fracturing and horizontal drilling, as involving an emerging technology that presents multiple types of risk. Citing the Understanding Risk report (National Research Council, 1996), he noted that to understand a risk issue, it is first necessary to characterize the risks, beginning by identifying the risk concerns of the interested and affected parties. He defined these parties as elements of the public that had educated themselves on the technology and that might be acutely aware of potential risks.
Webler noted that there are multiple ways to identify risk concerns. Resources for the Future (RFF) has conducted a study of experts’ concerns, reported later in the workshop, and Yale University recently released results of a survey of the general public. The results reported by Webler came from a different approach: seeking input from the interested and affected parties. Limited time and budget resulted in the use of an Internet-based survey that began with a search of Facebook and the Internet for groups believed likely to be concerned with shale gas development: 24 local antishale gas development groups, 17 regulatory agencies, 7 gas company groups, 6 groups in the consumer gas industry, and several additional groups in the finance industry, the energy media, or the renewable energy industry. Contact persons in these groups were identified and invited to send the elicitation instrument to anyone they thought would be interested.
The instrument asked two open-ended questions about shale gas concerns and topics that the respondent wanted to know more about. It also asked a few questions to determine the respondents’ states of residence and their connections, if any, to the industry. Overall, 372 responses were received, with a very wide range in format, tone, and level of detail; 40 percent of responses came from New York, 16 percent came from Ohio, and fewer were from other states. The great majority of respondents (76%) were from the four Marcellus shale states of New York, Ohio, Pennsylvania, and West Virginia, despite the effort to get respondents nationwide. Just over half (56%) of the respondents were members of groups opposing shale gas development; only small numbers were from members of groups supporting the industry or from employees of the industry.
Webler characterized the responses as falling into the five broad categories identified by Kasperson and colleagues (1988): hazards (e.g., fracking fluids), hazardous events (e.g., spills), the consequences of such events (e.g., ground water contamination), precursors (e.g., poor regulations), and risk amplifiers (e.g., obfuscation of information). The responses were examined using the constant comparison method of Glaser and Strauss (1967) for analyzing qualitative data: the responses were divided into codable data segments (2,567 in all), after which the three coders met as a group to develop a list of codes (which numbered 131) and then coded the data segments.
Although the sample was not representative of a specific population, the research group considers it meaningful in the sense that concerns that are frequently mentioned by interested and affected parties are important to investigate. It is also possible, of course, that concerns that are not mentioned frequently may also be important to investigate. The slide presentation summarized the frequencies of mention of concerns in each
of the five categories; consequences of hazardous events were the most frequently mentioned category.
Webler summarized the 131 coded concerns as falling under 9 themes:
- quality-of-life concerns (mentioned by 25% of respondents)—loss of rural character, crime, loss of beauty, community conflict;
- economic impacts (18%)—loss of property value, disruption to existing businesses;
- impacts distant from well sites (24%)—earthquakes, injection wells, wastewater treatment and disposal;
- climate change (17%)—including effects on renewable energy and overall energy consumption;
- quality and availability of information (18%)—insufficient disclosure, obfuscation of information;
- regulations and regulatory capture (46%)—poor regulations, flawed or biased science, inadequate oversight, exemptions from laws;
- ethics and environmental justice (10%)—procedural and distributive injustice concerns;
- wasted water resources (13%); and
- ecosystem and domestic animal impacts (22%)—such as effects on wildlife and domestic animals and habitat fragmentation.2
Webler concluded by saying that the respondents, some of whom had given a lot of thought to the issues, had raised a wide variety of concerns, some of which have already received careful analytical attention while others, such as quality-of-life and justice concerns, have not. He noted that the concerns go beyond what has been called NIMBY-ism;3 they include concerns about climate, ecosystems, and impacts distant from well sites. Also, many respondents expressed lack of trust in existing institutions to do what is needed to reduce risks. He noted some overlap between the concerns raised by these respondents and the expert concerns in the RFF study (e.g., concerns with air and water quality) but also suggested that the overlap may not be complete. Webler said that all these approaches may contribute useful information about concerns related to shale gas development.
2Habitat fragmentation, as the term is used here, refers to situations in which the preferred environment of a population of organisms is physically divided, for example, by land clearing or road construction, so that the population itself becomes divided.
3“NIMBY” stands for “not in my backyard” and refers to expressions of concerns about having undesirable activity close to one’s home, as opposed to concerns about the same activity at any location.
Several issues arose in the discussion after the presentation. One concerned method that might be used to “take the pulse” of stakeholders, including content analysis of local and national media, examination of social media, and the use of semistructured interviews to explore the ways that opponents and proponents of shale gas development think about the issues. There were also related questions about whether sampling of stakeholders was representative. Alan Krupnick (RFF) said there were great contrasts between this study and the RFF study, which he thought might be due to the fact that RFF tried to get wide representation from academics, government, industry, and nongovernmental organizations (NGOs). Another participant suggested that the use of Facebook to get respondents might have overweighted the sample toward the Northeast.
A third issue was whether concerns differed in different states. Krupnick noted that in RFF’s 1,500-person public survey, which was still being analyzed at the time of the workshop,4 there appear to be both similarities and differences in stakeholder concerns across states. For example, respondents in Pennsylvania and Texas seem to have similar attitudes about effects on ground water, but surface water is more of a concern in Texas, and habitat fragmentation is a big issue in Pennsylvania. Webler noted that in the elicitation study, there were some state-specific concerns, such as with impacts on the wine and tourism industries in New York and with wastewater issues in Ohio. A fourth issue was the possibility of systematic differences between concerns of people in places that have had and have not yet had direct experience with the industry and shale gas technologies.
One participant asked about concerns among people who believed that their own health had been harmed by shale gas development. Andrei Israel, one of Webler’s collaborators, said that respondents to this elicitation had expressed concerns about health generally, rather than specifically about their own health.
Nygaard, an engineer and senior stimulation consultant with the ExxonMobil Production Company, focused his comments on four key
4A brief report of the results of the survey is now available at http://www.rff.org/RFF/Documents/RFF-Resources-185_Feature-Krupnick,%20Siikamaki.pdf [May 2014].
topics: responsible operations, well construction and design, operational integrity, and potential impacts.
Responsible operations. Nygaard said that engineers use the term “fracturing” to refer to one part of a larger process to which the public typically applies that term. The larger process includes drilling, fracturing, extraction, and completion, and it occurs over a longer period of time than fracturing in the engineering sense. He said the keys to success in responsible operations are to identify the risks, the possible consequences, and the probabilities of their occurrence; to account for uncertainties (for example, in the subsurface conditions); to collaborate with stakeholders and regulators to share information and develop a common frame of reference; and to generate opportunities for meeting energy demand, producing jobs and revenue, and reducing environmental emissions.
Nygaard noted that the industry has been doing horizontal drilling for 20 years. What is new, he said, is development at large-scale and in high-population areas, which brings increasing public notice to the process. The American Petroleum Institute has published several documents showing recommended practices for well construction and operation; if these are followed, Nygaard said, wells can be operated safely. The challenge occurs when equipment is used outside contingent performance limits or if procedures are not followed. Regulations are critical, and local and state regulation are especially so because local geology varies, Nygaard continued. Strong company practices and policies are also necessary for risk management, including high standards and management accountability for adhering to regulations, the use of good engineering judgment, and employee training.
Well construction and design. These processes are dependent on local geology and local gas or oil resources. Water utilization and disposal also vary greatly by local geology. For these reasons, Nygaard stressed that local regulation is important. He noted that the surface footprint of wells is relatively small: a pad of about 3-5 acres on the surface allows access to 1-2 square miles of underground formations. He described the volumes of materials required for a typical well: about 8 Olympic-size swimming pools of water,5 about 20 rail cars of proppant materials (mainly sand to
5According to the Fédération International de Natation facilities rules, a swimming pool for the Olympic Games is 50 meters long between the touch panels, 25 meters wide, and a minimum of 2 meters deep (3 meters recommended) (see http://www.fina.org/H2O/docs/rules/facilities_20132017.pdf [July 2014]). Assuming a 2-meter depth, one such pool would contain about 2,500 cubic meters of water. Eight such pools would contain 20,000 cubic meters (20 million liters) of water, which is equal to about 706,000 cubic feet or 5.3 million gallons.
open small cracks and enable production), about 6 truckloads of chemical additives, and about 20-30 truckloads of stimulation equipment. He said that in well construction, the industry’s primary goal is to protect ground water resources. This is done by using multiple barriers to mitigate the risk (for example, multiple cement casings), and by designing for well completion at the start of the process. The industry follows robust mechanical engineering design approaches for pressure vessels in designing wells and custom-designs every well to fit the resources and local regulations. The industry, Nygaard continued, also strives to minimize the use of chemicals as much as possible, consistent with achieving a project’s technical objectives. He said that chemicals are used only as needed to mitigate corrosion and address other engineering issues, such as reducing friction. Extensive diagnostics are used to check on the fractures and the flowback.
Operational integrity. Nygaard said that the industry uses a range of approaches for monitoring casing and cement placement in a well as it implements treatments. He said that monitoring of a treatment may take 4-5 hours of continuous monitoring, with a commitment to shut down a well immediately if anomalous behavior is noticed.
Potential impacts. Nygaard referred to a recent study by George King (2012), which examined the probabilities and consequences of 21 different possible events and assessed the risk levels for each. Nygaard emphasized that potential ground water contamination is an issue in the public eye. He reported on one study (Fisher and Warpinski, 2012) that indicated that the heights of the tops of fractures in microseismic events were well below the depths of ground water in various shale gas plays and a second study that examined 396 documented incidents in Texas and Ohio and found a great decline in frequency of microseismic events over time.6 Nygaard noted that pit lining contributed to a large number of these events, indicating the importance of monitoring.
Induced seismicity has also been raised as an issue, Nygaard said. This can be triggered by changes in stress states near faults, he explained, adding that industrial operations of many kinds, not only shale gas extraction, can run this risk. He cited an NRC study (National Research Council, 2013) as having evaluated these risks and found low risk, though he noted that there can be unique high-risk cases.
Nygaard summarized his presentation by saying that each shale play is unique and requires unique solutions; that reliable and safe development can be achieved with collaborative engagement between the public,
6See Kell (2011).
regulators, and operating companies; that reasonable, locally specific regulations need to be combined with a responsible operations philosophy and effective risk management practices; and that transparency and reasonable regulations can enable natural gas to be economically developed in an environmentally responsible manner.
Zoback, who serves as Benjamin M. Page professor in earth sciences and senior fellow at the Precourt Institute for Energy at Stanford University, presented slides addressing multiple topics raised by the Secretary of Energy Advisory Board on which he served. His brief comments, however, focused on the earthquake issue (his primary area of expertise). He said that there has been an unusual amount of seismicity in the United States in the past few years, including in areas where shale gas operations are ongoing. Triggered seismicity has long been known to be a potential consequence of underground injection of water. He explained that injection of fluids reduces the coefficient of friction that keeps plates from sliding across each other, so that the timing of eventual earthquakes is advanced. He said that this effect of water injection is indeed something worth worrying about and that impoundment in reservoirs can also induce seismicity. The issue is to decide when it is worth worrying about induced seismicity from particular activities. It is possible to map the locations where induced seismicity is most likely to occur; when injection into those areas stops, the induced seismicity stops. Thus, he continued, there is a need to monitor potential seismicity, manage the pressures, and act accordingly.
Zoback said we know how to do this: “it is not rocket science.” He summarized the NRC report on induced seismicity (National Research Council, 2013) as having three main messages: (1) the very small seismic events associated with the fracturing process itself pose essentially no risk to the public; (2) the risks associated with wastewater injection pose a larger, but still low risk, which can be reduced by site characterizations and proactive planning; and (3) the potential for induced seismicity is much larger with carbon capture and storage7 than with shale gas development.
7“Carbon capture and storage” refers to processes to remove carbon dioxide from the air, chemically combine it into a nongaseous form such as a carbonate, and then inject the carbonate underground or otherwise store it indefinitely.
Mauter, who serves on the faculty in Chemical Engineering and Engineering and Public Policy at Carnegie Mellon University and conducts research on resource efficiency in water and energy systems, cited Webler’s presentation to support her point that the risks discussed in Nygaard’s presentation are not the only ones that need to be characterized. She noted that the probability and consequences of a risk event occurring are not the only important dimensions of risk, citing the work of Slovic and colleagues (e.g., Fischhoff et al., 1978; Slovic, 1987) on perceptions of risk, which found that perceived magnitude of risk is strongly affected by the degree to which a risk is dreaded and unknown. She suggested that unplanned fluid migration is a risk that seems uncontrollable, in contrast to risks like truck accidents, which may have higher consequences but are not so dreaded or unknown, so may not be perceived as being as serious.
Mauter reported on some as-yet unpublished research on liquid waste transport in the Marcellus shale formation in Pennsylvania. This research found that the mean transport distance for flowback water was 113 miles, with transport of water for shale gas development accounting for about 0.1 percent of all truck traffic in Pennsylvania in 2011. This transport traffic has associated risks including diesel emissions, accidents, and others. The research found, she continued, that the percentage of waste a company reuses was the strongest predictor of length of waste transport and that most other company characteristics were not predictive. In this research, company attributes and experience did affect how waste was managed: Larger companies reused less waste, and companies that drilled more wells in 2011 or that had longer experience in the Marcellus shale region reused more of the waste. Companies that had more wells in a cluster reused more of their waste. Mauter added that company attributes do not do a good job of explaining frequency of violations of wastewater regulations.
Participants’ questions and comments to the presenters raised multiple issues, which are summarized by the rapporteur below under the headings of injection of wastewater, green completions, industry communication about risks, safety culture issues, air emissions, water retention pits at well sites, and long-term issues with flowback.
Injection of wastewater. In response to a question about whether injection of wastewater into deeper formations is a viable idea, Nygaard said that in a number of shale plays it is feasible to recycle the water, but in other areas, where local water is not of sufficient quality for fracturing, disposal
is needed. In those cases, operators need to find the right place for disposal. Zoback noted that there are far more injection wells in Texas than in Pennsylvania because there is a large saline formation under much of the shale area that allows for recycling the water. He emphasized, though, the need to match practices to what nature makes possible. Even where injection wells have operated safely for decades, it is possible to go beyond the capacity of the formations to accept additional wastewater.
Green completions. In response to a question, Nygaard said that flowback equipment is designed to allow for separation of liquids, solids, and gases, so that, depending on local conditions, it is sometimes possible to earn money from the separated components. He said that green completion increases cost, but not to the point of doubling it.
Industry communication about risks. Bernard Goldstein, a professor at the University of Pittsburgh Schools of the Health Sciences, proposed that it is the industry’s responsibility to make sure the public understands the risks. In his search of the records in Pennsylvania on water issues, he found that companies and the state government mentioned that there was no water shortage problem as sometimes arises in the West, but they never mentioned that underground injection of wastewater is not feasible in Pennsylvania, as it is in Texas. He also said that people in Pennsylvania were commonly told that fracking was a short-term proposition lasting about 2 weeks, but they were not told that it is normal to have eight consecutive frackings. Goldstein cited George King [the petroleum engineer whose study Nygaard had cited] as having said that industry may have gotten it wrong by letting people think that fracking was just the release of gas from underground and not considering that public concern has been about whether ground water will be contaminated.
Nygaard said that industry has found that the more information it provides in community engagement, the better—whether in Texas, in the Marcellus formation, or internationally. Eight years ago, when the Barnett shale play was opening up, industry had not yet learned much about water management or about communication. He said that this is a developing process, in which industry is learning about the need to provide the public with much more detailed information.
