DR. WHICKER, thank you for the gracious introduction. On behalf of all of us you have taught and inspired, we express our appreciation. I am quite sure there is no teacher who has influenced so many students in the field of environmental health physics as you have. You influenced my career greatly, and I shall always be indebted to you for your leadership and friendship.
Dr. Boice, members of the Board of Directors, fellow members of the Council, thank you for this honor to be the 37th Lauriston S. Taylor Lecturer. I am truly humbled and appreciative, especially knowing so many of you with whom I have worked over the years and who have been so instrumental in my career. I am very pleased to have many members of my team present with me today. You deserve the credit for this honor as well, since any measure of success I have achieved is because of your steadfast support, diligence, and hard work. Finally, it is a very special pleasure to introduce many members of my family who are with me today, including four of my eight grandchildren.
I had the distinct pleasure of knowing Dr. Laurie Taylor during my early years with the National Council on Radiation Protection and Measurements (NCRP). I shall never forget that each time I came to headquarters here in Bethesda, he always greeted me as if I were a friend and colleague, even though I was very young in my career. The genuine warmth and respect that he afforded to me and to everyone who encountered him is an attitude I always remembered and have endeavored to pass along throughout my career as a scientist.
At the Board’s September meeting in 1979, a new task group was organized to prepare the first NCRP publication on assessing dose from radionuclides released to the environment. I distinctly remember receiving a call from Dick Foster, who was a member of the Board and a much-respected aquatic radioecologist, asking me if I would be a member of the committee and also be willing to co-chair the committee with Bill Templeton.
Fig. 1 is a photo of that task group, which prepared NCRP Report No. 76, Radiological Assessment: Predicting the Transport, Bioaccumulation, and Intake by Man of Radionuclides Released to the Environment (NCRP 1984). The photo was taken at the Grand Canyon in February 1981. It was a unique opportunity to work with some of the greatest contributors to our profession. Some of them, like Frank Gifford, Bill Templeton, Yook Ng, and Dick Foster, have passed away now. Others of us carry on this work. It was truly a great experience for me, especially because I was young in my career. I hope the Council will keep in mind how important it is for us to bring the younger generation into our work so we can pass the torch of knowledge on to them.
There is no question I would not be here today without my service in the United States Navy. The Navy taught me discipline and leadership, and it gave me the fundamental knowledge of my profession. By the time I reached my mid-20s, I had qualified in submarines and stood watch as the Officer of the Deck, giving me the responsibility of controlling the ship under the watchful eye of the Commanding Officer. There are few careers that grant this degree of responsibility to individuals who are so young. I recall vividly being summoned to Tecumseh Court as a new plebe at the Naval Academy on 1 August 1963 to hear President Kennedy say, “Any man who may be asked in this century what he did to make his life worthwhile can respond with a good deal of pride, I served in the United States Navy.” He was right. After serving 4 y on active duty, I continued my naval career as a reservist until retiring in 1999. My Navy career formed the foundation of my profession as a scientist.
In 1977, shortly after I had completed my master’s degree at Georgia Tech, my family moved to a farm in South Carolina where I had spent many summers as a child working with my grandfather and his sons. The move was not in our long-range plans, but it was necessary to keep the farm in our family and to ensure its survival. It was a dairy farm with nearly 400 cows, and I knew little about operating such a large, complex enterprise. Because of the remote area where our farm was located, there was little chance for employment in my scientific profession, so I created a company, Radiological Assessments Corporation,† and began working independently.
The farm recovered financially and is now a productive and state-of-the-art row crop operation where we grow corn and soybeans. The farm taught me perseverance, optimism, and respect for the land. It also gave me insight into one of the most important pathways of exposure from radionuclides released to the environment: the iodine-grass-cow-milk pathway. As I proceed through this article, it becomes increasingly clear how the Navy, the farm, and my profession as a scientist are inextricably linked together.
As a nuclear qualified officer, the Navy taught me to be a generalist. In the maneuvering room on a submarine where the control panels for the reactor, the steam panel, and the electrical panel are located, I was taught, as Engineering Officer of the Watch, to keep my eyes on all critical indicators, not just one. This approach to learning carried over to my scientific career. From the beginning, I wanted to understand something about each of the different components of risk assessment—not just one discipline—and how they were co-dependent.
