Environmental regulators would do a better job protecting air quality and public health if they worked with local communities.
On August 22, 1994, the Unocal refinery in Rodeo, California, along the north end of the San Francisco Bay, began to leak a solution of Catacarb through a small hole in a processing unit. While the prevailing winds blew the toxic gas (used by the refinery to separate carbon dioxide from other gases) over the neighboring community of Crockett, Unocal workers were instructed to contain the release by hosing down the unit, but to keep operating. Unaware of the ever-expanding leak, Crockett residents began to experience sore throats, nausea, headaches, dizziness, and other problems, their symptoms worsening over the next two weeks. Unocal finally shut down the unit on September 6, when a neighboring industrial facility, the Wickland Oil Terminal, complained that the refinery’s expanding leak was sickening its workers.
The 16-day release made the shortcomings of ambient air monitoring, which residents of Rodeo, Crockett, and other so-called “fenceline communities” had been complaining about for years, suddenly very visible. If there had been regular air monitoring in the area, or even if residents had had the capacity to test the air once they started to suspect that their symptoms stemmed from chemical exposures, the release likely would have been detected sooner, and the damage to workers’ and community members’ health—which in many cases seems to have been permanent—would have been mitigated.
The release galvanized residents of Crockett and Rodeo, who until that point had largely been complacent about the refinery’s presence. As a result of their activism, two new community-centered monitoring techniques emerged. Residents’ lawyers commissioned an engineering firm to develop an inexpensive, easy-to-use air sampler to give them a way to quantify chemical levels when air quality seemed particularly bad. The device, known as the “bucket,” was subsequently adapted by engineers and organizers with the nonprofit Communities for a Better Environment (CBE) for widespread dissemination, and it is currently used by fenceline communities around the world. Beyond helping neighborhoods closest to refineries know what they’re breathing, the bucket has become a cornerstone of advocacy for more comprehensive air monitoring: users take each bucket sample as an occasion to point out the lack of information being generated in their communities during potentially dangerous releases and to criticize industry and government agencies for their apparent lack of interest in finding out what fenceline communities are breathing.
The second innovation in air monitoring that arose from the Catacarb release, although less well known, has been an important complement to the bucket. In the wake of the accident, Crockett and Rodeo residents successfully demanded that Unocal’s land-use permit not be renewed unless a real-time, “state-of-the-art” air monitoring system was installed at the refinery’s fenceline. Residents were instrumental in designing “the Fenceline,” as the system has come to be known locally, and for nearly two decades it has been held up by bucket users around the country as the “gold standard” for air monitoring that should be required of all petrochemical facilities.
These two new monitoring technologies partially addressed and amplified fenceline communities’ long-standing criticisms of the way environmental regulators measured air quality. Monitoring sites established by state and regional agencies to assess compliance with the Clean Air Act are set up away from large sources like refineries with the aim of getting data that would represent the airshed as a whole. And the little monitoring that was being done by agencies in fenceline communities was conducted only after residents complained about odors, flaring, or other releases from neighboring facilities—and always, residents charged, many hours after the worst pollution had dissipated. Fenceline communities thus were left without information about what they were breathing.
With the invention of the bucket, community members suddenly had the ability to respond to releases themselves, taking 5-minute samples that represented air quality during the worst periods of pollution. These data helped fill the information gap. When regulators and industry criticized bucket data—raising questions about their credibility or arguing that they painted a skewed picture of air quality—community groups took it as an opportunity to highlight the shortcomings of agency monitoring, pointing out that, in most cases, the supposed experts had no data at all. Simultaneously, the development of the Fenceline not only gave Crockett and Rodeo residents continuous information about their air quality in real time; it also offered a concrete example of what regulators could and, according to residents, should be doing to assess air quality and protect communities.
Yet it still took two decades of sustained community activism around air monitoring to push regulators to change their approach. In 2013, the Bay Area Air Quality Management District (BAAQMD) proposed a new refinery rule that would require monitoring at the fencelines of the five northern California oil refineries it regulates (including the Rodeo refinery, now owned by ConocoPhillips spin-off Phillips 66) and in nearby residential areas. Fenceline air monitoring requirements are also a feature of the U.S. Environmental Protection Agency’s (EPA) new refinery rules, adopted in September 2015. (The BAAQMD’s rule was still in the process of getting final approval at the time this article went to press.)
