Using "sound science" to shape government regulation is one of the most hotly argued topics in the ongoing debate about regulatory reform. Of course, no one is arguing that the government should rely on unsound science for its decisions. But supposing, as some reform advocates apparently do, that even the best science will sweep away regulatory controversy is equally foolish.
My experience as the chair of a National Research Council (NRC) committee that studied the scientific basis for regulating high-level nuclear waste disposal drove home this conclusion for me. I learned that science alone could resolve few of the key regulatory questions. More often, science could only offer a useful framework and starting point for policy debates. And sometimes, science's most helpful contribution was to admit that it had nothing to say.
A short history of nuclear waste regulation
Both commercial generation of electric power and government production of nuclear weapons result in high-level (long-lasting and highly radioactive) nuclear waste. At present, these wastes are stored at nearly a hundred sites around the United States, but federal policy mandates that the wastes ultimately be placed in a mined underground geologic repository. In 1987, Congress decreed that the first such repository be located at Yucca Mountain, which is near Las Vegas, Nevada.
The basic idea of geologic disposal is to use permanent natural barriers as a principal means of isolating nuclear waste from the environment. Over time, however, some of the radioactive material will escape from even the best repository. At Yucca Mountain, for example, the casks in which nuclear waste will be initially stored will eventually break down, allowing the waste to migrate to the water table, which is located several hundred feet below the repository, and contaminate the flow of groundwater away from the repository site.
This process may take many thousands of years, but the nuclear waste will retain some of its radioactivity for more than a million years. Once the groundwater is contaminated, then the people who use it for drinking and irrigation will be exposed to radionuclides. Given this inevitability, the goal at Yucca Mountain is to design a repository that will limit, over very long periods of time, the human health effects associated with nuclear waste releases to an acceptable level.
Developing a standard that defines this acceptable level is one of Washington's longest running regulatory dramas. After ten years of work, the U.S. Environmental Protection Agency (EPA) first promulgated a standard in 1985. But following a successful court challenge in 1987, the standard was remanded to the agency for revision. Before EPA could issue the new standard, however, Congress enacted the Energy Policy Act of 1992, which mandated a new and different process for setting the standard for the proposed repository at Yucca Mountain.
Congress clearly wanted to curtail the debate over the standard. To do this, it reposed considerable faith in sound science. It required the National Academy of Sciences (through the National Research Council) to evaluate the scientific basis for a Yucca Mountain standard and directed EPA to promulgate a new standard "based on and consistent with" the findings of the academy. At the time, the idea of constraining regulators with the findings of a scientific panel was unfamiliar to the agency and the academy. Since a similar idea is afoot in regulatory reform, the Yucca Mountain experience may be instructive for that debate.
The Yucca Mountain standard
Developing a standard that specifies a socially acceptable limit on the human health effects of nuclear waste releases involves many decisions. As the NRC committee learned in evaluating the scientific basis for the Yucca Mountain standard, a scientifically best decision rarely exists. The trick is to make the best use of the science that is available.
The committee recommended a standard stated in terms of risk of death, in part because future scientific reviews are likely to tighten a standard stated in terms of permissible radiation dose—a situation deemed socially, politically, and administratively undesirable.
The first decision that EPA faces is how to measure safety. This decision entails setting a socially acceptable limit on some aspect of the repository's performance. As a technical matter, for example, the limit could be stated in terms of how much radioactivity the repository releases per year, how much radiation people will be exposed to as a result of releases, or people's risk of dying from this exposure. The committee recommended to EPA a standard stated in terms of risk of death.
The lack of people living in the Yucca Mountain area was one reason for proposing the site as the nation's first geologic repository for high-level nuclear waste. But science cannot predict where people will live thousands of years from now. Policymakers will need to decide what to assume about the distribution of future populations in the area
The evolving scientific understanding of the relationship between radiation doses and the health effects that they cause certainly influenced this recommendation. Over the years, successive scientific reviews typically have concluded that a given dose of radiation may cause more deaths than scientists had previously believed. As a result of this trend in science, it makes sense to state the standard as a limit on the number of additional deaths attributable to releases from the repository. Doing so would mean that the standard would not have to change as the science continues to evolve. This observation also weighed heavily in the committee's preference for a risk-based standard.
Although a scientific fact lies behind it, this recommendation is clearly not dictated by science. Changing a standard to incorporate new information is technically not a problem. The preference for a stable, risk-based standard rests on the belief that changing so controversial a standard as one that specifies the acceptable level of human health effects associated with nuclear releases is socially, politically, and administratively undesirable.
This intersection of science and policy permeates the other decisions that have to be made in setting the standard for determining whether the Yucca Mountain repository would adequately protect human health. In particular, EPA has to specify what level of protection is to be afforded, to whom, and over what time period. For only one of these decisions does science provide reasonably conclusive guidance.