Environmental risk assessment was evolving during the 1970s, but at that time it was still composed of many disparate sciences without a lot of coordination. In the early 1980s, there was a need to assemble the distinct sciences involved in risk assessment into a single source that could be used as a reference. The resulting book was supported by the U.S. Nuclear Regulatory Commission under contract to Oak Ridge National Laboratory and was published in 1983 (Till and Meyer 1983). It has since been updated by Till and Grogan (2008).
ENVIRONMENTAL RISK ASSESSMENT
On 15 September 2009, I attended a meeting of hydrologists and engineers in Charleston, South Carolina, related to tritium in the environment around commercial nuclear plants. A gentleman sitting next to me asked what I did professionally, and I responded by telling him about environmental risk. He responded by saying, “Oh yes, that fuzzy stuff.” I never had a chance to explain that environmental risk assessment is a science like any other, even though it is built on a number of different disciplines. Therefore, I use this article to explain that environmental risk assessment is not at all fuzzy, and that today it is widely used for a variety of purposes.
Environmental risk assessment, in the context of this paper, is defined as understanding and communicating the movement of radionuclides and chemicals released to the environment, the resulting external and internal exposure to humans, and the resulting dose or risk from exposure. Most of my work has focused on the public exposures that occur beyond the boundary of a facility.
Why do we need environmental risk assessment? There are many reasons we use environmental risk assessment today, and the list gets longer every year. For this discussion, I break down the uses into four fundamental areas.
Environmental risk assessment is used to demonstrate compliance at nuclear facilities so we can make sure authorized releases are below regulatory standards. These standards may be manifested in terms of dose to humans or predicted concentrations to environmental media such as food products. Furthermore, environmental risk assessment can be used to assist facility owners and regulators to identify key pathways of exposure to assist in the design of effective monitoring programs.
A second area that relies heavily on environmental risk assessment is exposure management—that is, understanding the potential health risk from radiation exposure. A recent example of this application was the evacuation of >100,000 residents during the Fukushima Daiichi nuclear accident. Risk assessment will also be relied upon heavily in the future to bring many individuals back into areas around the site when it is considered acceptable for habitation by residents.
We use environmental risk assessment in nuclear facility design to evaluate potential accidents to decide if design features need to be added to those facilities to meet safety criteria. These facilities include nuclear reactors and high- and low-level radioactive waste facilities, as well as small licensed facilities such as hospitals and laboratories handling nuclear materials.
Finally, we use environmental risk assessment to estimate historical and future health effects from exposure to radiation. Examples include past releases of radionuclides from the nuclear weapons complex and fallout from nuclear weapons testing. This is the area where the majority of my work has been focused.
The bottom line is that we use environmental risk assessment throughout our profession, and it is the basis for how we decide to spend billions of dollars each year.
COMPONENTS OF ENVIRONMENTAL RISK ASSESSMENT
I will use the algorithm below to explain the components of environmental risk assessment.
S = source
T = transport
E = exposure
D = dose coefficient
R = risk coefficient
u = uncertainty
v = validation
p = participation of stakeholders
c = communication
Each component is discussed briefly in the sections that follow, along with an example from our work at RAC to illustrate how it is quantified. In the end, we get a better idea about whether environmental risk assessment really is “fuzzy.”
We start with the source. In characterizing the source, the quantity, type, physical and chemical characteristics, and temporal distribution of a material released to the environment must be determined. My example is taken from our historical dose reconstruction of the former Fernald Feed Materials Production Center in southwestern Ohio (Till et al. 2000) that operated between 1951 and 1988. The project was supported by the Centers for Disease Control and Prevention. The goal was to reconstruct doses to the surrounding population/members of the public that could be used to determine if an epidemiological analysis could be justified. The primary source was uranium with lesser amounts of thorium that were released to the atmosphere from >120 different stacks on the site. In addition to uranium and thorium, significant quantities of radon and its decay products were released from two large concrete storage silos (called K-65 silos) containing pitchblende and located near the boundary of the facility.