Another participant asked if industry could be more aggressive in providing background information about water and air issues in advance of development, noting that governments lack the ability to collect all this information. Nygaard replied that transparency across groups is important and that doing more to be transparent is better.
In response to a question about public availability of data about toxic chemical exposure, Nygaard said that in the past 2-3 years, industry has
made significant strides in public information. He cited FracFocus as making much detailed information available and said that in selecting fracturing chemicals, the company looks for those that do best at moderating environmental exposures while meeting the requirements for fracturing.
Safety culture issues. A participant asked about managing the risks of unexpected events, especially when many different people and companies are involved on a well pad and face economic pressures. He offered an example of a driller who unexpectedly came upon an abandoned coal mine shaft into which the cement for the well lining kept flowing freely. He asked how operators inculcate conservative behavior [with respect to safety] among employees making many moment-by-moment decisions. In his reply, Nygaard emphasized two main points: corporate culture and a regulatory environment that is reviewing and assessing situations. Key to corporate culture in his company, he said, is the expectation that we will operate safely and the understanding that any release must result in notifying management. He said that his company has a culture of safety and empowerment: any employee can shut down the operation if something may be unsafe.
Zoback added that safety culture is fundamentally a management problem. He noted some similarities with offshore oil drilling. In working on the National Academy of Engineering’s study of the Deepwater Horizon accident, he found that a strong safety culture is more effective than just having operators follow regulations. He cited the American Petroleum Institute’s recommended practices, described by King (2012), which call for multiple barriers, as illustrating the best way to achieve a goal such as protecting surface aquifers. He said that regulations need to focus on the right issues: setting and achieving an environmental objective, rather than specifying a set rule to be followed.
In response to a question about how safe practices and the American Petroleum Institute recommendations achieve accountability for service contractors to well owners, Nygaard said that safety culture has to come from the chief executive officer on down and has to empower all the employees. He also said that the owner of the well is responsible for the service providers; his company evaluates contractors’ safety performance and environmental compliance, trains and orients them, asks whether their staff is capable of delivering his company’s needs, and lets crews go when not comfortable with their performance.
Air emissions. In response to a question, Nygaard said that although his presentation did not address them, air emissions are also of concern in operations. He noted, though, that all emissions across the shale gas life cycle need to be compared with those from other energy sources. Zoback
added that proper well design and construction is the first line of defense for air emissons as well as for water contamination.
Water retention pits at well sites. In response to a request for more information about how these pits are used, Nygaard distinguished pits used to store fresh water for fracturing from those for storing flowback water. He said that in designing such pits, his company considers berms and embankments, linings, sizing pits to accommodate expected rainfall, and avoiding floodplains. It then uses engineering design.
Long-term issues with flowback. An Internet participant asked what the industry is doing to address long-term risks from flowback after the lifespan of the well, especially with corrodible metals and concrete, and also about the liability situation after wells have exceeded their productive life. Nygaard said he was unable to comment on liability, but that in well construction, his company selects materials that will resist corrosion from the fracturing fluids and the flowback and designs wells for their intended lifetimes. Once the productive life is ended, he said the company returns to apply plug and abandonment procedures as established by state regulations to avoid long-term exposure.
Vengosh is professor of geochemistry and water quality and chair of the Water and Air Resources Program at the Nicholas School of Environment at Duke University; his research focuses on connections between energy and water quality. He presented work on which he collaborated with Robert Jackson, Nathaniel Warner, and Thomas H. Darrah, all of Duke University. He began by emphasizing that there are many gaps between what is known and what stakeholders want to know about the effects of shale gas development on water resources. His presentation, he said, would discuss only a few of the issues.
Stray gas contamination, such as the appearance of methane in drinking water wells near shale gas development, can cause fire and explosion hazards. Although direct health effects have not been found from drinking high-methane water, Vengosh, said, the presence of methane in water can cause water wells to be shut down, thus depriving users of their water supplies. He added that one study (not cited) also indicated that methane in water can decrease property values. Vengosh said that although some
studies suggest that methane in shale gas areas is a naturally occurring phenomenon (e.g., Molofsky et al., 2013), other studies that look closely at the chemical composition of the methane (e.g., Osborn et al., 2011) indicate that it is possible to distinguish different sources of methane in well water. The study by Osborn and colleagues found that water wells less than 1 km from active shale gas wells had a higher probability of having methane levels above the action levels defined by the U.S. Department of the Interior, implying that there is potential hazard to households using those wells. Newer data from Vengosh’s Duke University group reinforces these findings, suggesting that stray gas from shale gas wells was the cause of observed high methane levels in nearby drinking water wells. Several mechanisms could cause this contamination, but the most likely ones, according to Vengosh, include inadequate well integrity, improper cementing of wells, and improper well design that allows gas to escape along the well annulus. He emphasized the limitations of available knowledge. Although the existing studies are detailed, they cover only limited areas. A new study by his group and other coauthors (Warner et al., 2013) did not find a correlation between distance from gas wells and methane concentrations in areas of extensive shale gas exploration in Arkansas. What was found in the Marcellus shale may not apply to the Fayetteville shale, Vengosh concluded, so there is a need to look at every basin.
Surface water contamination can arise from spills of fluids, from normal well operations, and from disposal of wastewater, Vengosh said, adding that the risks likely vary across shale plays. In the Marcellus shale, the consensus view is that 10-20 million liters of water are needed for an average well and that wastewater volume averages 5.2 million liters, for a total of 3.1 billion liters across the Marcellus operations in 2011 (Lutz et al., 2013)—about four times the volume from conventional oil and gas production. Wastewater in the Marcellus formation has high salinity and bromide levels that could contaminate downstream water, toxic elements such as barium and arsenic, naturally occurring radioactive materials, and various organic compounds. The salinity in the flowback water sometimes poses a greater hazard than the toxic contents.
Vengosh explained that the wastewater has been treated at municipal treatment facilities or private industrial brine treatment facilities, disposed via underground injection, recycled for additional use in fracking (the use of about 70% of Marcellus wastewater in 2011), or spread on roads for ice control (a use that is still allowed in West Virginia but not in Pennsylvania). He noted that each treatment approach has risks: Treatment in municipal plants decreases the effectiveness of those plants for treating domestic wastewater; treatment in brine facilities is inadequate for handling halogens or radioactive materials. Deep injection has seis-
micity risks and is not possible in all locations. Recycling is the desired solution, but high levels of barium and sulfates could shut down well operations. Wastewater treatment does not remove all the contaminants; it leaves high salinity and bromide levels, so that treatment effluent may have several times the upstream levels of bromide for as far as several kilometers downstream. Such effluent could increase the formation of carcinogenic disinfection by-products when the water is chlorinated as a source of drinking water by communities downstream. For these reasons, Vengosh said, zero discharge of any effluent should be a goal for shale gas operations.
Long-term effects are of four main types. Long-term effects on water availability can be an issue in some places, Vengosh said. In the Marcellus shale, water use is 40-60 million cubic meters per year; in Oklahoma, 16 million cubic meters, which amounts to about 1 percent of statewide fresh water use. In the Barnett shale, 30 million cubic meters of water are used annually, which is equivalent to 7 percent of all water use in Dallas (Nicot and Scanlon, 2012). A prediction that Vengosh used for water use in shale gas extraction nationally is 150 million cubic meters per year, which he said is far less than is used for hydropower. However, in water-scarce areas in the West, such as Texas, where some counties have only ground water as a source for fresh water, he thought that competition for that water could become an issue.
A second long-term issue concerns connectivity between deep and shallow aquifers. The industry emphasizes a vertical distance of 3-4 km between ground water and the water used in drilling, Vengosh said, but even though the geological systems have low permeability, this may not be sufficient. More research is needed to quantify the connectivity between deep and shallow aquifers. Myers (2012) offered an initial estimate that contaminants will arrive in deep aquifers in about 10 years, though Vengosh said that there is a lot of debate about this estimate. Osborn and colleagues (2011) found that water wells in Pennsylvania valleys have a distinctive geochemical and isotopic fingerprint, with very high salinity similar to that of the Marcellus brine, but distinguishable with sophisticated isotopic fingerprinting of strontium concentrations. The study found that similarly saline water was found in the 1980s in the system, so the presence of saline water need not imply current contamination. However, Vengosh suggested, considering that this is an area with a high rate of water recharge, saline water would have been flushed away by now, indicating that there must be some continuous connection, possibly from shale gas development.
A third long-term issue, suggested by Harrison (1985), is possible ground water contamination from improper seals in gas wells. Gas and
brine might enter the annulus of improperly designed wells and be released into the surrounding formation, particularly in places where water and gas could flow from abandoned and improperly sealed gas wells into conventional water wells. There are areas of very high well density in southwestern Pennsylvania, which should be considered high-risk areas. Well density is much lower in northeastern Pennsylvania.
A fourth issue discussed by Vengosh concerns certain contaminants, such as radium, that remain in the environment for a very long time after they appear. After wastewater is disposed of, the amount of radium in the downstream sediment may increase over time because of the long half-life of radium. Tools are available to identify the sources of radium in water, and locations with high radium concentrations have been mapped (Lutz et al., 2013).
Vengosh concluded by reemphasizing that the scientific understanding of the risks to water sources from shale gas extraction is still in a very early phase and that decisions about risk are currently based on very limited data.
Nicot is a civil engineer and a research scientist at the Bureau of Economic Geology at the Jackson School of Geosciences at the University of Texas; his research interests include modeling of contaminant transport. He spoke first on the issue of methane in water wells, emphasizing the point made by Molofsky and colleagues (2013) that gas in shallow aquifers does not necessarily indicate contamination. He said that it is important to have baseline data on wells and that more drilling companies are now collecting such data, which will give a better idea of the sources of methane in well water. He also noted that there is extreme variation in methane concentrations in water at a single point and reiterated that it is important not to generalize across shale plays about either methane or the appearance of radioactive materials.
Regarding water use, Nicot said that in Texas, the industry uses about 100,000 acre-feet/year of the 15 million acre-feet available overall (0.67%) and that in Colorado, industry use is also a very small fraction of all available water. Although water levels are dropping, this appears to be mainly a result of drought and increased water use for other purposes. He referred to a new report from the U.S. Geological Survey, not further identified, which shows that water levels are dropping at the same rate across a geologic region, regardless of whether shale gas is being developed in parts of that region. Nicot said that when a single water well goes dry, this does not necessarily indicate depletion of the whole aquifer; the
key question is whether there is enough water at, for instance, the county level to support the multiple uses, including fracking.
On the issue of fluids migrating along faults, Nicot said that well operators normally don’t drill at geologic faults, which lowers the risk of fluid migration. He agreed with Vengosh that brine does move across geological levels, but said it is an open question how fast it happens.
Several issues were discussed in response to questions and comments from the participants. The rapporteur has summarized these under headings of regional planning, monitoring, and other issues.
Regional planning. Warner North, principal of Northworks, Inc., said that in the Marcellus shale, the release of bromides, radon, and other contaminants into rivers is a major problem. He suggested that there ought to be a carefully planned regional system for disposing of the wastewater, rather than doing this well by well. Furthermore, regional cooperation might be appropriate in the arid West, where the issue is moving large amounts of water. Nicot responded that in West Texas, where there is not much fresh water, the state has been getting good information on aquifers and the industry is increasingly using brackish water and asking the state to study brackish aquifers as possible water supplies. Vengosh added that he was unaware of any such cooperative efforts in the Marcellus basin.
Monitoring. Workshop chair Mitchell Small said that the U.S. Environmental Protection Agency (EPA) has an information collection rule for water treatment plants and wondered if a similar program of industry monitoring with regulatory oversight existed for shale gas, or if not, whether it would be helpful. Vengosh replied that he was unaware of any systematic monitoring program to produce public domain information before and during shale gas extraction operations. He said the industry has a lot of information, but it is not all publicly available; despite research efforts to monitor in particular places, there is no overall effort, which is why the “big picture” is lacking. Vengosh advocated a general public monitoring system using the best available analytical tools to delineate sources of contaminants in the subsurface. He said it may be too late to do this where drilling is already occurring, but it could certainly be done in new shale plays, such as the Monterey formation in California. Zoback commented that the Secretary of Energy Advisory Board on which he served called for a process of continuous improvement in the science, but the government has failed to follow through on that recommendation. He agreed about the need to collect more data.
Other issues. In response to Zoback’s suggestion that fracking will reduce the core pressure in gas wells so that fluids should be attracted downward into the wells rather than upward toward aquifers, Vengosh said that he had not seen a study that has tried to simulate that situation in real formations. He said that to answer this sort of question would require drilling research boreholes and monitoring pressures and flows.
There was further discussion of the possibility that thermogenic gas observed in water wells could have come from shallower shales than those being drilled for gas. Vengosh said that Molofsky’s research was based on very few observations. He noted that contamination from an intermediate level could come from a well casing failure rather than a natural process. Vengosh said that the studies by Osborn and colleagues (2011) and by Molofsky used different underlying assumptions. Robert Jackson added that there are many ways to tell different sources of gas apart and said that in the research by the Duke University team (to which he and Vengosh belong), there was evidence of both Marcellus gas and gas migration from shallower shales in the water wells investigated.
An Internet questioner asked about losses of water that is injected deep underground and does not return to the system (that is, the Earth’s hydrosphere). Nicot responded that he did not see this as a water loss issue because the oceans provide a huge reserve of water and because burning the extracted gas produces more water vapor than is lost by subsurface injection. He also said that evaporation of water used for irrigation and power plants takes much more fresh water than the shale gas industry does. Vengosh concurred, saying that even though shale gas extraction needs a lot of water, it is a small demand on freshwater resources relative to other energy-related uses.
Moore is an assistant research professor in the Division of Atmospheric Sciences at the Desert Research Institute, where his research focuses on the cycling of atmospheric pollutants. His presentation examined air emissions that can occur at each stage of the shale gas life cycle, summarized available data, and addressed areas where data may be lacking. He emphasized that hydraulic fracturing is only one part of the life cycle of shale gas development, which includes the phases of well development, production, distribution and storage, use, and end-of-life stage (e.g., well closure) (Branosky et al., 2012). The data on air emissions are almost entirely from the first three stages of the life cycle, he said.
Emissions of concern, Moore continued, include methane and ethane; benzene, toluene, ethylbenzene, and xylenes; hydrogen sulfide; ozone precursors; particulate matter; and silica. All these can have respiratory effects; benzene is a carcinogen. Modeling studies suggest that shale gas will have greenhouse gas implications similar to those of coal and conventional gas over 20- and 100-year timelines, but there has been a lack of direct measurements to support such expectations, Moore said, and estimates of emissions such as EPA’s (e.g., U.S. Environmental Protection Agency, 2013) have been volatile. As an example, he said EPA’s estimate of 2010 methane emissions from all gas operations dropped by 33 percent between 2012 and 2013. Getting accurate estimates is important because, according to Moore, natural gas extraction is the largest U.S. source of methane. Moore added that methane emissions from gas extraction processes appear to come mostly in the well development phase.
In the well development phase, he continued, the major air emissions are from vehicle traffic, which produces emissions from diesel fuel, and coarse particulate matter from road development. In the drilling and fracturing phase, 1 to 5 million gallons of water are hauled per well, bringing diesel emissions from trucks, and there are also emissions of fracking fluid and sand, along with releases of volatile organic compounds and silica. Particulate matter, hydrogen sulfide, and methane can also be released during drilling, Moore said. During well completion, flowback water is removed from the well bore and there is often venting and flaring, which he noted can release methane, hydrogen sulfide, and volatile organic hydrocarbons. Moore also noted that the well completion process has begun to be regulated more frequently.