The specifics of the air monitoring required by the two rules are very different: the BAAQMD rule calls for real-time monitoring of a number of chemicals, whereas the EPA rule requires benzene sampling that triggers remedial action if measured concentrations exceed a specified level. Although the very inclusion of monitoring in these rules is a victory for fenceline communities, arguably neither is up to the task of protecting air quality and, ultimately, residents’ health. Looking back to the story of how the Fenceline was set up and how it has evolved points to two important ways that the rules should be strengthened—if not now, then in subsequent iterations: by creating better mechanisms for presenting, interpreting, and using monitoring data, and by including neighboring communities in the design and operational oversight of fenceline monitors.
Issued by Contra Costa County just a few months after the Catacarb release, Unocal’s 1994 renewal of its land-use permit stipulated that the company would design, install, and test “an improved air pollution monitoring system” that would “include infrared or other state-of-the-art remote sensing technology.” But when Unocal brought its design to the county, residents found that “their [Unocal’s] idea of what state-of-the-art was and our idea of what state-of-the-art was were considerably different,” according to Jay Gunkelman, who lived in Crockett at the time of the release. Unocal wanted to use technology already common in the industry: hydrocarbon monitors at a few points around the refinery’s perimeter. Residents—including Gunkelman, a self-described “geek” who at one time developed brainwave biofeedback instruments and now specializes in computer analyses of electroencephalograms (EEGs), and Howard Adams, a Ph.D. chemist who worked in the research department at Chevron for 20 years before moving to Crockett in the mid-1980s—wanted instead to adopt remote sensing devices used by the military to detect chemical weapons in the first Gulf War. The monitors used infrared or ultraviolet (UV) beams of light in conjunction with advanced sensor technology to measure chemical concentrations along a section of the refinery fenceline, not just at a single point, by analyzing the wavelengths absorbed by chemicals along the light beam’s path. The residents’ proposal also included Tunable Diode Lasers (TDLs) to measure hydrogen sulfide and ammonia and, because “open-path” monitors of this sort don’t function well in inclement weather, hydrocarbon monitors similar to the ones in Unocal’s proposed system.
In the disagreement over what counted as “state-of-the-art,” county officials sided with the community, and Unocal was sent back to create a detailed plan that included elements of the community’s proposal. Meanwhile, Crockett residents worked with CBE engineer Julia May to do their own testing of the instruments for which they had advocated. Andy Mechling, who in his job at a camera shop was always the one to set up new equipment when it came in, found that testing out the monitors was right up his alley. Although he had moved away from Crockett with his family when a 1995 tank fire exacerbated chemical sensitivities they had developed in the wake of the Catacarb release, he commuted back to town to operate a borrowed infrared open-path monitor, known as an FTIR, on a neighbor’s roof.
The battle for data access
In the end, residents got the system they wanted. The final Memorandum of Understanding (MOU) that spelled out the monitoring required by Unocal’s permit included FTIRs, comparable UV systems, the TDLs, and organic gas detectors for both the Crockett and Rodeo sides of the refinery. But the residents’ fight for information about what was in the air wasn’t over, because the agreement did not provide for open access to the data. Instead, a video of the monitors’ computer interface was transmitted by modem to a terminal in one resident’s home. That resident, according to the terms of the MOU, could monitor measured chemical levels but was restricted from sharing any of the data until three business days after they were recorded.
Where the Catacarb release catalyzed innovations in community monitoring technology, the Richmond fire spurred a new wave of efforts to make fenceline monitoring standard at refineries.