In the Fernald dose reconstruction, the source was developed by reviewing historical records thoroughly, reconstructing operations at the site, and estimating resulting emissions to the environment. Uncertainties for different sources varied widely depending upon the type of information found to support the methods used to reconstruct the releases. The largest releases of uranium to air occurred in the 1950s and 1960s, and radon releases from the K-65 storage silos remained elevated through most of the 1970s. The predicted total quantities of radon released through 1988 are 5.9 PBq (5th and 95th percentiles of the distribution are 2.9 and 12 PBq, respectively).
Most of the uranium released from the Fernald Feed Materials Production Center from 1951 through 1988 went into the atmosphere, with much smaller amounts being released to surface water (Great Miami River and Paddy’s Run Creek). When uranium is released to the atmosphere in the form of large particles (as was the case at Fernald), it deposits on the ground near the facility and becomes relatively immobile, most of it remaining near the soil surface. Fig. 2 shows the concentration of uranium in soil above background, clearly demonstrating how concentrations in soil are elevated near the facility (Stevenson and Hardy 1993). Therefore, measurements of uranium in soil can be used to provide a qualitative alternative approach (called the “outside in” approach) to verify the magnitude of the more rigorous estimates of the source based on detailed plant records (called the “inside out” approach). Results of the two approaches for estimating the source at Fernald are shown in Fig. 3. The estimated source using the “inside out” approach (releases through dust collectors, scrubbers, and miscellaneous other sources) was a median of 310,000 kg of uranium, with an uncertainty distribution (5–95th percentile) of 270,000–360,000 kg. The median estimate of the source using the “outside in” approach (using soil measurement data) was 284,000 kg of uranium with an uncertainty distribution (5–95th percentile) of 130,000–720,000 kg.
In our work, reconstructing the source for a facility generally occupies a significant portion of the total work. Typically, we find there are always gaps in information needed to estimate developing the amount of material released to the environment, especially in historical dose reconstruction. These gaps, however, can be filled using knowledge about the facility for other time periods. As will be shown later, the source can be confirmed by comparing predicted releases and transport with historical measurement data.
This example of reconstructing the source releases at Fernald clearly illustrates how complex and quantitative the analysis can be. The methodology varies at each facility, but in all cases we have seen, estimating the source is a rigorous and quantitative process that, along with uncertainties, is certainly not fuzzy.
The goal of transport is to use the source to estimate concentrations in environmental media such as air, water, soil, and food products. I will use an example from our work reconstructing doses and risks to the public from the Cerro Grande Fire that burned across portions of the Los Alamos National Laboratory (LANL) in 2000 (Grogan et al. 2006). The fire originated on 4 May 2000 from a controlled burn that escaped containment at the Bandelier National Monument. The fire was not contained until 6 June. Altogether, ∼100 km2 were burned, including ∼30 km2 of the Laboratory. In the town of Los Alamos, 239 homes were destroyed.
Our team was asked by LANL and the New Mexico Environment Department to estimate risks to the public in nearby communities—both risks during the fire and longer term risks following the fire. The analysis involved an investigation of the different pathways of exposure, which included long-term risks from the transport of radionuclides and chemicals present on contaminated land areas where vegetation was destroyed by the fire and thus could lead to significant erosion and enhanced movement of materials to water. The water pathway was of particular interest to several Native American Pueblos in the watershed area, since water purity is of extreme importance to Native Americans.
The study domain (Fig. 4) was ≈3,300 km2 and included the cities of Española and Santa Fe, because many people were evacuated to these areas. There were relatively few measurements of radionuclides or chemicals in the plume during the fire. We were also surprised to learn there were very few previous studies investigating release of chemicals or radionuclides on contaminated land burned by a fire, so we had much less data to work with than we anticipated.
Calculating the transport of radionuclides and chemicals released into the air during burning of the potential release sites first required an understanding of the behavior of the fire itself. During forest fires, combustion products in the form of particulate matter are emitted in large quantities and provide a measurable material that can be used to estimate the concentration of other fire-related products as the smoke plume travels downwind.