Moore briefly discussed a few case studies that illustrate the state of empirical knowledge. In one, McKenzie and colleagues (2012) sampled around well completion activities at four well pads in Garfield County, Colorado. They evaluated risks to residents within and beyond half a mile from the wells and found increased health risks for residents living nearer to well sites. Moore considered this a good study, but one that only scratches the surface because it needs validation in other areas. An air quality study at eight sites in Fort Worth, Texas, did not find any air impacts related to gas drilling, concluding that a 600-foot setback adequately protected the public.8
In the production stage, the main emissions are of methane and volatile organic hydrocarbons from leaking valves and from diesel-powered compressor stations, Moore said. A study in Wise County, Texas (Zielinska et al., 2011) characterized emissions from gas facilities based on monitoring downwind of production areas and concluded that there is an expo-
8See City of Fort Worth (2011).
nential decrease with distance of concentrations downwind of condensate tanks. The study could not determine whether these emissions were significantly higher than normal emissions sources in the area, except that a significant level of benzene was generated. Moore briefly mentioned several other studies, one of which, in Utah, found levels of ozone above EPA standards.
Gas transmission and storage has been claimed to be a major source of leaks of methane and ozone precursors near pipelines. Moore cited as relevant in this regard one study that used mobile mapping along roads in Boston and found hundreds of methane leaks, some producing concentrations more than 15 times global background levels and occasionally at levels that pose an explosion risk (Phillips et al., 2013).
In conclusion, Moore emphasized the critical lack of studies offering actual measurements. What is especially needed, he said, are targeted studies with measurements before, during, and after drilling; studies defining the emissions signatures from all shale gas formations so that the sources of emissions can be traced; more data on surface atmospheric fluxes of methane, especially in urban areas; and characterization of silica emissions. These steps, he said, must be taken to ensure public safety in the near and distant future.
Petron, an air quality researcher and associate scientist at the Earth Systems Research Laboratory of the National Oceanic and Atmospheric Administration and researcher at the Cooperative Institute for Research in Environmental Sciences at the University of Colorado-Boulder, reported that in the Rocky Mountain region, ozone formation has been a major issue in Wyoming and in Utah, which set a national record for ozone concentrations. This phenomenon appears to be associated with temperature inversions, especially in winter. The states have begun to reduce these exposures by taking emissions inventories and regulating the precursors. She noted that reducing volatile organic compound emissions has a cobenefit by reducing emissions of greenhouse gases. She said that state ozone monitoring in Colorado does not yet occur in regions near gas fields. There is an effort ongoing to redeploy the monitors toward urban areas and areas of oil and gas activity.
Petron said that her group monitors emissions from towers, aircraft, and in situ, and has been monitoring from vans since 2007. It sees impacts of oil and gas activities daily, tries to pinpoint leaks, and has found measurement from airplanes very useful for identifying which parts of a region have higher methane levels. The measurements indicate that
benzene and methane correlate very highly in this region. However, there is a need to monitor other types of emissions as well, including emissions of additives such as acids, biocides, and solvents.
Petron discussed the importance of getting good measurements of methane emissions. She presented data indicating that the variation in EPA’s estimates has been even greater than the 33 percent change that Moore mentioned. She then reported on an as-yet unpublished study by her research group, based on measurements of methane losses at three locations. It estimated a loss of 4 percent of extracted methane in the Denver basin and a 9 percent loss in the Uintah Basin. This estimate is much larger than EPA’s estimates of a 1 percent loss from production and processing activities and a 1.9 percent loss from the full gas cycle. It is also higher than the 3.2 percent estimate given for gas in a study claiming a net climate benefit of gas compared with coal as the fuel for new power plants (Alvarez et al., 2012).
She concluded by identifying three priorities: (1) quantify actual emissions, especially fugitive emissions, with more field measurements: Petron said scientists have been using emissions factors from the early 1990s; (2) estimate emissions reductions from best management practices, which Petron said is being done in an ongoing University of Texas study; and (3) develop an effective and scalable leak detection program, with instruments that are usable by the industry for self-checking. Because there have been some strong, quantifiable impacts in some regions, new monitoring and detection techniques need to be deployed to assess and address the problem. The knowledge exists for capturing emissions for green completions, as has been done in Colorado, but that knowledge is not always used, Moore said, as indicated by high observed emissions at some sites in Utah and in Dish, Texas, where observed concentrations varied from less than 5 ppm to 30 ppm.
Participant questions and comments opened several topics of discussion, which the rapporteur has summarized below under headings of Petron’s measurements, relative emissions, monitoring, screening approaches, and variations in regulations.
Petron’s measurements. In response to a question about the durations of emissions measured by Petron’s equipment, she said that the measurements are over very short times—usually less than an hour—for the purpose of detecting problematic sites; they do not provide quantitative estimates over time. She said that to identify a chronic source of emissions would require measuring over several days. In response to another ques-
tion, Petron said that her measurements of methane releases in Colorado and Utah are for the whole of the natural gas system at the measured sites and acknowledged that not all the emissions are due to natural gas production (some may be from oil). She said that further measurements will help distinguish emissions from old and new wells. In response to a question about measuring aerosols, Petron said that this is being done by the National Oceanic and Atmospheric Administration, but her presentation focused on the emissions that are now of greatest concern to state governments.
Relative emissions. A participant asked how air emissions from shale gas development compare to other emissions; for example, how emissions from a gas well compare with emissions from a gas station on a freeway. Petron replied that in oil and gas development areas, emitted benzene is coming mostly from these activities and not from transportation, although this varies case by case. In the Barnett shale, there is very little benzene, but it may be much greater in other shale plays. Moore suggested that as more is learned about point sources of air emissions from gas wells, stakeholders may want to require controls as is done now in California for gas stations.
Monitoring. A participant asked about the status of methane monitoring technologies. Petron said the technologies are ready to use and their adopters are starting to produce publications. She said that measurements from airplanes can provide isotopic signatures that allow emissions to be attributed to particular sites. Monitoring in Colorado costs about $200,000 per month for a gas field, Petron added, while in Texas it costs about $350,000 per month because of the need to send a crew to the site.
Robert Jackson, of the Duke University research team, supported the need for monitoring in many places, not to document high concentrations or get averages but rather to examine the relationships among many gases from above (from a plane or a skyscraper) to identify the sources of the emissions. He also suggested that a lot of progress can be made by collaboration with industry. Moore agreed that long-term monitoring projects are desirable if there is agreement on what needs to be monitored. Petron agreed about the need to be able to differentiate between gas emissions from feedlots and natural gas production but said that current practical capability is far from that goal. She also noted that monitoring by the National Oceanic and Atmospheric Administration at eight towers and from aircraft has been cut back because of tight funding. Jackson said that air emissions represent the area where the greatest progress can be made most quickly and that it presents a multiwin situation because of safety, economic benefits, and air quality. He expected that in a year
much more will be known about air emissions than is known now. David McCabe of the Clean Air Task Force expressed the opinion that although measurement in collaboration with industry is very important, it is hard to assess the gross emitter problem in close collaboration with industry, so independent government and university measurements are also valuable.
Screening approaches. McCabe suggested that as with vehicle emissions where it is often one vehicle out of ten in a fleet that produces most of the emissions, a key approach for gas well emissions is to find a way to screen to find the few major sources. He suggested that one does not need isotopic signatures to “prove” that the source has been found. He mentioned an EPA study of gas processing plants a decade ago that found that leak rates varied by a factor of 50 across plants; in each plant, emissions were dominated by the top 10 leaks out of thousands, and the top 10 leaks from the single leakiest plant accounted for 35 percent of all the leakage in the entire study. Petron agreed about the value of easy-to-use screening devices, especially because the industry could use them to find and close their leaks and because there are only a few government regulators available—only 17 inspectors in all of Colorado, for example. She believes that within the next few years, it will become fairly easy to find the leaks.
McCabe said that the regulators need something specific to localize the source of a leak, and Petron replied that available one-second measurements make it easy to see where emissions are coming from. She disagreed with some of McCabe’s comments, however, noting that methane is not the only emission needing measurement and specifically mentioning volatile organic hydrocarbons. She also said that isotopic measurements are needed for observations from planes because that is the only way to trace the sources of the emissions.
Variations in regulation. There was a brief discussion about variations in regulations across jurisdictions, and even within a single state. Petron said that within Utah, regulatory authority differs between Indian land and other locations.
Adgate, who chairs the Department of Environmental and Occupational Health at the Colorado School of Public Health and whose research focuses on improving exposure assessment in epidemiological studies, organized many of his comments around a health impact assessment his group conducted in Battlement Mesa, Garfield County, Colorado. He
noted that public health risks result from contact with stressors, including both direct and indirect effects through environmental and social processes. Stresses may arise during the short-term well development process, the production phase, and afterward. Different kinds of stressors can arise at different phases, leading to public health risks.
Air exposures. The Adgate group’s health assessment focused a lot of attention on air quality issues. It gathered existing information from county officials and state records, such as local air monitoring data, traffic, and noise estimates; anecdotal reports of exposures and health symptoms; demographic and vital statistics data; data on cancer and other diseases; and school and crime data. It also examined the scientific literature to help think about possible exposures. It is important to note that complete exposure information and health outcomes data were not available. The group looked for potential impacts on acute diseases and cancer; accidents, fires, and explosions; and community changes that might affect activity levels, social engagement, and psychosocial stress among individuals.
Adgate said that the assessment led to some straightforward recommendations, which nevertheless proved controversial—for example, to reduce exposures, promote safe operations in residential areas and foster constructive interaction among stakeholders. Adgate noted that there is increasing concern about shale gas development in Colorado because it is now moving into populated areas.
The health risk assessment (McKenzie et al., 2012) emphasized the need to manage flowback, which seemed to be the most significant source of emissions. A 2011 EPA study found that methane, hazardous air pollutants, and volatile organic compounds were emitted at about 20 times the level in fracked gas wells as in unfracked wells. Using a limited number of “flowback” and nonflowback water samples, the group applied a standard screening risk assessment method that produced a hazard index for noncancer health risks and estimates of lifetime excess cancer risk. The hazard index fell below the line indicating that health effects might occur for most populations, but not for subchronic risks for people living near the wells. Using a 20-month exposure scenario, the hazard indexes were above the cut-off line of 1.0 for neurological, respiratory, and hematological effects, but not for developmental effects. The most important noncancer risk driver was trimethyl benzenes.
Lifetime excess cancer risks were assessed to be above a 1-in-a million target level but well below the 1-in-10,000 level that calls for EPA remediation. Benzene was the primary driver of lifetime excess cancer risk. Adgate emphasized the preliminary nature of these results, saying there are lots of limitations to the data.
Water exposures. Adgate said that the same risk assessment approach can be used with water exposures, but assessments using this approach have not yet appeared in the literature, despite the high level of public concern about water. A large number of chemicals go into fracking fluids, but much less is known about what comes out in the form of flowback from the high-temperature, high-pressure environment below the surface. Disposal practices are likely the most important source of risk because of the contents of the flowback fluids.
Industrial activities. Industrial activities and sand mining expose workers to silica and bring the risk of silicosis, Adgate said. In his group’s health impact assessment, truck traffic was a big complaint. Truck trips have been estimated at 1,000 per well in New York, with multiple wells per well pad. Thus, Adgate concluded, living near that level of traffic poses hazards such as exposure to diesel fumes and dust and risks to the safety of school children. Industrial safety culture can also be an issue. A report in Wyoming found worker fatalities occurring at two to three times the national rate, mostly on drill rigs and in transportation.
Other stressors. Adgate’s group found noise levels within 1,000 feet of wells in Colorado to be high enough to be a stressor. He noted an emerging literature on relationships between noise and cardiovascular disease. The health impact assessment also noted that increases in arrests and sexually transmitted diseases were positively correlated with the large increase in well starts from 2005 to 2009. Local residents also reported increased stress, insomnia, and other reactions. Adgate mentioned a recent report by Kyle Ferrar at the University of Pittsburgh that found that among 33 people who believed their health had been adversely affected by shale development in the Marcellus basin, the top stressors mentioned were all social (Ferrar et al., 2013). They included being denied information or being misinformed, corruption, and complaints or concerns being ignored.
Adgate emphasized the almost complete lack of information on exposures to health stressors and a lack of health tracking information, including information on occupational health. He argued in favor of greater transparency and better information aimed at the following objectives: (1) to characterize the range of activities and environmental factors relevant to smart setback policies; (2) to describe the variability in emissions, air levels, and human exposures; (3) to estimate toxicity factors; (4) to understand the effects of chemical mixtures and of noise, traffic, and accidents as stressors affecting health and quality of life; and (5) to incorporate better measures of stress into individual and community-level health assessments. He said that systematic data collection before, dur-
ing, and after shale gas development continues to be needed for data on exposure and health and that more study is needed on chemical mixtures as stressors and on nonchemical stressors, as these stressors affect both workers and communities. In addition, public health exposure-prevention strategies should also be directed at minimizing exposures during completion activities.
Brown is a public health toxicologist who has served as chief of environmental epidemiology and occupational health in Connecticut. He currently serves as an environmental health consultant to the Southwest Pennsylvania Environmental Health Project, a nonprofit group organized to assist residents of Washington County, Pennsylvania, who believe their health has been or could be affected by natural gas drilling activities. He reported on work done by that project to identify patterns of health effects in exposed residents in southwestern Pennsylvania, to track the exposures, and to advise residents on ways to protect their health. The project is supported by three foundations and has the goal of providing “accurate, timely, and trusted public health information and health services associated with natural gas extraction.” He emphasized the issue of trust because mistrust of health information in the region is so great that people are unwilling to even talk with strangers about these issues. In his project, an experienced nurse practitioner visits people who contact the project and provides examinations to arrive at health evaluations. The project offers the only physician education program available in the Marcellus shale region and provides clinical toxicology profiles to the nurse practitioner so that she will know the potential effects of exposure to relevant chemical agents when she talks to people who may have been exposed. The team also includes public health and occupational health professionals, a toxicologist, and a community outreach specialist.
The primary goal of the project is to identify and address health impacts by identifying the probable or possible causes of these impacts and determining actions that can be taken to reduce the stress level in this population. The nurse practitioner uses a structured interview schedule that asks about symptoms, when they appeared, the person’s proximity to environmental sources, and social or emotional factors that may be associated with the symptoms. She usually spends 45 minutes talking with people before asking any health questions.
The client population has reported skin rashes or irritation (48% of individuals), nausea or vomiting (45%), abdominal pain (38%), breathing difficulties or coughing (41%), and nosebleeds (21%). Other common
complaints have included anxiety and stress, nervous system problems including headaches and dizziness, and eye and throat irritation. The nurse practitioner’s reports include her assessment, based on the interview and her experience, of what she thinks is the client’s situation with respect to exposure-relevant symptoms.
The project creates a case file on each person. It considers symptoms to be attributable to shale gas drilling based on three criteria: the temporal relationship between gas extraction activities and the onset of symptoms, the presence of an identifiable exposure source proximate to the individual experiencing symptoms, and the absence of an underlying medical condition that was at least as likely to have caused the symptom. The case records are reviewed by a team including a toxicologist, an occupational physician, a nurse practitioner, a psychiatric nurse practitioner, and two public health researchers, with the objective of determining for each case whether shale gas exposure can be ruled out. The reviews consider only the individual interviewed in each household. The population is self-selected—it consists only of people who approached the project, and all clients are in Washington County.