The arrangement was cooked up to quell refinery official’s fears that data—especially data which had not undergone thorough quality assurance—might panic the public. To resident Ed Tannenbaum, it was technologically backwards: “I saw this screen and I said ‘Well, why isn’t this online?’ This was 2000 by now.” An electronic artist who already had experience building websites, Tannenbaum found a way to capture screenshots from the terminal in his neighbor’s house and put them on the web where anyone could look at the air data. Technically, his website violated the terms of the MOU between the refinery and the community. So the county stepped in: the government was also entitled to the data but not bound by the same rules. They agreed that the data should be made public, and Tannenbaum developed his website—which posted a .gif file every few minutes—under the auspices of the county. As Tannenbaum remembers it, the refinery was finally forced to allow the site to pull actual data directly from the monitors to make the website compliant with disability access law: his screenshots couldn’t be read by assistive technologies for the visually impaired, whereas the numbers could be. Still under pressure from the county, the refinery contracted with Argos Scientific to build a real-time website as part of satisfying requirements for a new land use permit in 2002. When Argos’s design did not meet with community members’ approval, the company helped Tannenbaum incorporate a real-time feed into a site he had designed.
As residents were trying to improve their access to data from the Fenceline, the limitations of the system were becoming clear. The TDLs could detect only large-scale releases of chemicals, making them good for emergency response but not for giving residents information about the odors they periodically experienced. The FTIRs went offline frequently because they had been set up with a path length that was too long, which resulted in too little light reaching the sensor for the system to quantify the chemicals in the air. And the UV systems couldn’t tell the difference between benzene and ozone.
Community members pushed for upgrades to the system, but—as with the initial installation of the Fenceline—the refinery agreed only when it needed permits for new land uses. The UV monitors were replaced in 2002, in conjunction with a new Ultra Low Sulfur Diesel project at the refinery; and when ConocoPhillips applied for a land-use permit for its Clean Fuel Project in 2006, residents used the company’s eagerness to get on with the project to negotiate an updated MOU: the FTIRs would be replaced, the UV systems upgraded, and all of the monitors would have to be online and operational an average of 95% of the time. Under the terms of the new agreement, Conoco also agreed to pay for a portable open-path monitor to be operated by residents and for laboratory analysis of a fixed number of ad hoc, short-term air samples collected at residents’ discretion—creating the possibility of representing air quality in the community itself and not just at the refinery’s fenceline.
The spread of fenceline monitoring
Crockett and Rodeo residents’ struggle to get and maintain a working fenceline monitoring system had ripple effects in other refinery communities. Even early in the history of the Fenceline, Denny Larson, former CBE organizer and founder of Global Community Monitor, a nonprofit that spreads buckets to communities around the world, was portraying the system in Crockett to communities in Texas, Louisiana, and elsewhere as an example of what the refineries next door to them could—and should—be doing. Communities wanting to emulate Crockett’s example gained an important resource and ally when the contract for operating the Crockett-Rodeo Fenceline was taken over by Cerex Environmental Services. Co-owner Don Gamiles, a physicist and entrepreneur with a passion for open-path monitors like the ones deployed in the Fenceline, saw in fenceline communities an important new market: although community groups could not afford to purchase and maintain the sophisticated instruments, organized communities could still create business for Gamiles by compelling companies to agree to monitoring.
With Cerex—and Gamiles’s next company, Argos Scientific—willing to rent monitors to community groups for demonstration projects, provide technical support to communities, and work with refineries compelled to develop monitoring systems that would satisfy residents, other communities began to advocate for, and win, fenceline monitoring. In Chalmette, Louisiana, residents working with the Louisiana Bucket Brigade, an environmental justice nonprofit that offers technical and organizing assistance to communities, independently operated an open-path UV monitor for several months, documenting a violation of the EPA’s 24-hour sulfur dioxide standard in the process. Their efforts not only prompted an enforcement action against nearby Exxon-Mobil, they helped residents win a real-time fenceline monitoring system that the Louisiana Department of Environmental Quality (LDEQ) operated for the next several years. Activists in Port Arthur, Texas, and Benicia, California, also successfully fought for real-time monitoring.