The computer model CALMET (Scire et al. 2000a) was used to estimate three-dimensional wind fields across the study area during the fire, and the model CALPUFF (Scire et al. 2000b) was used to estimate advection and dispersion of combustion products released into the air from the fire. The Emissions Production Model (Sandberg et al. 1984) was used to estimate the quantities of combustion products released to the air during the fire based on the type and quantity of fuel loading associated with each burn area. Twenty-four hour average concentrations of particulate matter <10 μm (PM10) were measured in air at a number of locations in the study area before, during, and after the fire. The computer model-estimated concentrations of PM10 were then compared with the measured concentrations and used for model calibration.
In the absence of any data to the contrary, it was assumed that the release and transport of particulate radionuclides and chemicals from LANL sources were proportional to the release and transport of PM10. The dispersion of PM10, therefore, served to track these particulate releases. Volatile chemicals were tracked using carbon monoxide, which is a gaseous forest fire combustion product. Concentrations were calculated for the radionuclides and chemicals identified as important through a screening process. Hourly meteorological data were used in the model to estimate concentrations of PM10 on a grid across the domain to produce concentrations of radionuclides and chemicals in air and to calculate deposition on ground surface. Figs. 5a and b illustrate PM10 concentrations on 6 May 2000 at the early stage of the fire and again on 10 May 2000 as it burned over LANL. Resulting concentrations in air on an hourly basis during the fire allowed us to estimate risks from inhalation and other pathways of exposure to people in the plume’s path.
Although atmospheric dispersion was used as my example in this section, it is important to realize that “transport” of radionuclides refers to movement in all types of media including surface water, groundwater, soil, and food products. Transport models are data intensive and rigorous and, as we shall see later in the paper, have been tested extensively for their validity. Certainly, you will agree estimating transport in risk assessment is not fuzzy and is highly quantitative.
Once we estimate the concentration of radioactive materials in environmental media, we can estimate exposure to people. Exposure can be either external (radiation outside the body that penetrates tissue) or internal (radiation that is inhaled, ingested, or enters the body through some other mechanism, such as a wound) and is characterized by personal or group-specific parameters that allow us to estimate the radiation dose. Such parameters include the amount of time a person stays in a contaminated area, the amount of shielding that protects a person, the type and quantities of food a person eats, or the volume of air a person breathes.
For exposure, I will use an example from our current work on the Atomic Veterans project.‡ The objective of the project is to estimate doses to military veterans who participated in eight series (CASTLE, GREENHOUSE, REDWING, UPSHOT-KNOTHOLE, PLUMBBOB, CROSSROADS, HARDTACK I, and TRINITY, the first nuclear test) of atmospheric testing of nuclear weapons in the United States. The series were conducted at the Pacific Proving Grounds, the Nevada Test Site, and in New Mexico. The dosimetry is being used in an epidemiological study to determine if there is a higher incidence of cancer within this cohort and, if so, is it correlated to dose. Although doses to atomic veterans have been estimated previously, dose estimates were specifically for purposes of compensation and were deliberately biased high in order to assure the veteran the benefit of the doubt in accordance with public law. Therefore, previous epidemiological studies investigating disease among this cohort were limited. Recent efforts by the U.S. Department of Defense to digitize the historical records supporting the veterans’ compensation make it possible to use this wealth of information to calculate doses that are not biased and can be used for epidemiological analysis. Our approach builds upon the available film dosimetry and other measurement data recorded at the time and incorporates detailed scenarios of exposure for each veteran based on their personal, unit, and other historical records. Although film-badge results are available for some, most veterans did not wear badges, and doses must be reconstructed. It is also known that film-badge readings for some tests were not reliable. Our initial goal is to estimate organ doses for approximately 700 cancer cases and 1,200 comparison subjects selected from the study cohort of ~115,000 atomic veterans.
The key to estimating exposure to veterans in our study is knowing where they were located during and after the tests. Fig. 6 is a photograph of Shot BAKER during the CROSSROADS series at the Pacific Proving Grounds. Target ships without crews aboard were anchored near the blast to evaluate the effects. Following tests, support ships would reenter the area and send boarding parties to target ships to facilitate evaluation and decontamination of damaged ships. The water was typically still very contaminated and, as a result, the hull and inboard piping of the ships became heavily contaminated. The most extensive exposure from this contamination for steam propulsion ships was to crewmen who worked in the engineering spaces; therefore, it was important to determine who worked in these spaces.