The analysis of the cases found dermal effects in seven people, all attributed to water exposure, and respiratory (13), neurological (3), and eye irritation (4) effects, all associated with air exposures. The group initially assumed that exposures would be through water, so was surprised to identify people with no water exposure who had symptoms attributable to shale gas operations. The results were consistent with some other research studies (e.g., Ferrar et al., 2013; Steinzor et al., 2013). The group noted that clients also reported health problems in pets, which created additional stresses for the clients.
The project’s nurse practitioner also surveyed 279 people who presented with complaints to a clinic in Burgettstown, Pennsylvania, during 2 months in 2012-2013. She found that this group scored below norms on each subscale of a standard psychiatric test and that at least 30 percent of respondents were at risk of depression, compared with a national average of 19 percent. These findings go beyond anecdotal data, but are still clinical. Brown sees the findings as indicating a significant public health problem.
The project tried to reconcile its clinical data with literature indicating that there have been no exposures in the study area. To do this, it undertook some environmental measurements, such as monitoring airborne fine particulates for 4-5 days in homes located about 1,000 feet from compressor stations. It found long periods with fairly stable background particulate counts of 1,000-2,000 per cubic foot of air, but there were shorter periods with peak counts of 7,000-8,000 per cubic foot. Measurements of volatile organic compound concentrations over weeks, months, and
the entire year showed similar wide fluctuations in readings and some periods of very high exposure. The data indicate that cloud cover, wind speeds, and other environmental conditions strongly affect air mixing and observed concentration levels. Brown concluded that serious modeling is needed to better understand the exposures.
The project’s primary objective is to tell people how to reduce their exposures. Measuring fine particles can be a surrogate for all air exposures and can help people know when it is all right for their children to go out and play. The project hopes to be able to offer a simple screening test to allow people to connect measurements to action recommendations. It advises people to reduce outdoor activity and to remove children from polluted sources. It asks them to use filtration systems to reduce exposure to particles and gases, and it is evaluating different filtration systems in collaboration with a group from the University of Pittsburgh. Inventorying emissions near clients’ homes can reassure people who are not exposed: some people are panicked even 7 miles away from sites. Finally, the project asks clients to maintain environmental and health diaries.
The project offers clients “three good things to do”: clear the air by managing home ventilation, cleaning the house often, and avoiding tracking in dust; use clean water for cooking, showering, and drinking, and see a doctor if water use appears to burn skin or cause rashes; and look for health changes by keeping a health diary, checking water, monitoring air, and paying special attention to children, the elderly, and the chronically ill. The project also suggests reducing noise and light pollution in homes. If clients cannot follow these guidelines, the project advises them to consider relocating temporarily or permanently.
Brown concluded by saying that to understand the health effects, it is necessary to do basic public health work: conduct needs assessments, get health information and information about the chemicals, identify the plausible routes of exposure, and make recommendations for reducing it.
Bredfeldt is a senior toxicologist at the Texas Commission on Environmental Quality (TCEQ), where she focuses on human health risk assessment in relation to ambient air quality. She began by noting the concerns raised by several workshop participants about lack of data and suggesting that because of differences in geological and meteorological conditions, the public health issues are likely to be region-specific. She also noted that regulations need to be data-driven and tailored to community needs.
She described the Barnett shale region, where she works, as highly urbanized, with many air monitoring stations. For evaluating the impacts
of shale development, Bredfeldt said that Texas is adding mobile monitoring stations at a cost of about $250,000 each, as well as canister samplers that cost $75,000 to $125,000. She noted the rapid development in the Barnett shale: when the TCEQ first started thinking about impacts on air quality, there were fewer than 1,000 Barnett gas wells; there are now over 15,000. Bredfeldt showed a TCEQ graph showing that although ozone levels are above health guidelines, they have not increased in the region over the past 20 years despite the large increase in the number of gas wells. Benzene concentrations are well below levels of concern and also have not increased with the increase in shale gas wells.
She did identify some site-specific issues. For example, benzene concentrations in Longview increased over the decade before 2007 until in that year they exceeded the 1.4 ppb level that the TCEQ had set as the level of concern. Mobile monitoring identified the source of the emissions as a single well. By attending to this well, benzene concentrations in 2008 were brought back below the level of concern. The 7,000 TCEQ helicopter flyovers, which can identify volatile organic compound plumes by infrared photography, have found 88 cases of emissions causing concern.
Short-term monitoring efforts indicated that carbonyls, nitrogen oxides, and sulfur compounds did not exceed levels of concern for short-term exposure and that less than 5 percent of volatile-organic-compound canister samples exceeded levels of short-term concern in terms of health or odors. Bredfeldt said that the TCEQ responds to complaints, most frequently about odors but sometimes about runny noses or scratchy throats, and that a complaint from an area near facilities with a history of noncompliance brings TCEQ staff to the site within 12 hours to take measurements and get the specifics of the complaints. This approach has been very helpful in finding the sources of problems.
Bredfeldt said that the TCEQ collaborated with the U.S. Department of Health and Human Services on a study of the town of Dish in which blood and urine samples were collected from 28 individuals to look for biomarkers of exposure to volatile organic compounds. The TCEQ did not find concentrations high enough to conclude that there were problems specifically from volatile organic compounds.9
She said that the TCEQ experience indicates that nearly all the documented issues arose from human or mechanical failures that were quickly remedied and could have been avoided through increased diligence on the part of the operator. The needed corrective actions normally amounted to little more than replacing worn gaskets, closing open hatches, and repair-
ing stuck valves. The commission has engaged in outreach and education for operators and developed new rules in the form of best management practices for well sites to avoid exposures. This information, along with much else, is available on the TCEQ Website.10 TCEQ public education efforts have also included open houses. Bredfeldt invited workshop participants to explore the TCEQ Website for further information.
Participant questions and comments covered several topics of discussion, which are summarized below under headings of measurement and methodology issues; quality of epidemiologic knowledge; the case of Dish, Texas; and behavioral changes.
Measurement and methodology issues. In response to a question about whether variations in chemical concentrations in the homes studied by Brown’s project could have been due to cooking and other household activities, Brown replied that measurements are made every hour for 24 hours and daily for multiple days, the monitoring takes account of in-home activities, and examining variability over time both within and outside the home can determine whether the source is inside or outside.
A participant asked if the high risk of depression reported in Washington County could be attributed to high rates of unemployment, poverty, or other causes. Brown replied that Washington County is not a low-income area and added that his project has extensive information on social and economic stressors for each individual and will be teasing these issues out. Another participant noted that the Washington County data were from people attending a clinic, who may be more prone to depression than average residents. Brown indicated that the sample also included people who drove patients to the clinic and added that his project intends to consider this comparison and other possible sources of insight.
At another point in the discussion, Brown commented on the dangers of using national reference values for populations that are under stress and are experiencing a combination of chemicals and fine particles. He said that it is known that fine particles act synergistically with chemicals and suggested that therefore, when fine particles are present, general reference values are not very useful. He was concerned that by relying on reference values, experts might be in the position of telling people who present with valid symptoms that they are not sick.
Quality of epidemiologic knowledge. In response to questioning about whether there are any strong epidemiologic studies being conducted about the public health effects of shale gas development, either generally or on children in particular, the panelists said that they did not know of any. Alan Krupnick of RFF indicated that the Geisinger Health System, which operates in Pennsylvania, is working with RFF on a screening study examining their clients’ health data. He believes the Geisinger Health System is doing a major epidemiologic study in collaboration with a researcher at Johns Hopkins University. Another participant emphasized the need to start any public health study before the population is exposed, follow the population over time, and also follow workers’ health.
Bernard Goldstein raised questions about the willingness of responsible authorities to support good epidemiological studies. He said that of the 52 members of the shale gas commissions set up by President Obama and the states of Maryland and Pennsylvania, none had any health background; although there were good environmental organizations involved, no health organizations were represented. There were many state and federal government departments, but no health departments. He said that scientists are willing to do the studies, such studies are feasible (e.g., the Geisinger System and other groups could compare health conditions in areas with and without shale gas development), and Colorado had made a good start on a health impact assessment but was not able to complete it. Goldstein expressed doubt that the potential funders of such studies are willing to have them done.
The case of Dish, Texas. Some participants asked about the TCEQ’s failure to find evidence of serious exposures in Dish, Texas, where there have been multiple complaints. Bredfeldt replied that the TCEQ does not have monitors everywhere in Dish and speculated that much of the concern there was attributable to nuisance factors such as noises and odors. Gabrielle Petron responded that her research group has data on Dish and commented that some of the well pads there are very close to the town hall and to playgrounds. Dish has the dirtiest well pads she has seen in 5 years in the field, she said, and she offered to work with the TCEQ and share measurements. She said her group’s night measurements showed high levels of the subset of toxic chemicals they monitored. She expressed concern about the need for more exposure assessment. Bredfeldt expressed interest in sharing and comparing the two groups’ data and reiterated her point that the TCEQ data are incomplete.
An Internet participant asked if there is an explanation for the lack of any increase in air pollutants in Texas shale gas areas despite the very great expansion of shale gas development there. Bredfeldt responded that Texas has been extracting oil and gas for a long time, so it has the best
record for developing and implementing these technologies safely. She noted that the biggest problems are user error and that these do not occur when operators take care. She said that the state’s monitoring is extensive enough to detect emissions.
Behavioral changes. In response to a question about Brown’s clients’ receptivity to the information they were given on how to improve their health prospects, Brown said that although they do not monitor to see if the clients had made recommended changes, he believes that once they are confident that the project is focused on their health, they do pay attention. He reported anecdotally that they do buy meters, try to filter the air in their homes if they can, take their shoes off when they enter the house, and so forth. But he added that to get these changes, you must speak with them in person. He added that his project plans long-term follow-up on the question of behavioral change.
Bowen is Ecosystem Dynamics branch chief at the Fort Collins Science Center of the U.S. Geological Survey. He began by acknowledging a large group of collaborators in federal agencies and elsewhere who contributed to the content of his presentation, which is based on examinations of unconventional oil and gas development generally, not only hydraulic fracturing for gas. He emphasized the variety of ecosystems in which deposits exist and noted that over the past century, oil and gas development has occurred in many areas where new deposits are being developed or may be.
Terrestrial ecosystems. The direct effects of shale gas development on terrestrial ecosystems include removal of habitat, fatalities of animals by collisions with equipment, and the introduction of invasive species, typically plants, brought in by disturbance of soil and by human activity that may introduce seeds. Indirect effects may result from dust generated by trucks and construction activities, noise, light, avoidance by wildlife species of the development area, and at a larger scale, habitat fragmentation that can alter habitat use. The end points of concern are physiological changes that affect survival or reproductive success. Cumulative effects can result from the increasing scale of activity.
Bowen said that surface disturbance is easily quantified and that several research methods are available to measure it and estimate its effects.
These include spatial analysis (mapping and estimating development patterns); species-based modeling of population changes, behavioral responses, and habitat modeling; vulnerability assessment of habitats and of species distributions; and ecoregional assessment that considers multiple species and multiple drivers of change in larger geographic areas. Analysts increasingly realize the importance of incremental development, which can have effects going beyond those at particular well sites. Bowen illustrated some difficulties of measuring surface disturbance due to such factors as different types and degrees of disturbance at different spots at a well site or a single well pad being connected to multiple wells, which often creates greater surface disturbance than does the pad itself. He noted that renewable energy development, such as for wind energy, also creates significant surface disturbance and that there has been little measurement of this.
Bowen illustrated the state of research on terrestrial ecological impacts with a few recent studies. A study of the effects of gas development in Wyoming on mule deer habitat (Sawyer et al., 2006) showed that the deer avoided areas around wells and roads and moved into areas they formerly did not use. Further studies of the effects on migration routes, winter range, and reproductive success are being used to consider strategies to mitigate effects. Other research in the West has considered the habitat of sage grouse, a species being considered for federal listing as endangered or threatened (e.g., Knick et al., 2013). This research has quantified the extent to which the grouse avoid areas of human development, including areas with power lines, pipes, etc. Other research overlays changes in the landscape with model-based predictions of habitat suitability for species of interest, to develop maps of vulnerability. Ecoregional assessments consider effects at larger scale and involve mapping the concentrations of species of interest against maps of development. He emphasized that the location of disturbance matters a lot to the ecological effects.
Bowen briefly described a water quality study at watershed scale in which he is involved. It found that of 837 watersheds examined that were potentially affected by shale development, only 153 had adequate water quality data to look for trends. This finding indicates that the existing network of water quality monitoring stations is not extensive enough to evaluate effects of shale gas development on water quality in enough places to draw general conclusions.
Bowen summarized by making these points: (1) the distribution of shale gas resources and the methods used to develop the resource determine potential surface disturbance; (2) habitat requirements and behavioral responses to development are species-specific; (3) species responses must be known or estimated to predict effects of development, but population responses are difficult to predict precisely; (4) vulnerability of
species, communities, or ecosystems to potential development is typically assessed by examining areas of overlap and optimally considers the sensitivity of the affected species; and (5) ecoregional assessments examine multiple natural resources and are potentially useful in identifying priority areas for development or conservation. Bowen closed with the statement that we know a lot more than we used to about species responses.
Aquatic ecosystems. Aida Farag, a fish biologist who is station leader at the Jackson Field Research Station of the U.S. Geological Survey, spoke about aquatic ecosystems, pointing out that several stages of the hydraulic fracturing water cycle can affect them. She noted a history of looking at effects on aquatic organisms: the National Pollutant Discharge Elimination System under the federal Clean Water Act includes a permitting process for potential discharges into surface water under which the discharger can be required to test the effluent on aquatic organisms to determine whether the water is acceptable for discharge. Shale gas development can directly affect water quantity, quality, and infiltration through the levels of salts and trace organic compounds in produced waters. Indirect effects may include effects of produced water on the solvent absorption ratio and the ability of waters downstream to absorb water, the effects of dissolved solids and trace metals, alterations of flow rates and seasonal cycles, reduced diversity of habitat patches, and increases in non-native species.
Farag illustrated aquatic effects with photographic illustrations and reported findings from some studies. She noted that produced water is useful in some cases for agriculture but can also be toxic to aquatic organisms. Work to assess the effects begins in the laboratory and then moves to the field; it progresses from studies of individuals to studies of populations. At the intersection of laboratory and field studies and of individual and population effects are the mechanisms of toxicity, such as effects on ionoregulation, enzyme effects, and effects on estrogen and androgen receptors. Toxicity thresholds are set based on laboratory studies first and then on effects found in field settings. A study of a fish kill in Kentucky (Papoulias and Velasco, 2013) illustrates the histological and physiological effects of exposure. Such studies are preludes to basin-wide studies using in situ bioassays and reporting survival rates of species in untreated discharge water. Farag said there have been recent studies of brine contamination in wetlands near development sites and of chemicals in drinking water. Although these studies have not examined ecological effects, such studies could be conducted.
Farag concluded her presentation with the following points: mitigation of surface disturbance can maintain diversity of aquatic habitat patches, an integrated scientific approach needs to balance beneficial use with potential toxicity, studies defining mechanisms of toxicity at the indi-
vidual level can provide explanations and possibly provide early warning at the population level, establishing toxicity thresholds and then conducting field studies can expand understanding, and long-term water quality monitoring data are essential for estimating effects on aquatic ecology.
Brittingham is professor of wildlife resources and an extension wildlife specialist at the College of Agricultural Sciences at the Pennsylvania State University, with research interests that include the effects of habitat fragmentation on bird populations. She discussed some ecological issues that are appearing in the eastern United States. In the East, and particularly in Pennsylvania, Brittingham said, there is complete overlap between areas of shale gas development and core forested areas, which have high ecological value. She especially noted the ecological importance of neotropical migrant songbird populations to the forest for insect control and other ecological reasons and the importance of amphibians. For example, 18 percent of the world population of scarlet tanagers breed in Pennsylvania. She expressed the goal of having shale development and restoration proceed in ways that maintain these populations because maintaining them is much easier than restoring them.