All of these projects were, however, plagued to some extent by the same issues that Crockett residents faced. How the information would be provided was a frequent bone of contention. In Benicia, for example, monitors operated for two years without data ever becoming available to the public because residents and the Valero refinery could not agree on a data-presentation format. A significant issue, which had also arisen in Crockett, was whether monitoring readings should in some way be flagged as dangerous (“red”) or potentially dangerous (“yellow”) when they reached certain concentrations—and, if so, what concentrations would mark the change from safe to unsafe. The sustainability of the monitoring programs was also a problem: Crockett and Rodeo residents had to remain active on the monitoring issue, at times appealing to the county’s authority to withhold the refinery’s land-use permit, to ensure that the Fenceline was kept up to date and in good repair. In Benicia, residents were not able to convince Valero to extend the monitoring program beyond their initial two-year commitment, nor could they persuade the city government to take it over. And the LDEQ-run monitoring program in Chalmette was pared back dramatically after a “final report” concluded that air quality met all relevant standards.
On August 6, 2012, the Bay area saw another major refinery accident. At the Chevron refinery in Richmond, a corroded pipe ruptured, releasing flammable gas which subsequently ignited and sent up a large smoky cloud over the East Bay. Hundreds of residents went to area emergency rooms to treat respiratory problems and vomiting presumably caused by exposure to the large amounts of sulfuric acid and nitrogen dioxide released by the fire. Yet in the immediate aftermath of the accident, residents did not have access to information about the chemicals they were breathing. Contra Costa County officials told reporters and the public that their monitors had detected no hazardous chemicals. No fenceline monitoring was in place at the time of the accident, either: Chevron had been talking for years to residents, City of Richmond officials, and Don Gamiles about establishing a system similar to the one at the Rodeo refinery but, in the absence of a deadline from the city, it had not been completed.
Where the Catacarb release catalyzed innovations in community monitoring technology, the Richmond fire spurred a new wave of efforts to make fenceline monitoring standard at refineries. Chevron moved quickly to establish real-time air monitoring at its perimeter and in the community with the help of Argos Scientific, whose design for the system was informed by the company’s experience with the fenceline at Rodeo. By the spring of 2013, residents could look at real-time data on a website, modeled on the one developed for Benicia but never made public.
The fire also prompted the BAAQMD to make the development of new refinery rules a priority. In March 2013, they released a draft rule that, among other things, requires the five refineries in its jurisdiction to establish fenceline and community air monitoring systems. The accompanying guidelines for monitoring, informed by both public comments and an “expert panel” that included Gunkelman and Larson, clearly reflect lessons learned from residents’ experience with the Fenceline in Crockett and Rodeo. Guidelines include requirements for data completeness—a minimum of 75%—and specify that refineries’ monitoring plans must provide for making monitoring data available to the public in real time, via a website or some similar means.
During the same period, the EPA was revising its own refinery regulations. The rule it put out for comment in May 2014 and adopted in September 2015 also calls for ambient air monitoring at refinery fencelines, but its focus is on controlling “fugitive emissions” from leaky valves and seals. Real-time, open-path monitoring is among the approaches the rule considers, but it concludes that fenceline monitoring with open-path UV systems, although technically feasible, is cost prohibitive. The rule opts instead for passive sampling for just one chemical, benzene, with each sample representing a two-week period. Environmental justice activists, including those associated with Bay area refinery communities, have criticized the proposed requirements for their limited scope and their poor temporal resolution, both clear weaknesses of the strategy relative to the kind of fenceline systems the BAAQMD rule calls for.
But the EPA’s monitoring scheme has an additional feature that highlights weaknesses in BAAQMD’s rule. While the Bay area rule focuses exclusively on generating and providing air quality information, the EPA’s rule lays out a plan for assessing and acting on the air data generated. Data from two-week samples are to be compiled into an annual average, which in turn is compared to a “concentration action level.” If the average benzene concentration at a refinery’s fenceline exceeds nine micrograms per meter cubed for any 26-sample (52-week) period, the refinery must take action to reduce its fugitive emissions.