Fig. 7 is a schematic diagram of a Sumner Class Destroyer, similar to those present during the Pacific testing. The ship has two boiler rooms where steam is produced from freshwater and two engine rooms located just aft of the boilers. These two compartments are separated by watertight bulkheads, and each has a separate hatch for personnel entry. The engine rooms contain a significant amount of piping where seawater is taken in and discharged. The water is used primarily for condensing steam discharged from the turbines and also for evaporating seawater to make freshwater used throughout the ship. The boiler rooms used the freshwater to make steam, and there are no primary sources of seawater in the boiler room.
The U.S. Navy employs a system of job descriptions among enlisted personnel that explains specifically what each person does while on duty. These specialties are known as ratings, and there are dozens of them. For example, boatswain’s mates spend most of their time above deck and also operate small boats carrying crewmembers ashore; machinist’s mates operate the ship’s propulsion system; electrician’s mates tend to electrical systems; and so forth. Although these ratings have changed over time with naval technology, knowing a sailor’s rate and rating is a good indicator of where they were aboard ship and what they did.
The engine room was generally manned by machinist’s mates, electrician’s mates, and a chief or junior officer. The boiler room was manned by boiler technicians and firemen. Contrary to what we expected, enginemen did not spend much time in the engine room but performed their duties in other engineering spaces aboard ship and thus received less exposure. This knowledge about where specific ratings were located aboard ship was very important ultimately in assigning appropriate values for time spent in the engine room, boiler room, and other spaces aboard ship. Table 1 shows several ratings and our assumed breakdown in the crew’s time aboard ship when the ratings have been taken into account.
In this example, having knowledge about where individual veterans worked aboard ship resulted in significant differences between resulting dose and risk. Estimating exposure is often tedious work, but the effort can gain credibility in risk assessment, especially when we work with the people exposed to collect information. As with the source term and the transport elements of the risk assessment equation, estimating exposure is not fuzzy at all.
Dose and risk coefficients
Once exposure is determined, we estimate dose and risk using coefficients that have evolved over many years. The science of determining dose and risk began as soon as scientists began to recognize the potential health effects of exposure to radiation, making it one of the best studied and documented areas of science related to understanding health effects of exposure of humans to any type of environmental contaminant. As a result, we have a vast resource of compilations of dose and risk coefficients that can be readily employed in risk assessment.
The disciplines contributing to dose and risk coefficients cannot be addressed adequately in this article and are mentioned only briefly. Nevertheless, their importance to risk assessment cannot be overemphasized. It is clear that without the extraordinary contributions of scientists and organizations who have focused on dose and risk coefficients over the years, environmental risk assessment would not be where it is today. As a result of their efforts, today we have databases of dose coefficients for internal exposures and of dose-rate coefficients for external exposures for hundreds of radionuclides, especially the radionuclides known to be of radiological importance. These databases are based on an extensive library of biokinetic models that tell us where radionuclides concentrate in the body and the dose to each organ. Risk coefficients are based on epidemiological studies investigating radiogenic diseases in specific groups of individuals. These include the Japanese atomic bomb survivors (Life Span Study cohort), patients exposed to radiation for treatment or diagnosis, workers exposed to radiation in the workplace, and members of the public exposed to radioactive material released into the environment.
Many organizations have contributed to this knowledge and to synthesizing dose and risk coefficients into highly accessible and usable information. The International Commission on Radiological Protection (ICRP) continues to build on its publications of internal and external dose coefficients (ICRP 1992, 1994, 1995a, 1995b, 1995c, 2002a, 2002b, 2004). The United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR 2010) and the National Research Council in the United States (NA/NRC 2006) provide information on risk coefficients as our analysis of radiation exposure continues to grow.