Brittingham noted that gas wells change the landscape very differently in the East than they do in the West and that the ecological effects of deep shale gas development are very different from those of shallow gas development. Shallow wells tend to take about one-fourth acre each, and forest cover restores fairly easily, she said, compared with the industrial style of deep gas development, which uses pads plus other local disturbances covering an average of 6.7 acres and sometimes as much as 50 acres, not counting the large impoundments that are sometimes present. The habitat fragmentation pattern is also very different with deep gas development—for example, roads are much wider. The extent of disturbance is indicated by the number of pads being built: over 2,350 in Pennsylvania between 2005 and 2011, half in farmland and half in forest land, with about a quarter going into core, formerly unfragmented, forest. She expects that roads will have a larger ecological effect than the pads themselves. For some species, they will act as corridors of dispersal, invasion, or hunting; for others, such as amphibians, they will act as barriers. A study in Alberta found, for example, that pipelines and roads increased wolf predation on caribou. The width of the corridors probably determines the ecological effects.
Brittingham cited a study in Bradford County, Pennsylvania, that found that loss of core forest (i.e., forest more than 100 m from a for-
est boundary) due to pipelines occurred at a rate twice that of the loss of overall forest. Roads have similar effects. A recent overview paper (Northrup and Wittemyer, 2013) offers a general summary of knowledge about effects of energy developments on wildlife and identified these major concerns: habitat fragmentation, the balance of species (with fragmented habitat favoring generalist over specialist species), the spread of invasive species, disturbance to sensitive habitats, and negative effects on biodiversity and ecosystem functions. Her group’s research so far indicates a decline of forest interior species, an increase in human-associated species, and no change with early successional species. Invasive plants are showing up at 60 percent of surveyed pads, with their appearance dependent on road access. Some areas become drier and some wetter, affecting habitat for amphibians as well as stream erosion and flooding patterns. Increased noise and light, strongest during well construction but continuing near compressor stations, can affect songbird territories and the reproductive success of some wildlife species, with some benefiting and others losing out.
Brittingham mentioned some clear nonecological differences between the East and West that have implications for ecological effects. Much of the development in the East is occurring on private lands (93% of the pads in Pennsylvania are on private land). Private landowners generally lack the resources for planning that public landowners have, and governments cannot control the location of roads on private lands even if they know which locations would minimize ecological impact. Also, the surface land owner often does not own the mineral rights. These factors increase risk and uncertainty compared to the situation in the West. She also noted that many sites in Pennsylvania have only one to two wells. This pattern may be occurring because companies need to show some activity to keep their leases, and it may indicate that habitat disruption will continue for several decades before completion occurs. In Pennsylvania, only 16 percent of pads have been reclaimed and most reclamation is to grass, not to forest.
Brittingham concluded her presentation by identifying the following research needs: studies of thresholds for change for different species and groups of species, mechanisms underlying species responses (avoidance, mortality, reproductive disruption), and restoration methods (including intermediate restoration while development is occurring).
Participant questions and comments opened several issues for discussion, which are reported below under the headings of collection of baseline data, possible factors affecting ecosystems, modeling ecological impacts for decision making, and the ecological importance of the Appalachian area.
Collection of baseline data. A participant asked what triggers the collection of baseline data on private lands, which are dominant in shale gas areas in the East. Susan Tierney noted that the Secretary of Energy Advisory Board on shale gas development, on which she served, suggested that a triggering process be used to collect baseline data, similar to what states use when development processes trigger environmental reviews. Brittingham added that state agencies fund monitoring, but there is no system for looking at the ecological changes occurring.
Possible factors affecting ecosystems. A participant asked if research has examined effects of belts that include roads, pads, and pipelines, to determine any cumulative effects on particular species. Brittingham said that researchers at Pennsylvania State University are developing indexes of fragmentation for 3 × 3-mile blocks and that the Nature Conservancy is also looking at fragmentation at various scales to develop measures of this type. Bowen added that the U.S. Geological Survey is trying to do ecoregional assessments that capture all sources of disturbance, but because of gaps in knowledge, a lot of seat-of-the-pants thinking is still needed to determine the sizes of buffers needed to protect particular species.
In response to a question about ecological effects of the conjunction of shale gas development and climate change, Brittingham said this is an important point to examine, particularly in Pennsylvania, because it is at the southern border of the range for many northern species and at the edge of stress from several harmful tree pests. In response to a question about whether the per-gas-unit surface impact of deep gas production differs from that of shallow production, Brittingham noted that the landscape change is structurally very different, in addition to differences in scale.
Modeling ecological impacts for decision making. A participant said that the Nature Conservancy is developing a model to help companies locate their sites in ways that would allow them to incorporate habitat considerations in addition to economic factors and regulatory constraints. This participant asked whether the science is well enough developed to allow models to deal with regional-scale ecological issues. Brittingham said that enough is known to allow scaling up, but that it is not clear how flexible industry is regarding where pipelines go, whether companies can share pipelines, and other larger-scale issues with ecological implications. She noted that the model is being built to allow modification as more is learned.
Robert Winthrop of the Bureau of Land Management said that his agency’s ecoregional assessments examine large areas; as an illustration of size, one such assessment is for all of eastern Utah. The assessments,
which consider certain biological and physical conditions and change agents, can be used to identify areas needing protection in the development of oil and gas master leasing plans and in setting conditions for leases—for example, by requiring phased development—that consider various kinds of values. In the Utah case, these include scenic values.
The ecological importance of the Appalachian area. In response to a question about the ecological importance of the Appalachian area compared with other regions around the planet, Brittingham, while noting that she has an Appalachian mindset, said that this area does have global ecological importance. She noted the concerns with worldwide amphibian decline and said that the Appalachian basin is the heart of salamander distribution; similarly, neotropical migrant songbirds depend on this area and the boreal forest of Canada, which is also being disrupted.
Newell is Gendell professor of energy and environmental economics and director of the Duke University Energy Initiative. He previously served as administrator of the Energy Information Administration (EIA) at the U.S. Department of Energy. Newell characterized his presentation as a first foray into understanding the impacts of shale gas development on climate change. It does not consider other risks of shale gas development or compare these risks with the risks of other energy technologies. He distinguished two main questions to be answered: accounting for the greenhouse gas emissions associated with shale gas development, and the implications of that accounting for decisions being made by producers, policy makers, equipment manufacturers, and individual consumers. Accounting efforts include those that seek to estimate total life-cycle emissions and those focused at the sectoral level (e.g., comparing emissions from gas versus other technologies for particular purposes, such as for power generation). To inform decisions, Newell said, we need to consider the future both with and without changes in policy.
Greenhouse gas accounting. Newell explained that the available evidence includes baseline national emissions statistics, EIA data, studies from academia and NGOs, technology life-cycle analysis, and energy modeling projections such as those in the EIA’s Annual Energy Outlook and projections from the International Energy Agency. The EIA models incorporate several scenarios, including a reference case and other cases that vary the
outlook for gas and oil. The high oil and gas resource case (which includes development of tight oil)11 doubles shale gas development over the reference case. Similarly, the International Energy Agency offers a world outlook that includes a case with greatly increased gas development globally.
In the United States, Newell continued, total natural gas use divides roughly in thirds among power generation, industry, and commercial/residential uses, and substitution can occur in any sector, not only in power generation. In 2011, natural gas accounted for about 26 percent of U.S. carbon dioxide and methane emissions. Globally, shale gas could technically increase reserves by 40 percent, although with current costs and prices, the economically recoverable resources are much smaller. Almost all the current development is in North America. Despite interest, exploration, and development in many other countries, Newell said, there is so far not much production. Shale gas has gone from virtually a zero share of U.S. gas production in 2005 to about 35 percent now, with one model projecting that it will reach about 50 percent by 2040. The effects of these developments have included a significant decrease in actual current prices of gas, as well as lower projected future prices of natural gas out to at least 2040.
The effects on climate are both direct and indirect. Lower prices, Newell said, cause fuel substitution of gas for coal, oil, renewable energy sources, and nuclear power—so substitution affects sources that are both higher and lower than shale gas in greenhouse gas emissions. Lower prices also lead to lower overall energy prices and therefore to increased overall energy consumption. Combining these impacts and their consequences for carbon dioxide and methane emissions (from both extraction and combustion processes) yields the net climate impact. Policy also impinges on these effects by affecting emissions, technologies, and production processes.
Newell noted that modeling these effects on a complex system requires some assumptions. He pointed out that in the U.S. economy, natural gas expenditures are 13 percent of energy expenditures and 1 percent of the total economy, which suggests that lower gas prices will increase gross domestic product (GDP), but probably not by much. He said that the substitution effect (using gas instead of other energy sources) will likely dominate the aggregate demand effect. Aggregate energy demand is mainly driven by population growth, overall economic growth, and the share of the economy represented by manufacturing versus services. The effects of prices are represented in economic models as demand elasticity: the aggregate demand effect is the percent increase in consumption associated with a percent decrease in price. EIA modeling indicates a very low
11“Tight oil” refers to underground petroleum sources such as oil shales and tar sands.
elasticity of aggregate energy demand with respect to natural gas price changes (elasticity less than 0.1, or a 1% increase in aggregate demand for a 10% decrease in price), a low-to-moderate elasticity of natural gas demand with respect to natural gas prices in the residential/commercial (elasticity less than 0.3) and industrial sectors (less than 0.5); and a much greater elasticity of demand for natural gas for electricity generation (1.5 to 2.5).
In the EIA models for the high-resource case, which assume a doubling of expected recovery of natural gas, gas prices are about 45 percent lower in 2040 compared to the reference case. Total energy use goes up by 3 percent, GDP increases by 1 percent, and cumulative emissions between 2010 and 2040 in carbon dioxide equivalents decrease by 0.4 percent. This modeling result, Newell said, indicates that fuel substitution dominates the other effects, leading to lower emissions, though not by much. Using other target years before 2040 presents a similar picture, he added.
According to the 2013 U.S. EPA greenhouse gas inventory, Newell said, 87 percent of greenhouse gas emissions from natural gas come from combustion. Noncombustion emissions have been variable, but have decreased in the past few years: Newell reported that according to the 2013 inventory, upstream emissions have decreased by 11 percent since 1990 per unit of production. He noted, however, that these estimates have fluctuated from 1 year’s inventory to the next due to changes in methods of calculation. So different studies may get different results, depending on which year’s EPA estimates they use.
Newell presented results from Weber and Clavin (2012), who compared several published estimates of noncombustion greenhouse gas emissions from conventional and unconventional gas development and found considerable variation across studies and no consistent difference across studies between shale gas and conventional gas. Newell indicated that if the latest EPA estimates had been used in all these studies, the average emissions levels would have been lower than they appeared in the studies as published.
Looking by sectors, Newell said, the studies mostly indicate that emissions from combustion are 40-50 percent lower with gas than with coal. The debate has mostly been about upstream emissions. The one outlier is a study by Howarth and colleagues (2011), which concluded that gas was worse than coal for greenhouse gas emissions. Newell thought that it may be an outlier for one or more of the following reasons: Howarth and colleagues used a 20-year global warming potential, rather than the conventional 100-year horizon; they assumed a relatively high methane leakage rate and assumed that it is all vented rather than flared; and they did not account for the greater efficiency of natural gas combustion compared to coal for power production.
Newell said that coal-based power production has decreased in the United States by 496 GWh between 2005, when serious production of shale gas started, and 2012. There have been accompanying increases in gas-fired power production of 470 GWh and in power production from renewable sources of 138 GWh, as well as a reduction in power from petroleum of 87 GWh. National greenhouse gas emissions are now said to be the lowest they have been since 1992, Newell said, adding that this is the combined effect of recession, lower gas prices, and increased regulation affecting coal power. The data, Newell added, indicate that fuel substitution of gas for coal has so far outweighed any effect of gas on power generated from renewables.
The EIA analyses, said Newell, suggest that in the power sector, a high oil and gas scenario reduces cumulative greenhouse gas emissions by 5 percent through 2040, for the same reasons as in the whole-economy model. As a rule of thumb, he added, if gas replaces more coal than nuclear, there is a net benefit for climate change, and this is what this model projects.
In the residential and commercial sectors, the evidence reviewed by Newell suggests that gas-fueled space and water heating has lower greenhouse gas intensity than electricity-generated heating, though this outcome depends on where in the country the gas is used. The modeling results are similar to what they are for the power sector: in the high oil and gas scenario, cumulative greenhouse gas emissions are 3 percent lower than in the reference case.
In the transport sector, studies comparing gas-powered and gasoline-powered light vehicles show decreased life-cycle emissions for gas of around 10 percent, according to Newell. In a similar comparison for heavy vehicles, which are conventionally diesel-fueled and more efficient, he explained, natural gas does not do as well as a substitute fuel. The industrial sector shows some of the same dynamics, he said, but the EIA model projects a slight increase of 0.3 percent in cumulative emissions.
International implications. Newell said that the international implications of all these model projections are important. The International Energy Agency’s “Golden Age of Gas” scenario indicates that by the end of its projection period in 2035, greenhouse gas emissions will be 3 percent lower than under a reference scenario (International Energy Agency, 2012). These models involve a lot of behavioral assumptions about substitution that need to be examined, Newell noted. For example, there is a concern about U.S. coal exports, which will have a net climate effect only if they affect global coal prices; if they do not, they will only substitute for coal from other sources. U.S. coal exports account for only about 5 percent
of the international coal trade, Newell continued, so U.S. exports seem unlikely to have a major effect on global coal prices.
U.S. policy implications. Newell concluded by discussing some policy implications. Low natural gas prices may make certain climate policies easier and others more costly. For example, lowering the price of natural gas will make it easier to meet some climate policy targets because it substitutes for coal. It may, however, increase the relative cost of renewable energy standards, which will increase with low-price gas. To achieve substantial long-term goals for reducing greenhouse gas emissions, Newell said that increased use of natural gas would need to incorporate carbon capture and storage at reasonable cost to continue as a competitive option.
Newell summarized by saying that the greenhouse gas emissions intensity of natural gas has fallen and that further reductions in noncombustion emissions and improved combustion efficiency are feasible and could further this trend. Thus far, he said, shale gas has led to decreased greenhouse gas emissions by lowering prices and displacing more coal than renewable and nuclear energy sources; using current life-cycle estimates, natural gas tends to lower greenhouse gas emissions relative to coal-fueled electric power generation, gasoline-fueled personal vehicles, and electricity for space/water heating. But natural gas abundance alone will probably not, according to Newell, have a substantial effect on future greenhouse gas emissions. He sees policy as the key factor and added that natural gas abundance could influence relevant policy in ways that could have a substantial effect on future emissions.
Bordoff is a professor of professional practice and director of the Center on Global Energy Policy in the School of International and Public Affairs at Columbia University. He formerly served on the White House National Economic Council and the Council on Environmental Quality. He emphasized that the key questions about climate effects are how much gas actually is available, what the substitution and demand effects of increased supplies will be in the United States and overseas, and how much better natural gas really is from a climate perspective.
Every piece of data suggests that shale gas is booming and will continue to do so, he said. The latest estimate of U.S. reserves is 26 percent greater than the previous one. New estimates continue to surprise on the up-side, Bordoff continued, though uncertainty remains about decline and recovery rates from unconventional wells. He said that substitution of gas for coal, changes in the economy, and wind-energy development
are among the main reasons for the decline of U.S. carbon emissions of about 12 percent since 2005.