Despite its other shortcomings, then, the EPA’s rule addresses one problem that has plagued fenceline systems: how the data should be interpreted and presented to the public. Where communities like Crockett and Benicia have struggled with refineries over the levels at which readings should be coded “yellow” or “red,” the EPA’s rule specifies exactly what concentration is considered to be of concern. It remains unclear whether the presence of the Fenceline in Crockett and Rodeo has resulted in any change at the refinery. Although residents maintain that the refinery is more vigilant because it knows it is being watched, there is no evidence that the data are informing refinery practices, or even that refinery officials pay attention to the data. In contrast, the monitoring required by the EPA rule has the potential to trigger corrective action that would measurably reduce the refinery’s fugitive emissions, which, because they occur at or near ground level, are thought to have an especially significant impact on community health.
The lesson from Crockett and Rodeo is as much about the value of community involvement and local learning and innovation as it is about the technical specifications of a monitoring system.
If government agencies are going to succeed in protecting the health and safety of fenceline and other communities vulnerable to airborne toxic chemical releases, rules for air monitoring at refineries and other large point sources will need to incorporate the best features of both the EPA rule and the BAAQMD’s regulation. As in the BAAQMD’s guidelines, air quality measurements should be taken often enough and be available quickly enough to bring to light releases from accidents and inform emergency response. Residents should also have information about the highest levels of pollutants they are being exposed to and the durations over which they are exposed—information not available from long-term sampling strategies. And monitoring should measure as many as possible of the potentially deleterious chemicals released by refineries, not just a single proxy chemical, no matter how strategically chosen. However, for extensive, temporally fine-grained monitoring programs to be effective in reducing communities’ exposure to toxic chemicals, they need to be implemented in the context of a well-specified framework for understanding the results, one that identifies not only what levels may be a cause for concern—as websites for data from Crockett-Rodeo and Richmond monitors now do—but also what levels require action to be taken by facilities responsible for the pollution.
The still-missing masses
Different as they are in their visions for fenceline monitoring, the BAAQMD and EPA regulations share one feature that misses a larger lesson of the Crockett-Rodeo Fenceline: monitoring design and implementation is left up to experts. The EPA rule prescribes in detail the monitoring approach to be used and lays out the site selection and data analysis tasks to be undertaken by technical people at the refinery. Community groups do not figure into the process of siting, sampling, or evaluation (although the rule-making itself, of course, included a public comment period). The BAAQMD’s rule is less prescriptive, outlining basic components of an acceptable monitoring system and requiring each refinery to make a plan and submit it for the agency’s approval. To the agency’s credit, its expectations for monitoring plans have been informed by residents’ experience designing and trouble-shooting the Fenceline in Crockett and Rodeo. In addition, refineries are expected to invite public comment on their monitoring plans—but BAAQMD’s guidelines suggest that community involvement is to happen after refinery experts have come up with a plan.
BAAQMD and the EPA fail to learn from the crucial contributions that residents played in setting up the fenceline air monitoring system in Crockett and Rodeo, and, just as importantly, the contributions that members of the Fenceline’s Community Working Group continue to make in overseeing the system’s operations and understanding its results. Engaged citizens were integral to designing a fast, sensitive monitoring system, at a time when there were no models for fenceline monitoring at refineries. They also catalyzed further improvements to the system by identifying weaknesses in the original design and remedies for them. They created means for data from the monitors to reach the community at large, and they now work with Argos on improvements to the website—most recently by asking the company to show wind direction on a map alongside readings from the monitors. Residents’ on-going engagement with air monitoring issues has even helped improve local emergency managers’ understanding of the effects of chemical exposures on human health: in June 2012, the refinery released hydrogen sulfide from a sour water tank, which stores wastewater during its treatment process. Monitors at Philips 66’s north fenceline and a monitor in a residential area of Crockett registered high levels of the gas, and many residents felt ill as a result of the incident. Yet the measured levels of about 10 parts per million (ppm) hydrogen sulfide did not trigger a shelter-in-place warning that would have indicated to residents that conditions were potentially dangerous to their health; such a warning would have been triggered at 15 ppm. Because fenceline monitoring enabled residents to make a clear link between the 10 ppm level and adverse health effects, they were subsequently able to convince Philips 66 to lower the threshold for a shelter-in-place specified in its MOU, thus improving the health benefits of the local Community Warning System.