The next four components of the risk equation – uncertainty, validation, participation, and communication – are common to the first five primary elements and apply to each of them individually and as a whole. No area of risk assessment has expanded and progressed over the past two decades as rapidly as uncertainty. This expansion has resulted because the science of uncertainty analysis has been given considerable attention over the past two decades, but also because the technology used to implement uncertainty analysis has improved profoundly. In risk assessment today, estimating uncertainties is an integral part of the calculation. Uncertainty is not only introduced by a lack of knowledge but also by natural or inter-individual variability. This uncertainty can be quantified and used successfully in decision-making and in epidemiology.
I will use an example here from our historical dose reconstruction work of the former Rocky Flats Plant near Denver, Colorado. At Rocky Flats, components were made for nuclear weapons, and the primary source of concern was 239Pu. Our goal was to reconstruct routine and accidental releases of plutonium to the environment and to estimate risk to the public. Estimates of risk to hypothetical scenarios of exposure were to be bounded with uncertainty.
Although uncertainty analysis at the time typically accounted for the source and transport, little work had been completed to estimate the uncertainty associated with risk from exposure of inhaled plutonium. As described in Grogan et al. (2001), an approach was developed using four key sources of information about plutonium risk to develop a distribution for each component, to weight the different components, and to combine them into a single distribution.
The risk per unit dose to the four primary cancer sites for plutonium inhalation exposure (lung, liver, bone, and bone marrow) was calculated by combining the risk estimates derived from four independent approaches. Each approach represented a fundamentally different source of data from which plutonium risk estimates could be derived. These were:
- epidemiologic studies of workers exposed to plutonium;
- epidemiologic studies of persons exposed to low linear-energy transfer radiation combined with a factor for the relative biological effectiveness of plutonium appropriate for each cancer site of concern;
- epidemiologic studies of persons exposed to alpha-emitting radionuclides other than plutonium; and
- controlled studies of animals exposed to plutonium and other alpha-emitting radionuclides and extrapolated to humans.
This procedure yielded the following organ-specific estimates of the distribution of mortality risk per unit dose from exposure to plutonium expressed as the median estimate with the 5th to 95th percentiles of the distribution in parentheses: lung, 0.13 Gy−1 (0.022–0.53 Gy−1); liver, 0.057 Gy−1 (0.011–0.47 Gy−1); bone, 0.0013 Gy−1 (0.00006–0.025 Gy−1); and bone marrow (leukemia), 0.013 Gy−1 (0.00061–0.05 Gy−1). Because the different tissues do not receive the same dose following an inhalation exposure, the mortality risk per unit intake of activity via inhalation of a 1 μm activity mean aerodynamic diameter plutonium aerosol was also determined. To do this, inhalation dose coefficients based on the most recent ICRP models at the time and accounting for input parameter uncertainties were combined with the risk coefficients described above.
The intrinsic merit of each approach was scored on a scale of 1–10, with a score of 10 indicating the approach is ideally suited for estimating plutonium risk in humans. These scores were used to calculate the weighting factor for each independent risk coefficient distribution. An overall risk coefficient distribution for a given cancer site was generated by randomly sampling the independent distributions with a frequency determined by its weight (Fig. 8).
The results showing the range of uncertainty for risk to each organ investigated are shown in Table 2. Since Grogan et al. (2001) was published, significant new data have been collected on plutonium exposure (Jacob et al. 2005; Suslova et al. 2012; Gilbert et al. 2013), and our understanding of uncertainty in risk coefficients (NCRP 1997) and the methods we use to estimate uncertainty in risk assessment have also progressed. It could be argued that if risk assessment as explained up to this point appears to be fuzzy to skeptics, uncertainty quantifies this fuzziness and fully defines the range of possible risks to those exposed.
The next subscript in the risk equation is for validation. Validation is an analysis that attempts to delineate the domain of applicability of a model (Caswell 1976; Kirchner and Whicker 1984). The domain of applicability of a model refers to the set of conditions under which a model can be assumed to adequately represent the system of interest. The validation process determines the accuracy of the model and the range of input factors over which the model provides accurate estimates (Grogan 2008).