The question of climate effects is, according to Bordoff, mainly dependent on the net of substitution and demand effects of increasingly available gas. Even if the net effect in the United States, as Newell’s presentation indicates, is a reduction in emissions, Bordoff said that this will not “solve global warming.” In the high-supply case, Newell’s slides show only a 0.4 percent reduction in greenhouse gas emissions. The number is probably as small as it is, Bordoff surmised, because hydraulic fracturing produces oil as well as gas and the rise in oil production counteracts some of the net benefits of increased gas production.
Bordoff agreed with Newell that the study by Howarth and colleagues (2011) of fugitive methane releases is generally seen as an outlier. Most studies show that natural gas is roughly half as carbon-intensive as coal for power generation; with recent EPA rules on green completion, he said, this proportion will decrease, though much remains unknown about leakage in transmission and distribution. Still, Bordoff said, the sources of leakage, such as fugitive methane, can be reduced at a fairly low cost. Thus, in all, he concludes that shale gas production in the United States produces a net climate benefit.
At the international level, Bordoff said the questions are also about substitution and demand effects of increased gas availability, whether locally sourced or imported. There are enormous shale reserves in China, Argentina, and elsewhere, but he believes development will take time for reasons including difficult geology; requirements for water and for industrial and transport infrastructure; investment policy; and human capacity. There are reasons for pessimism about shale gas development in Europe, he said, partly because of difficult geology and partly because the gas there is mainly “dry,” without the valuable hydrocarbon liquids. Even though development will take time, he expects the global supply of liquefied natural gas to increase. Gas prices in Europe and Asia, which have been much higher than in the United States, can be expected to drop because liquefied natural gas that might otherwise be exported from the Persian Gulf to the United States will become available for export to Europe. If more liquefied natural gas flows to the Pacific Rim, Bordoff continued, there could be downward pressure on gas prices there as well, leading to substitution from coal.
The International Energy Agency’s Golden Rules for a Golden Age of Gas study (International Energy Agency, 2012) projects a three percent decline of global emissions, due to the combination of natural gas substitution for coal, oil, nuclear, and renewables, and increased demand. Bordoff said that it would probably take at least 5-10 years for new gas-fueled power plants to come online globally. He noted that electricity demand growth
is starting to slow in China, which implies a decreasing need for new power plants—unlike the United States, where there are many old power plants needing replacement, which makes gas-fueled power generation look more attractive. He believes that substitution for renewables and nuclear energy will mainly be policy driven rather than price driven. For example, if the European Union continues its renewable-energy mandate and its subsidies for nuclear power, substitution for these power sources would be less intense than in scenarios where greater substitution would increase the climate benefits.
In his summary, Bordoff said that although the effects of increased gas supplies on climate push in both directions, on net, they are still positive. The most important point, he concluded, is that whatever the economic effects, the main impact of natural gas will be to make it less expensive to enact policies to solve climate problems.
Many issues arose in the discussion, and the rapporteur has organized these under headings of costs to consumers, responses in the industrial sector, methane releases, effects of particulates, liquefied natural gas exports, global economic development issues, and “what will be the ultimate effects on humanity?”
Costs to consumers. A participant noted that for consumers in some states, much of the retail price of gas is due to distribution and delivery costs and asked about the implications of this for the analysis. Newell agreed that the proportion of costs to consumers due to the price of gas vary greatly and are sometimes relatively small, so that decreased gas prices sometimes have a relatively small effect on prices to consumers. He said that these factors are built into the demand models.
Responses in the industrial sector. A participant said that there is very little in the literature on the market effects of the one-third of gas that goes to industry. Although those markets will respond in the short- and long-term, we don’t know to what extent, the participant said. Newell agreed that there has been relatively little attention to substitution in industry, adding that some companies are relocating internationally and analysis needs to consider what emissions would have been if they had not moved.
Methane releases. A participant proposed that the uncertainties about methane releases are greater than the presenters suggested and expressed the opinion that the EPA data and the studies built on them do not repre-
sent the range of uncertainty. He said that a number of ambient-air studies show emissions in several basins of over five percent, indicating that the EPA-based numbers may be considerably too low. He also pointed out that studies are not looking at methane emissions from coal-fired power plants, which needs to be done for a fair comparison within the power sector. Newell agreed that there have been significant variations among studies. Bordoff added that there are methodological issues with the EPA analyses of methane, as well as with the ambient studies. He reiterated his view that whatever the emissions are, the problem is solvable at relatively low cost.
Another participant asked how much is known about fugitive methane emissions from old and abandoned wells and what might be done to get data on that. Neither presenter could answer the question, but Bordoff pointed out that the problem is not new: methane emissions have long been an issue with conventional gas wells.
Effects of particulates emitted from coal-based power production. In response to a participant’s question, Newell said that his and other studies comparing gas and coal in power production have not looked at particulates, but his sense is that they are a small contributor.
Liquefied natural gas exports. A participant noted that these exports are very greenhouse gas-intensive and need analysis. Newell expressed the view that although liquefied natural gas is more energy-intensive than pipeline gas, the difference is not enough to outweigh the benefits of substituting gas for coal in electricity production. Bordoff added that liquefied natural gas exports to the Pacific Rim would likely have a net climate benefit, though there is considerable uncertainty about that.
Global economic development issues. A participant asked about the effects of gas supply and price in places where more people are seeking middle-class life-styles. Newell reiterated his view that the demand effect will not be the major one globally but that once people can afford such things as motor vehicles and domestic water heating and electricity, the main issue will be which fuel is used to meet these new demands.
What will be the ultimate effects on humanity? A participant said that because both gas and coal are fossil fuels, the debate about which has lower emissions is somewhat beside the point. She noted that even the Golden Age of Gas report indicates that a golden age for gas will not be a golden age for humanity. Bordoff responded that global economic growth and decreasing poverty are also policy objectives, and he pointed out that cheaper energy helps achieve these objectives. Susan Tierney, the modera-
tor of the discussion, commented that she had heard no one suggest that shale gas was a solution to climate change.
Jacquet is assistant professor in the Department of Sociology and Rural Studies at South Dakota State University. His research focuses on social and economic impacts of energy development. His presentation aimed for a broad overview of what is known about the blessing and “curse” of natural resources to communities, discussing four types of risks and identifying four main gaps in knowledge. He pointed out that although community outcomes are usually long term, shale gas development is relatively new, so most of the relevant knowledge comes from other types of energy development in the past and from studies of environmental contamination and technological disasters.
Blessings and costs of natural resources. Jacquet indicated that the blessings are mostly related to jobs and economic stimulus. Natural resource development increases employment and tax revenues, especially in rural areas lacking other opportunities, but these blessings are relatively short term and come in booms and busts. The costs are longer term, he said, and include volatility, instability, and de-diversification of the economy; higher long-term unemployment, poverty, and inequality; and lower educational attainment compared to similar areas not experiencing natural resource development. World Bank country-level data show a negative correlation between fuels, ores, and metals as a percentage of national exports and economic growth rates, Jacquet said, and county-level data in the United States support the resource curse hypothesis, showing that total personal income in energy-focusing counties follows boom-bust cycles while other counties show steadier income growth. Data on nonlabor income shows a slowly increasing gap, he added, favoring nonenergy-focusing counties. A meta-analysis of 369 studies of economic impacts (Freudenburg and Wilson, 2002) showed that resource-dependent counties fared slightly better than other counties on measures of income but worse in terms of poverty, unemployment rates, and overall economic performance.
Boom town effects. The risks to communities of rapid industrialization, found by Jacquet in studies done over the past several decades, involve strained municipal services, poor quality of life, out-migration of residents, and a legacy of overbuilt and unplanned construction. Jacquet said that resource boom towns show rapid short-term population growth, with
the effects of this depending on the initial population that can absorb the impact, the pace of development, and the availability of funds to mitigate the impacts.
Inequality. Inequalities in the distribution of costs and benefits tend to increase with time, Jacquet said, and they often produce what Freudenburg and Jones (1991) called “corrosive communities.” Landowners may get benefits that others cannot. In some places, split estate—the disconnection of surface land ownership from subsurface minerals ownership—is likely to contribute to the inequality effects, especially where mineral owners do not live in the community. The results, Jacquet continued, include fierce community conflict, distrust, litigation, uncertainty, and confusion about what is happening, blame-placing, and distaste toward those who benefit. The community conflict is often worse than the environmental impacts: community decision making suffers, communication breaks down, scientific facts become harder to obtain because of litigation, and out-migration and disinvestment often occur over the long term. Much social science literature, he added, indicates that unequal distribution of costs and benefits affects perceptions of risk and harm, attitudes about acceptability of the activity, perceptions of fairness, and trust within communities.
Jacquet’s (2012) research among landowners in Pennsylvania found that 60 percent of those without gas development on their land thought development was making the community worse, people with leases but not yet with development were split in their opinions, and people with leases and development believed gas development was making the community much better off. General environmental attitudes also had a major influence on these judgments, he said.
“Contaminated communities.” Research on communities that have been stigmatized as contaminated (regardless of actual contamination levels), such as the Love Canal community in New York and the area around Three Mile Island in Pennsylvania, indicates that this experience has powerful effects on residents’ self-image and subjective sense of well-being (Edelstein, 1988). This kind of stigmatization is evident in numerous communities where shale gas development is occurring, Jacquet said.
Stress. Data on public health impacts often do not address social-psychological stress, said Jacquet, but there are effects on powerful place-based identities related to ideas of “what kind of place do I live in, what is my role in the community, and who are in my social circle?” A study in Gillette, Wyoming, in the 1970s (Weisz, 1979) found that on a self-reported stress scale, the average resident scored as having “major life stress” even though there was no environmental contamination. Half of the people
with that level of stress reported physical illness, compared with nine percent of people who were not stressed. Jacquet added that stress was found to be among the greatest impacts of gas drilling in Garfield County, Colorado. Around the Exxon Valdez oil spill, many people experienced symptoms of post-traumatic stress disorder.
In concluding his presentation, Jacquet said that the risks of rapid shale gas development are likely to be broad-based and to operate through social, psychological, and economic mechanisms. They are long term in nature, and perceived inequity is a major stressor. He emphasized that with these effects, perception is reality because it is perception that causes stress.
Jacquet identified four knowledge gaps: (1) What happens to the generated wealth? We know that wealth is generated, but know little about how or whether it stays in these communities. (2) What are the magnitude and effects of stress? We know that community change creates stress, he said, but we do not know its magnitude compared to other types of stress people are experiencing or its effects on health, conflict, and economic development. (3) What are the long-term effects of corrosive communities, inequality, and stigmatization on disinvestment or in- and out-migration, and how can these negative effects be overcome? (4) What is the long-term development picture? For example, should communities plan for multiple booms and busts? Jacquet concluded that targeted funding is needed to address these issues, conduct longitudinal analyses, revisit previous studies, and assist communities in planning.
Christopherson is an economic geographer and a professor in the Department of City and Regional Planning at Cornell University whose research interests include economic development in older industrial regions, including the impact of natural gas drilling in the Marcellus shale. Her comments focused mainly on knowledge needs regarding community impacts. She emphasized that very few academics have studied community impacts in shale gas development areas, the topic is very controversial, and it is important not to neglect the intangible effects. She stressed the importance of an issue she heard in many of the presentations at the workshop: risks extend far beyond the well site. She cited risks related to trucking, sand mining, and other activities outside the shale gas plays and pointed out that 27 states have regions that will be affected by shale gas development. Also, although the costs will be concentrated in certain regions, benefits, especially in the form of lower gas costs, will be distributed across the entire population.
Christopherson agreed with Jacquet that available knowledge, which is mainly based on what has happened in rural communities, indicates
that the long-term effects on communities are generally poor. However, many of the shale gas sites in the 27 states are not in rural communities, and these places may be affected differently.
She noted that the available information has mainly come from case studies, which do not provide the strongest evidence. Some aspects of communities can be measured quantitatively (e.g., crime), but other important aspects have not been measured well, if at all. One is the stage of development of the shale gas resource, which cannot be measured by consistent methods across states. Another analytic problem involves understanding the financial impacts, which requires understanding who owns the land and mineral rights and where they live. A recent study by researchers at The Pennsylvania State University indicates that only 25 percent of the people obtaining lease and royalty payments live in the communities where the leases apply. Thus, Christopherson concluded, the economic data do not tell who is benefiting economically and who is assuming the risks.
Effects on pre-existing industries, particularly tourism and agriculture, and on the jobs in those industries, seem to be negative, she said, but have not been carefully measured. Long-term public costs (roads, professional emergency services, public safety and crime control, administrative and monitoring costs, health care) seem to be significant in some places, but again, there are no good data. Another unmeasured type of risk relates to local control issues. In New York, for example, many communities have taken action, usually passing development moratoria, out of distrust of the state and the industry. Christopherson concluded by saying that even though the wide variety of shale gas plays will have to be governed in many different ways, comparative information is needed to inform these choices.
Topics and issues raised by participants’ questions and comments are summarized below under six headings: effects of in-migration, community health effects, community-level effects, community relations with industry, “social capital” and resilience, and possible best practices.
Effects of in-migration. A participant asked about the effects of in-migrants on fishing and hunting pressure on wildlife. Jacquet noted that although actual data are hard to get, the demographics of the new workers suggest hypotheses: the workers are mostly young and male and come from areas that are more urban than the communities in which they are working. Christopherson cited other ways the incoming workforce can be very different from the pre-existing local population. In Pennsylvania, for
example, local hospitals have had to hire Spanish translators to deal with incoming workers of Mexican background. She noted that communities can respond better if they have a good idea of what they will face and what it will cost.
Community health effects. A participant said that it has been impossible to get worker health statistics for this industry because the work was so heavily subcontracted that it is hard to identify the workers. Another commented on the view that health effects in affected communities are mostly due to stress, saying that it is hard to tell whether this is true. He cited an example of a worker who presented serious neurological problems that could not be fully diagnosed because it was impossible to get access to the data on the chemicals to which the worker may have been exposed. Some of the most serious health effects are from accidents, he added, which are not stress-related. In addition, some stress-related responses, such as family break-ups, are not normally reported. Jacquet agreed that some of the health effects are due to contamination and some to stress, but he said that we do not yet know how large each portion is, or what the total is. Christopherson added that federal data collection is needed. She said that the Occupational Health and Safety Administration has not been engaged, which is one reason there is so little information about the risks workers face.
Community-level effects. A participant who had studied impacts in Pennsylvania said that communities are disintegrating: people have lost their social networks, and many people who have not yet left but who have developed plans to leave are the people the community most needs. He added that school attendance rates have fallen in the shale gas communities he studied.
Community relations with industry. A participant asked how community relations with the industry are affected when the companies in the industry are small and when the owners of the mineral rights are not in the community where the gas is extracted. Christopherson said that there are good relations between communities and some operating companies and wondered whether large oil companies behave better in Texas, where their employees live, than in the East.
“Social capital” and resilience. A participant asked whether research on social capital, community resilience, and responses to disasters has entered into the work on community risks from shale gas development. Jacquet said that although lessons need to be drawn from past research on these topics, generally speaking that has not been done. Research-
ers who try to draw lessons from experience, even with other kinds of energy development, receive criticism that it does not generalize. Jacquet said that the research on “corrosive communities” by Freudenburg (e.g., Freudenburg and Jones, 1991) has generally become a lost literature and needs to be reexamined.
Possible best practices. A participant suggested that the long history of extractive industries in the United States should offer instructive experiences, from places that claim to have done reasonably well, for formulating best practices to address negative impacts on communities, anticipate problems, and spread the benefits. An example might be the use of receipts from taxes on new industries to support education and other local public services. Christopherson agreed that there are possibilities of this kind, although there are important differences between states: in Wyoming, taxes go to the state government; in New York, to localities. Jacquet noted that after the Western energy booms of the 1970s, some state governments changed policies with the apparent result that the negative effects of more recent booms were lessened. He added that in the East, states do not have that experience as a basis for adjusting their policies.