Establishing the level of risk that the standard will allow is a question of policy, not science. In other contexts, however, EPA and other organizations have set limits on a variety of nuclear risks that range from one additional death per hundred thousand persons to one in a million. At best, this information provides a scientifically defensible starting point for debating the acceptable level of risk at Yucca Mountain. It certainly does not predestine the outcome. Acknowledging this reality, the NRC committee could only recommend a reasonable range of risks for EPA to consider in crafting its regulatory proposal.
To determine whether a repository provides the acceptable level of protection, the risk that repository releases could impose on a specific individual or group must be calculated. How this person or group is defined can determine whether the standard is met. It has a particularly significant effect on whether the standard is met at Yucca Mountain, because the geology of the site lends itself to the creation of spots—for example, places in a groundwater plume—at which radiation tends to concentrate. A clever opponent of the repository could define the person to be protected as someone drawing water for drinking and irrigation only from one of these hot spots. An advocate for the repository would naturally assume that the affected parties were located at a safe distance from these areas.
As a matter of policy, the NRC committee preferred to avoid these extreme assumptions. Given this policy, it looked to science (or at least to careful scientific thinking) to contribute a methodology for calculating compliance with the standard that resists extreme cases. The methodology that the committee chose was the "critical group method," which calculates the average risk to a member of the group at greatest risk.
Guidance for the time period over which the standard should provide protection is provided by the fact that radioactivity associated with high-level nuclear waste will not dissipate for more than a million years. Ideally, then, compliance with the standard would be tested over the full duration of this period in order to determine the time at which the greatest effect on human health occurs. Whether this determination is possible depends on the ability of scientists to evaluate the behavior of the repository over very long periods of time.
Here, for a change, is a question of science rather than policy. The committee answered it by saying that compliance assessment is feasible for most physical and geological aspects of repository performance on the order of a million years at Yucca Mountain. Still, this answer is based on the expert scientific judgment that the fundamental geologic structure will be relatively stable for this long, not on the testable hypotheses of scientific method. Thus, other experts might reach a different conclusion.
Running out of science
The NRC committee was able to recommend the foregoing elements of the standard with at least one foot in the realm of science. Unfortunately, however, science can contribute little to answering three of the most controversial questions that bothered Congress about the standard in the first place. For two of these questions, the scientific basis for decisionmaking essentially does not exist.
The absence of a scientific basis for predicting the behavior of humans many years into the future is probably a help in deciding whether we should continue to study the risk of human intrusion upon repositories after they close.
What is a negligible risk?
The main concern of a standard for a nuclear waste repository is to protect populations living near the repository. In principle, however, a very large and dispersed population could be affected by releases of nuclear waste. In the case of Yucca Mountain, radioactive carbon dioxide gas could escape from nuclear waste canisters and be inhaled by people living far away from the repository. The carbon-14 problem, named after the radioactive isotope present in the waste, is one of the most vexing problems with which EPA must deal. Because carbon-14 releases from Yucca Mountain would be mixed with the global atmosphere, the health risk to any one individual is exceedingly small. On the other hand, the number of people exposed worldwide over the life of the repository is astronomical. If we multiply the very small risk by this very large number of people, we can calculate that many additional deaths could occur over a very long time period.
But how do we interpret a number computed in this way? No adverse health effects may occur at the very low doses of carbon-14 to which people would be exposed; but lacking data to show that this would be the case, experts in the field say that the prudent course is to assume that health effects will occur. Making this assumption could produce a scenario that leads either to abandoning the Yucca Mountain site or to spending a great deal of money to contain carbon dioxide gas.
To the dismay of policymakers, science cannot make this problem go away. Faced with this dilemma, the committee could only observe that the risk to any one individual in the global population would be very small—perhaps ten thousand times lower that the one-in-a-million level at which the basic standard might be set. A responsible decisionmaker could conclude that such risks are so negligible that they should not affect the design of the repository, but he or she would have to do so without much definitive guidance from the scientific community.
Can we guard against future human intrusion at a repository?
One way to project significant human exposure to radiation releases from repositories is to assume that someone intrudes after they close. For example, a future oil explorer could drill into a waste canister and bring radioactive material directly to the surface. In crafting its charge to the NRC, Congress specifically asked whether any scientific basis exists for evaluating this risk or for assuming that it can be prevented.
The answer to both questions is no. The committee found no scientific basis for predicting the behavior of humans thousands of years into the future. Since neither the probability of human intrusion nor the effectiveness of preventive measures is predictable, the committee concluded that these issues should not be considered in the assessment of compliance with a risk-based standard. (We did, however, offer an alternative analysis to test the resilience of the repository to an assumed intrusion.)
In this case, the absence of a scientific basis is probably a help to decisionmaking. Admitting the limits of science should greatly reduce the considerable analysis and controversy lavished on speculation about the likelihood of human intrusion. I should note, however, that if regulators were deciding whether to dispose of waste at scattered surface sites instead of in a geologic repository, as at Yucca Mountain, analyzing the risks of human intrusion might be crucial.
What assumptions do we make about exposure scenarios?