My example of validation is taken from work our team performed reconstructing historical doses to the public who lived in or around a small town in southwestern Colorado called Uravan (Rood et al. 2008). Uravan had a long history of mining, first as a radium mine that began operating in 1912, followed by United States Vanadium Corporation, which began operating in 1936. With the beginning of the atomic age in the early 1940s, the production shifted to uranium that was abundant in the ore. Hence, the name of the town: uranium plus vanadium equals “Uravan.”
Fig. 9 is a photograph of Uravan and the milling facility circa the 1940s. The operation and town were located in a deep valley along the San Miguel River. Our goal was to reconstruct historical doses over the period from 1936–1984. The atmospheric dispersion modeling was particularly difficult, and we had to use a dispersion model that could account for the complex terrain in the area. Although we lacked meteorological data for the entire period of interest, we did have data between 1981 and 1986. We also had measurement data of 238U in air between 1981 and 1986 (noted in Fig. 9 as S-7 F-block sampler) and measured uranium in soil at several soil depths (noted as soil samples in Fig. 9). Fig. 10a and b show comparisons between the measured and predicted concentrations for air and soil. The agreement was excellent, giving us confidence that our source term and transport models were reliable.
The process of validating models used in risk assessment does not introduce fuzziness. Rather, as in the case of uncertainty, validation is a quantitative process that makes risk assessment more transparent.
Participation of stakeholders
Since the public is the target of the risk assessment in most of my work, I have learned that the more stakeholders can be involved in the work, the more credibility the work receives. To illustrate the importance of stakeholder participation, I use work we completed at the Rocky Flats Site to derive cleanup levels [also called radionuclide soil action levels (RSALs)] for plutonium in soil (Till and Meyer 2001). This work was supported by the Rocky Flats Citizens’ Advisory Board. The Board wanted to work with scientists to understand how the cleanup levels were derived and be sure that as citizens they supported the proposed levels, since the property was to be turned into a wildlife refuge accessible to the public. The U.S. Department of Energy (DOE) was responsible for the cleanup in coordination with the Colorado Department of Public Health and Environment.
Fig. 11 shows plutonium contamination in soil based on measurement data at the Rocky Flats Site before the decommissioning and cleanup began. Most areas of the site had plutonium concentrations below 37 Bq kg−1, but in some areas the concentration exceeded 1,000 Bq kg−1. Obviously, the lower the RSAL used as a target for cleanup, the more it costs to make the site acceptable for public use. All parties wanted a reasonable RSAL that would be an acceptable level of risk and also manageable from a cost perspective.
We had to work within several guidelines. The dose constraint not to be exceeded was 0.15 mSv y−1 from all pathways of exposure. The RSAL was to be calculated assuming unrestricted use of the land—even though the long-term plan was to make the land a wildlife refuge. Uncertainties were to be addressed. The technical work would be peer reviewed independently by a group of experts selected by the panel.
We worked with the RSAL panel over a period of 18 mo. During this time, we explained to the stakeholders the models we planned to use and model input parameters, scenarios of exposure, and uncertainty. With regard to uncertainty, the panel understood the low human health risk associated with the dose constraint of 0.15 mSv y−1 and suggested that a 10% probability of exceeding the dose constraint would be reasonable. Another important outcome of our discussions with the panel and peer reviewers was to consider the potential for a fire to burn prairie grass onsite and the impact this pathway could have on the RSAL.
The result of our work is summarized in Fig. 12. This graphic shows the probability of exceeding the dose as a function of the RSAL for three scenarios of exposure: (1) a resident child, (2) a resident rancher, and (3) a resident rancher assuming that a fire burns across the land. The goal was to select a scenario of exposure and ensure that the probability of exceeding the dose constraint resulted in an acceptable RSAL. The panel rejected the fire scenario because although the occurrence of a fire at the site was possible, the likelihood of such an event was small. The panel ultimately recommended a RSAL of 1,300 Bq kg−1.
Decommissioning and cleanup of the Rocky Flats Site has been completed, and this success was due in large part to the cooperation between the RSAL panel, DOE, and the Colorado Department of Health and Environment. This example illustrates clearly how stakeholder participation can make risk assessment more understandable, less fuzzy, and more acceptable in the end.