In response to a question about whether there are success stories from local communities, Jacquet replied that rebound sometimes occurs: a longitudinal study in Utah indicated that quality of life decreased during the boom but rebounded by 20 years later. However, he could not identify a shining example of success. Christopherson said that anticipation of the costs and developing a plan are the best strategies for limiting community impacts, but she noted that development often happens in places that are not prepared and lack government capacity. Many rural communities in the Marcellus shale region have volunteer mayors with no paid staff. Where there is greater governance capacity, she said, communities are in better condition to cope with risks.
A participant asked about impact fees as a mitigation strategy and wondered how Pennsylvania (the only state with impact fees dedicated to helping communities) is faring with that approach. Christopherson replied that right now, there is no relationship between public costs and taxes or impact fees. In Pennsylvania, a community gets $10,000 for every conventional well and $50,000 for every unconventional well, she said, with communities that oppose development not getting any impact fees. She said there are no data on how these impact fees compare with the actual costs communities face; New York is collecting baseline data on some community conditions, such as road conditions, and there are some agreements being made with companies to keep up local roads.
Jacquet offered as a positive example the community of Evanston, Wyoming, where in the 1970s Chevron rapidly developed oil and gas
wells, quadrupling the size of the town. Chevron created the Overthrust Industrial Association to help the community, rebuilt the police station, bought school buses, and spent many millions of dollars on community facilities through partnering with local officials. He said that although there are many such examples of socioeconomic mitigation, the investments usually pale in comparison with the mitigation investments for wildlife and are usually limited to what is required by the landowner or the government. In the East, he said, people are impressed when a company donates a few thousand dollars to the 4-H Club, but there are greater needs.
Another participant said that he had once suggested to the state and federal governments in Australia, where the state owns all mineral rights, that they should take funds from the government’s royalties from extraction to establish community trust funds. He asked what such funds should best be used for, if they were created. Jacquet said that trust funds can be a great solution and suggested that spending begin with community infrastructure (sewer and water systems, schools, health facilities, etc.), as has been done in Norway. This helps people see the community benefits of development, he noted. Christopherson suggested that the funds might also be used to level out boom-bust cycles, except where local officials are required to spend tax money in the year that it is received.
Krupnick is director of the Center for Energy Economics and Policy at Resources for the Future (RFF), where he is engaged in a series of studies related to the sustainable development of shale gas. He reported on some research recently completed at RFF, organized around a risk matrix that identifies activities (e.g., horizontal drilling, flowback and produced water disposal), burdens (e.g., air pollutants, fracturing fluids), intermediate impacts (e.g., on ground water, air quality, habitat), and final impacts (e.g., on human health, ecosystems, climate, quality of life). The matrix identifies 264 “risk pathways” from activities to intermediate impacts. For example, site development includes on-road vehicle activity, which creates burdens of conventional air pollutants and carbon dioxide, noise pollution, and road congestion and intermediate impacts on air quality and community disruption. The survey did not ask about final impacts.
The RFF project surveyed experts who worked in four types of organizations: environmental NGOs, research universities, federal and state government agencies, and companies in the industry. The survey
sought the people most knowledgeable about these risks, and 215 experts responded (30% of those asked). The experts were asked to identify the highest-priority risk pathways in terms of the need for risk reduction through government or industry action. They were also provided with 14 accident categories and asked to identify those most in need of mitigation. By comparing the top 20 risk pathways identified by each group of respondents, Krupnick said the study found a lot of consensus about which risk pathways most need attention: 12 pathways were in the top 20 for all four groups, which suggested that it would be possible to reach agreement to work on those. There were also six pathways prioritized by the industry group and no other group; these were all community effects, indicating the industry’s sensitivity to these effects. The project conducted several statistical analyses about the risk pathways, did a state-by-state regulatory comparison, and conducted a general public survey, the results of which Krupnick said were not yet available.
The “consensus” risk pathways included seven involving surface water, two involving ground water, two affecting air quality (both related to methane venting), and one involving habitat fragmentation. Seismicity was not a consensus concern. The people who identified themselves as “top experts” mostly agreed with the other experts but were also concerned about casing and cementing failures, either through leakage or accidents, leading to ground water contamination. These two pathways were among the top three concerns of all four groups of experts.
The comparison of experts’ judgments of the highest risks, defined by their judgments of the product of probability and consequence, indicated some differences between groups. For example, the NGO experts considered some risk pathways to be high in both probability and consequence, said Krupnick, but very few industry experts identified any pathways of that type and were mainly concerned with risk pathways they judged to be of low probability and medium severity. The main differences among groups were in the judgments of probability. The risks that were judged differently in probability were casing and cementing failures and accidents, impoundment failures, and truck accidents.
The RFF group also conducted a study of surface water quality risks that examined statistical relationships based on the locations of shale gas wells and water monitors over the duration of shale gas development in Pennsylvania (Olmstead et al., 2013). It looked for chloride and total suspended solids downstream from wells and for the impacts of shale gas waste treatment and release of materials from waste treatment plants. It found no significant effects of shale gas wells on downstream chloride, showing no indication that spills had created a systematic problem for water salinity. Krupnick noted that the study did find elevated chloride concentrations in waters below waste treatment plants and identified rela-
tionships between the presence of wells and downstream concentrations of total suspended solids.
Krupnick described in some detail the conceptual framework his group uses for thinking about types of risks. He distinguished cumulative risks (which arise when multiple risk pathways affect the same actors) from synergistic risks (which arise when multiple associated pathways act together to make things worse). He also distinguished scale effects and interaction effects. He noted that risks that arise from flows (e.g., water withdrawals or air pollution) may not be cumulative if the stock recharges quickly enough, but can be cumulative if the pace of development increases beyond the capacity of the system to recharge. He suggested that habitat fragmentation from pipelines is an example of a cumulative stock burden because as the number of pipelines increases, habitat fragmentation increases. He noted that this relationship can be highly nonlinear, in that the results of fragmentation may appear only when a threshold level is crossed. Nonlinearity of effects also arises with regulations, which establish a threshold beyond which costs increase, and with water salinity because beyond a threshold level, salinity may make the water useless for crops. Methane emissions from shale gas production also produce a cumulative risk; the potential for explosion is probably a threshold phenomenon, Krupnick added.
Interactions among risks include chemical interactions involving similar burdens that produce hazards (for example, volatile organic hydrocarbons and nitrogen oxides interact to produce ozone), physiological interactions (e.g., exposures combined with pre-existing disease), interactions of burdens involving dissimilar pathways (e.g., surface water withdrawals and pollution of the same stream can cause greater damage than either burden alone), and interactions between shale gas burdens and other things in the environment. Krupnick noted that cumulative risk reductions can also be important. Some industry actions, such as recycling wastewater, reduce risk from multiple pathways simultaneously. Nonindustry actors (e.g., exposed individuals moving away from locations with exposure) can also reduce risk via multiple pathways.
Krupnick summarized by noting that there is much consensus about which risks most need attention; that a lot is known about the magnitude of many of the risks, but much less about others (such as those of habitat fragmentation); and that there is a great need to think about risks cumulatively rather than only in isolation.
Perrow is an emeritus professor of sociology at Yale University whose research has focused on risks associated with structures and interactions
in large organizations and complex social and technological systems. He began his comments by saying that shale gas involves what some have called “destructive technology” in two senses. It brings innovation and introduces new products, so it is destructive to what it replaces. But it is also destructive in the usual senses of the word: its most important aspects are hidden from view, far below Earth’s surface, and are impossible to anticipate or monitor. Contaminated water travels thousands of feet underground and interacts with nearby abandoned wells in surprising ways: in one case, sending a geyser of methane 80 feet into the air from a well that no one knew was there. Induced seismicity potential also exists and can sever lines from nearby wells and cause leakage, a possibility that Perrow said the recent NRC report on seismic risk did not consider (National Research Council, 2013). Although that report found only one case of induced seismicity from fracking, Perrow said that other sources in the literature indicate many more cases.
Perrow stated that some risks, including from toxic substances added to fracking water, are deliberately hidden by the industry. In many states, he said, it only takes a declaration that these are proprietary information to prevent disclosure. This makes it impossible, he said, for a homeowner or a community to prove that water that poisoned their wells and livestock came from a fracking operation.
Adequacy of regulations. Because this new enterprise has grown very rapidly, Perrow said, adequate regulations are not in place and regulating unexpected interactions producing risks will be very difficult. He said that most regulations are state and local and that regulators at those levels are poorly staffed, trained, and equipped to deal with the risks. Also, the 2005 Energy Act exempted the industry from provisions of the federal Clean Water Act, so that when the U.S. EPA can intervene, it has to do so under legislation that is limited in scope, such as toxic waste legislation. Thus, he concluded, shale gas development is a new destructive enterprise without compensating regulatory institutions. He said that the economic and political power of the oil and gas industry is effective in convincing states not to regulate and that states and counties are in the awkward position of protecting humans and the environment when they are also interested in the economic benefits that the industry can bring. He added that in some states, local governments are prevented from preventing or regulating shale gas development because of the state’s economic interest.
Economic justice issues. Perrow stated that the benefits of this industry are unequally distributed to an extreme extent. He stated that the gas industry has received about $13.5 billion in subsidies in recent decades and that oil and gas combined are said to receive $10 billion in subsidies
each year. The gas industry is taxed at only 0.3 percent of its profits, which he said is probably close to a record low for any industry, even though this industry is one of the most profitable in the United States. Another economic justice question he raised involves property owners, some of whom receive large signer fees and royalties, while neighbors who do not receive the fees must use bottled water and live in fear of gas explosions. Property values decline (in some cases, by 75%), but full information about damages is lacking, he said, because gag rules associated with damage settlements hide information about the extent of damages. Perrow added that although local tax revenues can be substantial, they disappear in a few years after wells are exhausted. In Texas and Colorado, the state can deny petitions by landowners to prevent fracking activities on their land.
Methane releases. Perrow said that the industry releases methane at rates of two to seven percent and cited one scientific report estimating that it is more polluting than coal. He claimed that the public is not getting good independent estimates of methane emissions. Perrow agreed with what other speakers had said about methane leaks being correctable.
Corporate culture issues. He was pessimistic, however, about the possibility of changing corporate culture to reduce accidents. His research on organizational behavior indicates to him that the chances of changing corporate culture quickly for the better are very remote. He offered the example of BP, which had a massive leak from a facility in Prudhoe Bay, Alaska, after being warned by company engineers. Then, 2 years later in Texas City, Texas, BP had a deadly explosion after corporate leaders had been previously warned of the need for more safety and better maintenance at the plant there. Perrow concluded with an analogy to the case of nuclear power accidents, which he said present two levels of reality: the reality on the ground for local exposed people, who experience a great deal of damage, and the reality seen by governments and some nonprofits, which say that radiation levels are too low to be detected and are not a serious health threat. He sees this same situation occurring with the risks of fracking.
Kasperson is research professor and distinguished scientist at the George Perkins Marsh Institute at Clark University; his research interests have included the vulnerabilities of people, places, and ecosystems and ways to reduce these vulnerabilities and build resilience. He spoke about the issue of social trust and risk. He identified some issues that need further exploration, as there is no literature yet on social trust and risk
in relation to gas exploration. He said that people in government and industry often tell him that trust will not be a major issue because their organizations have good relations with the public. However, he pointed to trends showing a huge decline in public trust in the federal government over the past 50 years and said that the trends are similar for state governments and for corporations. He said that the current, very rapid development of natural gas and the related transformation of our energy system will require a high level of social trust at a time when social trust has sunk very low.
Kasperson said that people now are being socialized into very low levels of social trust. Two decades ago, he said, someone who proposed a new technology would probably get an immediate favorable reception, but now innovators cannot count on trust and must expect suspicion. He cited research by Paul Slovic (1987) indicating that events that might be expected to increase social trust have smaller effects on trust than events that might be expected to decrease it. This evidence indicates that once trust is lost, it is very difficult to regain, Kasperson said, adding that now we need to proceed from high levels of social distrust, which is a new situation in American history.
He noted that concerns about shale gas development occur at scales from local to global and that issues of trust arise at all these levels. He emphasized that uncertainty is dangerous for social trust: the greater the uncertainties, the greater social trust needs to be. He said that the many uncertainties presented by shale gas development create the problem of managing the risks and uncertainties under conditions of low social trust.
Kasperson concluded with several observations. He noted that where those bearing risks lack trust in those making decisions, they demand a greater role in decision making. He said that loss of trust is systemic in the United States. Shale gas, Kasperson said, is like low-level radioactive waste in presenting a very difficult combination of a highly dreaded hazard, large uncertainties, and low social trust. These are conditions that create unusually difficult management and regulatory challenges and call for different kinds of governance processes. He said that the usual methods of risk assessment and command-and-control regulation tend not to work very well when social trust is low.
Questions and comments from participants following the above presentations are summarized here under headings of Krupnick’s presentation, social trust issues, climate change issues, and prospects for solutions.
Krupnick’s presentation. One participant suggested that Krupnick’s idea of cumulative risk assessment leaves out the social part, including community impacts and environmental justice issues, and thought that a different term might be used. He also took issue with the word “accidents,” saying that these are incidents and are preventable through stronger safety culture, Krupnick agreed.
In response to a question about Krupnick’s surprise that surface water issues were so prominent among experts’ concerns, Krupnick replied that he would have expected more concern with ground water than surface water, and possibly more concern with seismicity. He noted that the respondents who described themselves as having the greatest expertise focused particularly on surface water issues related to casing and cementing.
Another participant suggested that expert consensus on the top priority risk pathways, though promising for reaching agreements, may not indicate that those are the most important areas for regulation. He also asked whether agreement on the most important risk pathways implies agreement on how to reduce those risks. Krupnick agreed with the first comment, but noted that each respondent was asked to identify the risks he or she thought most important to address. He said his study had gathered little information on experts’ views about how to reduce the risks, though it did ask whether respondents thought industry or government should take primary responsibility. For the consensus risks, industry agreed with others that government should take primary responsibility; for the nonconsensus risks, industry respondents were more in favor of industry taking primary responsibility.
Social trust issues. A participant asked whether distrust applies to all levels of government and particularly, whether local governments still have the public’s trust. Kasperson replied that this is highly variable from place to place. Where there are good personal relationships with local officials, levels of trust are sometimes high, but in some instances, trust in federal agencies is stronger than trust in local governments. With energy facilities, both the risks and the levels of social trust are very site-specific.
Another participant commented that in the communities in the Marcellus shale region that she has studied, people seek local action because of mistrust of higher levels. She also noted the increasing incidence of litigation, which does not resolve the mistrust question. Kasperson noted that litigation is particularly typical of the United States and is usually polarizing. With high mistrust, he said, willingness to rely on political systems gets weaker and, in the United States, that often translates into court action in the form of liability cases. Perrow added that people rely
on the courts because they have lost trust, and they hope that a court ruling in their favor will restore trust.
Climate change issues. Perrow expressed pessimism because shale gas provides additional cheap fossil fuels, and 80 percent of fossil fuel resources have to be left in the ground to avoid surpassing average global warming of 2°C by 2050. He sees this industry as speeding the world along the path to global warming by cheapening energy and enabling China, India, and other countries to use more of it. Krupnick said that global warming is being conflated with the risk issues. He said that cheap energy is a good thing: what is not good is the failure to internalize the resulting damage in energy prices. If these damaging consequences are internalized, energy will not be as cheap, but the resource will be good for social welfare. He argued that putting a price on carbon would increase the prices of the highest-carbon energy sources, which would boost renewable energy forms cost-efficiently, without the government “picking winners.” The lower social cost of a cleaner fuel will be reflected in fuel cost savings when and if the government implements a carbon tax or other greenhouse gas reduction policy. He said that blaming shale gas for this mismatch of social costs and market price is conflating too much, though there are overlaps.