In all of the above issues, the committee walked the line between science and policy without dissent. But consensus failed when it came to specifying the exposure scenario to use in calculating compliance with the standard.
The exposure scenario describes how radiation that is released from the repository passes through the biosphere to expose humans. The scenario thus must specify whether and how water wells are drilled into the groundwater underlying Yucca Mountain, whether the water is used for drinking or irrigation, how much of a person's food intake is contaminated by this irrigation, and so on. Science can put bounds on many of these assumptions; for example, people can drink only so much water, and plants retain radionuclides at predictable rates. Developing exposure scenarios, even for the distant future, is therefore not entirely a blue-sky exercise.
Still, science cannot predict human behavior. This consideration is important in the Yucca Mountain case, because the area is sparsely settled—one good reason for locating a repository there. Given this, what should an exposure scenario assume about whether someone is present to be exposed to any release that might occur?
Remember that the committee recommended a standard that would protect the people at greatest risk, while avoiding the trap of extreme assumptions. It would be inconsistent with this principle to base the exposure scenario on, say, the expectation that millions of people will move into the Yucca Mountain neighborhood. A more reasonable assumption is that farmers scattered about the area will comprise the population at greatest risk. Insisting on such a cautious but reasonable approach to narrow the range of assumptions about the distribution of population in the distant future is no small accomplishment. Indeed, doing so would considerably circumscribe the current debate about Yucca Mountain.
Even within this narrowed range of options, however, members of the committee disagreed on the exact population-distribution assumption that should be used. One member felt strongly that the exposure scenario should assume that a subsistence farmer will always be living at the place where exposure to radiation will be highest over the life of the repository. The other members believed that the physical features of the site naturally lead to a dispersed population and that the exposure scenario should take account of this fact.
These alternative views can excite considerable passion on the part of their proponents. In my view, however, such controversy obscures two crucial points. One is that the population-distribution assumption cannot be resolved on the basis of science. No one can predict where people will live in the future; therefore, regulators must make a judgment call in choosing an assumption about population distribution in the exposure scenario. The other point, noted above, is that the debate is over a fairly narrow range of assumptions. Despite the passion attendant on it, this debate is far more manageable than the open-ended debate to which EPA might be exposed if the committee had not narrowed the range of assumptions.
The role of science in regulatory decisions
The lessons that the NRC committee learned in studying the scientific basis for the Yucca Mountain standard may be important to those involved in the regulatory reform debate. The chief lesson is that the soundest science rarely provides black-and-white answers for regulatory decisionmaking; it only brightens a bit the familiar gray space in which decisions are made.
Science cannot protect public officials from hard decisions. Whether the risk from carbon-14 emissions is so small as to be negligible is a tough political call that science cannot—and should not—make.
To be sure, science can sometimes have a conclusive effect on a regulatory decision. In the Yucca Mountain case, the conclusion that the standard should be applied without time limit rests almost entirely on expert scientific judgment. By contrast, the current EPA standard applies only over a 10,000-year duration. Accepting the scientific judgment of the Yucca Mountain study would thus have a profound effect on the design of the standard.
Admitting that science has nothing to say also can powerfully affect decisionmaking. For example, the committee found no scientific basis for evaluating the probability of human intrusion. Therefore, it concluded that the issue should not be considered in assessing compliance with a risk-based standard. If EPA accepts this conclusion, a significant line of argument that could distract the regulatory debate will be closed off.
Mostly, however, the Yucca Mountain study shows that science is helpful, but not conclusive, in arriving at reasonable decisions—such as setting the acceptable level of protection, defining the people to be protected, and specifying the exposure scenarios to be used for compliance analysis. In these instances, the committee avoided asserting that sound science provided a complete answer, but did try to use scientific judgment to define a reasonable starting point and a bounded range of options for EPA to consider. In this way, science can be quite helpful in fostering constructive debate.
Finally, the Yucca Mountain study indicates that science cannot protect public officials from hard decisions. Advocates of the Yucca Mountain repository would like nothing better than for science to make the carbon-14 problem go away. But science cannot do that; it can only note that the risk from carbon-14 emissions to an average individual in the global population is exceedingly small. Whether these risks are so small as to be negligible is a tough political call that science cannot—and should not— make.
In short, the Yucca Mountain study clearly illustrates that excessive faith in the power of sound science is more likely to produce messy frustration than crisp decisions. A better goal for regulatory reform is the sound use of science to clarify and contain the inevitable policy controversy.
Robert W. Fri is president of RFF and recently chaired the National Research Council's Committee on Technical Bases for Yucca Mountain Standards. The views expressed in this article are his own and may not reflect those of the committee or the National Research Council. The complete report, "Technical Bases for Yucca Mountain Standards," is available from the National Academy Press by calling 202-334-3313 or 1-800-624-6242.
A version of this article appeared in print in the June 1995 issue of Resources magazine.
A version of this article appeared in print in the June 1995 issue of Resources magazine.