In the mid-1980s, the public became aware that the United States had released large amounts of 131I to the atmosphere during early operations at the Hanford Nuclear Site in south central Washington State. These emissions caused outrage from many long-term residents in the area because they had not been publicly disclosed previously. There was also concern that exposure to the radioiodine could have caused health effects among the exposed population. The states of Washington and Oregon asked for an independent study to evaluate potential doses to the population. The study was called the Hanford Environmental Dose Reconstruction Project (HEDR).
The project was to be supported by DOE, and technical work was to be performed by Battelle Pacific Northwest Laboratories and directed by an independent technical steering panel (TSP) that was to have complete direction and authority for the work. In 1988, I was asked to serve on the independent panel, and over the next 6 y, this work would transform me greatly as a scientist.
Fig. 13 shows the TSP near the end of our work in 1994. When we first met in 1988, most of us had not crossed paths professionally, we came from across the country, and we each represented a different aspect of risk assessment, essentially covering the disciplines in my risk equation. We did not realize at the beginning how difficult the work would be, and I shall always be grateful to my colleagues on the panel and also to the staff of Battelle Pacific Northwest Laboratories for their support. What we accomplished together, in many ways, helped shape how risk assessment is carried out today.
I will use the HEDR Project for my final comments and to address the last component of the risk equation, communication (Till 1995).
Up to this point in my lecture, I have shown how risk assessment is a quantitative science. I say with confidence that risk assessment is not “fuzzy” at all. So perhaps, if the gentleman at the Charleston conference believed it to be fuzzy, the answer lies in communication. In almost every example discussed previously, the risks to people from exposure to radiation were small. Yet the perception of these risks was huge.
On 17 May 1988, the TSP had its first meeting,§ and I was asked to be the chair. My first job when I came out of the meeting room was to meet with the press. I was confident, I knew my stuff, I lived 2,500 miles away, and I was a dairy farmer. I thought I was prepared.
Then I was asked this question:§ “Do you think a system of having a major contractor on this site, performing this work, and having the oversight committee chosen in effect by DOE is going to hamper your credibility in regard to the outcome, regardless of the outcome?” This was not what I expected. I was not prepared.
During the years that followed in my work with the TSP, I realized how difficult it was to communicate with people who did not trust us. What communication came down to was building credibility and trust from the ground up. I knew at the moment in 1988 when I was asked that question, the TSP had to figure out how to build credibility and trust in the HEDR project. If we did not, we could carry out the best science, spending millions of dollars, and it would have been wasted. Failure to communicate our objectives, our methods, and our results in the HEDR Project was not an option.
So how can we build credibility and trust so we can effectively communicate what we do?
First, we must be transparent. The TSP shared information as we received it and opened our meetings to the public. We stressed that anyone could review our work and provide comments. The transparency we achieved set a new standard for openness, especially for studies related to DOE facilities.
Second, we must maintain the highest standards of science. Peer review of the HEDR Project was extraordinary. The body of work that was produced was peer reviewed by independent scientists, journals where the work was published, and also by the public. Creative new ideas were introduced that advanced the state of the art of risk assessment. The science produced during the HEDR Project has endured even 20 y later.
Third, we must recognize that credibility and trust have to be earned. We must go beyond what is expected of us. Just doing our job does not earn credibility and trust.
The TSP requested and received support from DOE to declassify, for the first time, production records at Hanford. Without these records, estimates of releases of 131I at Hanford could not have been peer reviewed by independent scientists, raising doubts about their accuracy and reliability. We transferred funding of the project from DOE to the Centers for Disease Control and Prevention to assure our independence. We worked with nine tribes in the area, under contract, to help us develop diet and lifestyle information so we could estimate realistic doses to Native Americans in the area. These steps were not expected of us, but they were essential to the ultimate success of the project.
Communicating risk is based on trust and credibility that is built during the risk assessment process. There is no prescription for steps that need to be taken, and every risk assessment I have performed has been unique.
Perhaps if I had had more time to communicate with the gentleman who thought risk assessment was fuzzy, I could have convinced him otherwise. I certainly hope I have convinced you.