Prospects for solutions. Susan Tierney asked if anything in the literature or experience can offer guidance for problems like those raised by this set of presenters. Krupnick responded that any industrial activity has this level of complexity and that we have coped with it in the past, adding that this sub-industry is not yet mature, the technology is improving all the time, and many industry actors show good will. He said that we have worked through these problems with the pulp and paper and the chemical industries, and we can work through them in this industry as well, though it has growing up to do. Perrow said he was not hopeful, since this industry has a payoff structure that is worse than that of the chemical industry in terms of the distribution of costs and benefits, and it would be harder to change.
At the end of the previous session, workshop chair Mitchell Small asked the participants to think about the risks they consider to be of greatest significance and to identify any of these that have not yet been discussed. For the risks they see as being most in need of further analysis, he asked the participants to consider (a) the state of scientific understanding of the risk, (b) the availability of methods and procedures to address
the risks, and (c) high-priority needs for further data and studies. He distributed a table with a list of risks, organized under the topics of the workshop presentations, and invited participants to present their thoughts during the final discussion. The summary below organizes the participants’ comments under headings of mishaps in gas extraction; human health; society-level economic and political risks; ecological effects; waste treatment; management of produced water and water withdrawals; routine air emissions and methane leakage; well design, construction, and quality control; unequal distribution of costs and benefits and community risk issues; and methods of risk analysis.
Mishaps in gas extraction. Small suggested that scientific understanding of these risks is at a medium level; there is a fairly good understanding of what to monitor and what effective responses would be. The problem, he said, is that this understanding may not be uniformly applied across the industry. Although there are methods to reduce risks, the extent of their use is unclear. He identified two issues needing further study: (1) development of sensors on the equipment and for monitoring nearby air and water conditions, wellbore integrity, and so forth; and (2) assessment of the costs of this equipment and its ability to make adequate measurements. He said there is a need for supervisory control and data acquisition systems, such as are used in the water supply industry for integrated data collection, reporting, and real-time adaptive management. He sees this topic as a high-priority area for research because knowledge and methods are developed well enough to be used soon and to identify high emitters early.
A workshop participant questioned the quality of the data on mishaps, saying that in that sense, the level of scientific understanding is low. Warner North of Northworks, Inc., suggested distinguishing between general understanding of the processes in question and local understanding of conditions at particular sites. Roger Kasperson noted that the problems of response systems involve human behavior as well as technology, and Small agreed that even with a good technical warning system, human response systems may be inadequate—especially with warning systems that have high false-positive rates.
Human health. Lisa McKenzie, Colorado School of Public Health, proposed human exposure and health monitoring as a high priority, saying that scientific understanding is weakened by a lack of data and that methods and procedures to address the risks are not well developed. She said that public health is a major issue that is too often left out of the discussion. Other participants suggested adding stress and trauma due to traffic accident mortality and morbidity to the list of health concerns.
Krupnick suggested that statistical studies of historical data are important for understanding health effects of shale gas development activities and for identifying health issues needing further study. In response to a question about whether there are biomarkers that could provide indicators of extended exposure, a participant said that the U.S. Department of Health and Human Services has conducted biomarker studies, with data available for analysis, and that the Occupational Health and Safety Administration’s monitoring of workers might also be used for comparisons with nonworker populations.
Society-level economic and political risks. Susan Tierney raised a complex of issues around climate change, fuel substitution, and energy use, related to social, political, and economic risks connected to the ways policy on shale gas is made. She said that rapid shale gas development, negative impacts and community dislocation, and debates about development of renewable energy versus shale gas could lead to large amounts of stranded investments and locked-in greenhouse gas emissions. She said that scientific understanding of this complex of issues is low and suggested that better elucidation of the risks can help inform the public in dealing with them. In response to a participant’s question, Small noted that this issue concerns risks to the economy, which need greater understanding and require integration of social and ethical considerations, which affect human behavior. Newell added that there are a number of unanswered questions about how different relative prices will affect fuel substitution and other economic processes and proposed that increased understanding of the complex effects of policy actions on the larger economic system are important for informed choice.
Ecological effects. One participant identified ecological impacts, such as habitat disruption, as needing attention. She rated the state of scientific understanding as medium for some habitats and low for many others, and she saw the availability of methods to address the risks as medium to low, depending on the habitat. This participant proposed that major research needs include understanding thresholds for change, conducting landscape-level analyses, and exploring the ability of restoration to mitigate disruption of natural habitats, partly using controlled experiments. Aida Farag identified air and water toxicity as topics affecting ecological risks and suggested that scientific understanding is at a medium level for salt and low for trace organics. She said that methods are available to address risks, but their economic costs are unclear. She would give this area high priority for research on thresholds of change, first in the laboratory and then in the field, with monitoring related to species-specific impacts and restoration planning. We may know what needs to be done
for mitigation and remediation of ecological impacts, she said, but we don’t know how to do it.
Waste treatment. Perrow said that industry knows how to use returned water, but he sees a serious problem in the lack of knowledge about treatment and disposal of radioactive materials and other waste constituents. This suggests a need for studies of treatment methods. Another participant said that there are technologies available for treatment and that some are in use in the Marcellus shale region, although there is a need to reduce treatment costs.
Management of produced water and water withdrawals. Abbas Firoozabadi of Reservoir Engineering Research Institute and Yale University suggested that reinjection of produced water into deep formations was a promising risk management technique, adding that with very deep injection, seismicity risk is much reduced. There was some discussion of the ability of the tight deeper formations to accept this water. Jean-Pierre Nicot expressed the view that injection of produced water, at least in Pennsylvania, is a regulatory problem, not a geological one.
Nicot raised the issue of water withdrawals and said that much is known about water needs, though in some places companies may have to be required to report their withdrawals. He said there is a need for better understanding of the availability of subsurface brackish water for use in shale gas extraction operations. Regarding water leakage, Nicot said that monitoring is made difficult by the lack of knowledge about the mechanisms of leakage and that this problem applies not only to shale gas. He identified needs for better understanding of the subsurface behavior of the chemical additives used in extraction and for developing tracers for identifying leaks.
Routine air emissions and methane leakage. A participant said there is good understanding of the mechanisms producing emissions, but very poor understanding of what is happening in the field: for example, is the issue one of good versus bad operators, or is there a manufacturer who is making poorly performing products? He suggested that data are needed on a large number of operations. Tierney added that the frequency of leakage in local distribution systems is also poorly understood. Krupnick said there is good scientific understanding of most of the chemical relationships producing emissions, as well as of how to control them; he suggested that reducing them is mainly a governance issue.
On the subject of local air toxics, a participant said that Texas has much more extensive data than other states but there are still questions about what the air toxics are, how they are distributed through the air,
and how current levels compare to those prior to development. Outside Texas, the state of knowledge is poorer. She said that monitoring technologies exist to measure environmental levels, though making them more affordable may require collaboration between communities and industry. Monitoring should be given high priority, to inform people of the health risks to which they may be exposed, she continued, and biomonitoring can be done along with environmental monitoring, when there have been expressed concerns. She said that improved standard procedures on site and improved communication with the community will help address this issue.
Gabrielle Petron disagreed with Krupnick, saying that the level of understanding of ozone emissions is not very good. She said that Utah has been spending $2 million per year for the past 2 years to understand where winter ozone emissions are coming from, while Colorado has been struggling with ozone in the Front Range since 2004 and is not close to fixing the problem, which comes 55 percent from oil and gas. The area has been in nonattainment since 2007 [of EPA recommended ozone levels]. She expressed distrust in the numbers that some people present, based on the experience in Colorado of spending years to understand the issue without success. She said that although ozone levels can be measured very well, different kinds of measurements are needed to know how to mitigate it. Texas is doing well with monitoring, she noted, but many communities elsewhere do not have the support to do this. She referred to a recent report from the Office of the Inspector General on the low quality and limited quantity of data EPA uses to model emissions. She sees an emergency need for states to monitor their air quality better in order to understand exposures.
Petron also commented on the very different views of the state of knowledge from economists, engineers, and people on the ground. Krupnick said his understanding comes from air quality models that are commonly used. Petron responded that those models were developed for urban emissions, but the needed measurements and monitoring are lacking for rural shale gas development. She added that Western Colorado will soon be in nonattainment because of ozone emissions from Utah, which Colorado cannot control. Krupnick agreed with her position that states need to work together, especially if ozone standards are improved.
Well design, construction, and quality control. A participant accepted Nygaard’s claim that the industry knows how to do well design and construction right but said that that doesn’t mean it is getting done. Site-specific data are not available. Although methods exist for mitigating the risk, there is neither enforcement nor documentation, governance is not consistent, and no one knows whether well quality control is being done.
Unequal distribution of costs and benefits and community risk issues. Jacquet said that there is fairly good understanding of this issue, as well as of methods to address it, but there are governance issues. He said that further research is needed to develop frameworks and best practices for lessening the unequal distribution of costs and benefits. Another participant pointed out that communities do not always have the authority to undertake plans that would help them; they also may not have access to the knowledge that could support good planning. He advocated an extension-service capacity to give communities access to the knowledge about the impacts they face. Jacquet agreed that most communities have neither the data nor the planning capacity to use it.
Small said that the nation is in the midst of a large-scale adaptive management process but lacks adequate data collection. He suggested that there should be studies in some communities to assess different processes for cost sharing. Jacquet added that we know how to do such studies. Christopherson said that information is lacking on some kinds of communities that will be affected, particularly suburban and urban communities, including information on their governance capacity. She judged the level of scientific understanding to be low on differences among communities and on the distribution of costs and benefits because of incomplete knowledge of who owns the land and the full costs of development.
Simona Perry offered the judgment that we do not understand community impacts very well and said that long-term research is needed on psychosocial stress at the community level. She noted that the various risks discussed under the headings of community, health, ecological, and water risks are all related to each other and hard to separate. Qualitative ethnographic and historical studies are needed in communities, she said, to understand stressors and inequalities in relation to past experience. Perry said that economic and census data can help with historical analysis and noted the potential of community-based participatory research for addressing inequality of costs and benefits. She believes there is only low to medium understanding of community trust, but there are methods for mitigating the risks, such as using more collaborative community forums and decision processes, as opposed to the usual approach of private decisions by landowners.
Methods of risk analysis. Thomas Webler referred to the Understanding Risk report, which recommended that understanding be developed with the stakeholders (National Research Council, 1996) and contrasted it to the focus in this workshop on what the scientists know. He said the workshop is missing what stakeholders and the public know and understand, and he suggested that the project consider what it takes to
engage with stakeholders in discussion of the risks, as Understanding Risk recommended. He also commented favorably on the presentation from RFF, which emphasized cumulative and synergistic risks and said that in addition to considering particular kinds of risks, their interactions need to be kept in mind.
Small invited presenters and participants to offer summary comments. He reminded all participants that the workshop project is intended not to develop recommendations but to clarify the state of knowledge. Paul Stern commented briefly on the risk governance workshop scheduled for August 15-16, which was designed to address how risk reduction could be made to happen, given what is known about governmental and other approaches to identifying and disseminating best practices.
Tierney reported on various comments from Webcast viewers of the workshop in Pennsylvania who were quite vocal about wanting the shale gas development stopped. North said that he hoped the NRC would remain open to ideas from people living with shale gas development. He endorsed the suggestion that the kind of analysis being discussed at this workshop should lead into an analytic-deliberative process of the kind defined in the Understanding Risk report. He expressed the hope that the governance workshop will distinguish between risks that arise when best practice is not followed and the larger problems of risk planning and coordination, which are where an analytic-deliberative process ought to be applied. Small noted that input from the Webcast viewers can point researchers toward doing “the right science,” as Understanding Risk put it. Finally, Perry referred to a paper she recently published on the use of an analytic-deliberative process in shale gas development (Perry, 2013). She noted that such a process can be difficult because of transparency issues that need to be overcome and the need to engage the industry in the process, as well as people who may be affected.
This section provides the rapporteur’s comments on trends and patterns that emerged during the presentations and discussions. One such pattern indicates that considerably greater analytic attention has been given to some of the risks related to shale gas development than to others. The risk domains that appear to have received the greatest attention—based on this workshop—include effects on water systems, seismicity, and methane leakage from wells. The domains of potentially significant risks that, based on presentations and comments from various participants,
have not been as carefully examined include risks to public health, ecosystems, air quality, human communities, and global climate.
Research questions that were posed by one or more workshop participants are summarized below under the topics of risks to public health, ecological risks, risks to air quality, risks to communities, implications for climate change, risks to water resources, and other risk issues.
Risks to public health. Several workshop participants argued that strong epidemiological studies about the public health effects of shale gas development appear to be lacking. Key research needs identified by one or more participants include studies to: (1) estimate toxicity factors for substances used in shale gas development; (2) measure exposures to toxic substances and other health stressors and describe variability in emissions and exposures; (3) track the health of workers and residents near shale gas operations; (4) understand the effects of chemical mixtures, noise, traffic, and other stresses on health and quality of life; and (5) incorporate stress into individual and community-level health assessments. Several participants emphasized that systematic before-during-after data collection is especially needed on exposures and health outcomes. Region-specific studies were also highlighted by several presenters and participants as particularly important.
Ecological risks. Studies of the ecological effects of shale gas development activities appeared to a number of participants to be at a very early stage of development. Key research needs identified by one or more participants and presenters include (1) assessments of effects of different patterns and degrees of surface disturbance on terrestrial and aquatic species, (2) studies defining mechanisms of species responses (e.g., toxicity, avoidance, reproductive disruption), (3) field studies of thresholds of change for different species and groups of species, and (4) research to model ecological impacts and studies of restoration methods.
Risks to air quality. Research needs identified by one or more workshop participants include studies based on actual measurement of air constituents before, during, and after drilling activities; studies defining emissions signatures from shale gas formations; studies measuring atmospheric fluxes of methane, particularly in urban areas; characterization of silica emissions; measurements of emissions reductions from best management practices; and development of scalable leak detection methods.
Risks to communities. Several participants noted that there have been very few studies of effects of shale gas development on communities. Among key research needs that one or more participants suggested are
studies of: (1) the distribution of new wealth and the effects of different distributions on community processes, (2) the magnitudes of stress, (3) effects on communities indirectly affected (e.g., by materials transport), (4) effects on urban communities experiencing gas development, (5) the long-term development implications of different patterns of shale gas development, (6) effects on pre-existing industries, and (7) the community effects of possible best practices in development.
Implications for climate change. Most analyses of this issue, according to several of the participants, have been narrowly focused on issues of fuel substitution in the electric power sector and methane releases. Key research needs that one or more workshop participants suggested include studies of the effects of low-cost gas on aggregate demand for fossil and renewable energy sources, studies on fuel switching in the industrial and commercial sectors, studies on the emergence of renewable energy supplies, and studies on the pattern of energy use in the economies of developing nations. Analyses of the implications of various energy policies on these relationships were also identified as important.
Risks to water resources. Some participants identified as research needs studies estimating the degree of stray gas contamination of water wells and the long-term effects of shale gas operations on water quality and availability.
Other risk issues. Some workshop participants identified a need to better understand the distribution of risks and benefits from shale gas development and the risks that shale gas development may pose to trust in various social institutions.