Neither the BAAQMD nor EPA rules seems designed to encourage the kinds of opportunities to improve the Fenceline that Crockett and Rodeo residents have had as members of the Fenceline Community Working Group, and that they have made for themselves through their attention to local permitting processes. Community groups are not invited into the design of monitoring systems. Opportunities for public comment, where they exist at all, happen after a monitoring plan has already been sketched out. And public involvement ends once a plan is in place, suggesting that the plan can be implemented in its optimal form by experts with no need to learn from subsequent experience.
The history of community involvement in Crockett and Rodeo shows the limits of that approach. Community members’ experimentation with the system and knowledge of its operations were critical in the design phase, not just to ensuring the acceptability of the system to the community, but also to creating a robust system. And once operational, community involvement helped to improve the system, and to highlight data that could feed back into local government policy to improve outcomes. Given the particular geography and chemical profile of every refinery, the lesson from Crockett and Rodeo is as much about the value of community involvement and local learning and innovation as it is about the technical specifications of a monitoring system.
More effective government regulations would set up the design of fenceline and community monitoring systems as a collaborative process from the start—one in which community members, industry, contractors, and regulators have the opportunity to learn from each other. In the Bay area case, for example, this could be accomplished through monitoring guidelines that require community members to sign onto a monitoring plan before it can be approved. Regulators could further facilitate these collaborations by offering to lend their expertise in both air monitoring and community collaborations; both BAAQMD and the EPA can now point to projects that have been quite successful in the latter regard.
Regulations should also make provisions for ongoing community involvement. Oversight, in particular, should be a collaborative process, since community members have different incentives than refineries to keep monitoring systems up-to-date and operating well; their outside-the-fenceline perspective could also offer insight into how adequately the monitoring system, once it is up and running, is representing local conditions. The most recent MOU between Phillips 66 and Crockett and Rodeo residents builds in opportunities for continued involvement by stipulating that the company will meet with the Community Working Group at least quarterly “to review fenceline monitoring system performance.” Agency rules could specify that a similar agreement should be one component of a company’s monitoring plan.
Finally, although making fenceline data available online was an important step in the development of monitoring systems, more work remains to be done to create interfaces that allow members of the public to explore the data, query it, look for trends, and connect it to what they see, hear, smell, and feel when chemical concentrations spike. As part of creating frameworks for interpreting and acting on data, regulatory agencies should attend to this need for better interfaces, whether that means expanding monitoring guidelines to include issues of accessibility and ease-of-use, or finding ways to integrate fenceline monitoring data into the interactive tools for accessing environmental information that the EPA and other agencies already host. In any case, of course, members of fenceline communities will be important contributors to design teams for the interfaces.
In the realm of fenceline air monitoring, the 1994 Catacarb release was a seminal event. It inspired the buckets, which have become a potent symbol of activists’ calls for better air monitoring in fenceline communities. And, equally important, it led to the behind-the-scenes work that produced a “gold standard” fenceline monitoring system that bucket activists could advocate for—and that at least one California regulatory agency is now using as a model for the fenceline monitoring that it hopes to require of all oil refineries. But just as the fenceline monitoring requirements in the EPA’s proposed refinery rules would benefit from the BAAQMD’s more comprehensive monitoring approach, the BAAQMD rules would be strengthened considerably by the development of predetermined thresholds that trigger action to address toxic releases. And both should make provisions for community participation in all stages of monitoring design, implementation, oversight, improvement, and data interpretation. Not every resident of a fenceline community will have the interest, dedication, or time to delve into the intricacies of ambient air toxics monitoring. But as the history of the toxic releases in Rodeo and Crockett shows, local knowledge and motivation can be powerful sources of technological and policy innovation on behalf of public health and the public good at fencelines that separate everyday life from chemical hazards.
Christine Overdevest and Brian Mayer, “Harnessing the Power of Information through Community Monitoring: Insights from Social Science,” Texas Law Review 86, no. 7 (2008): 1493-1526.
Gwen Ottinger and Rachel Zurer, “New Voices, New Approaches: Drowning in Data,” Issues in Science and Technology 27, no. 3 (2011).
Gwen Ottinger is an assistant professor in the Department of Politics and the Center for Science, Technology, and Society at Drexel University.