7
STATEPRIORITYSETTING
INTRODUCTION
Numerous states have encountered the need to develop ranking systems as an aid to setting priorities for the remediation of abandoned hazardous waste sites. States often have multiple statutes that provide authority for remediation of waste sites that are not covered under federal Superfund. The Environmental Law Institute (ELI, 1989, 1991) found that 24 states had their own priority-setting systems. Based on ELI's survey results, the committee asked a number of states to provide written descriptions of their systems. This chapter discusses the ranking systems of the states that provided descriptions. The approaches considered fall into three categories: systems similar to the EPA Hazard Ranking System (HRS) model; other explicit numeric systems leading to a site-specific score; and systems that categorize sites with the highest priority into three or more groups based on a narrative description of the severity of effects. This chapter examines several of the state-ranking models with respect
to how they compare with each other and how well they achieve general objectives of a ranking system.
This chapter does not describe how the various state-ranking systems help to obtain a final priority for cleanup or the policy context of their application. In general, if observations at a site make it dear that a severe problem exists, a response is triggered even if that site did not receive a high numerical ranking. The documentation reviewed by the committee represented various stages of drafting and reformulation. Although some of the ranking methods considered in this chapter have not been made final, it is useful to examine them as examples of the different ideas on ranking methods that have arisen at various state environmental regulatory agencies.
STATES WITH RANKING SYSTEMS SIMILAR TO THE EPA HAZARD-RANKING SYSTEM
California, Ohio, Oregon, and Washington use indices of contamination severity—such as chemical toxicity, quantity, and mobility in the environment—that closely resemble those used in the EPA hazard ranking system (Federal Register, 1990; California Department of Toxic Substances Control, 1991; Oregon Department of Environmental Quality, 1991; EPA, 1992; Ohio Division of Emergency and Remedial Response, 1992; Washington State Department of Ecology, 1992). Scoring methods were provided by these states containing enough detail to allow a comparison of the scoring element values, routes of exposure, and algorithm structure.
Scoring Elements
For each of the four models, subscores are determined for up to four attributes of the contamination situation. Sites are ranked on a "worst first" basis with larger subscores reflecting greater concern for human health or environmental damage from the site. The following descriptions represent a composite of the four state models. The first attribute in the overall architecture is a subscore, called release, reflecting the strength of the evidence that contaminants are indeed present at the site in places and concentrations with the potential to migrate or cause a problem for direct on-site contact. This conclusion is supported, for example, by measurements of contaminant concentrations or observation of leaking containers.
For the most part, the second attribute—substance characteristics—scores the qualities of an individual contaminant in terms of its toxicity (human and environmental), mobility or water solubility, quantity, degree of persistence, and characterization of containment on site (e.g., landfill, above-ground container, or spill). This score describes the chemical contaminant in isolation from the topography and geology of the site.
The third attribute, migration, scores the same quality of chemical mobility as in substance characteristics, only from the perspective of the characteristics of the environment rather than the chemical. Parameters such as slope of the land, precipitation, and potential for flooding are rated so that greater potential for migration results in a greater score. California, in contrast to the other states, includes an exposure estimate in the migration section. Indices of exposure include distance to nearest structure and sensitive environment. This estimate would give higher scores to sites with close-by structures where the potential for human exposures would be greater.
The fourth attribute, targets, gives all indication of the receptors (human and environmental) that are in the vicinity of the waste site. The more and closer the potential receptors, the higher the score. Factors considered include: number of people living within 1 mile, presence of surface water bodies within 2 miles, population size served by wells, and the distance to the nearest well. The California model includes these sorts of considerations in both the migration and targets categories.
For each of these four attributes, a numeric score is determined by choosing from among site-specific numeric values. For example, migration potential is gauged by quantifying such parameters as amount of precipitation, slope of the land, and soil type. The degree of toxicity of a chemical is scored with parameters such as reference dose (RfD) for noncarcinogens and slope factor for carcinogens. For the most part, states have selected similar ranges of values to describe a particular parameter;, however, significant differences exist in some cases. Table 7-1 shows a comparison of the states' middle values for several parameters pertinent to the groundwater pathway. Also shown for comparison are the corresponding values from the EPA HRS.
The most significant differences among the four states are in the scoring of the substance characteristics attribute. For example, the middle value of the quantity (in tons) of a contaminant is five times as large for Ohio as for California. This means that what would be considered a medium amount of waste in Ohio is considered a large amount in California. There are substantial differences between Oregon's treatment of the chemical toxicity indices for acute and chronic health effects as compared with those of the other three states. The LD50 leading to a medium score for Oregon is almost three times that for the other states. The differences in the RfDs is even more pronounced with Oregon's middle value 100 times higher than that of the other states. Therefore, the relative weight assigned to chemical toxicity compared to mobility and targets is lower in Oregon than in California, Ohio, and Washington.
Another point of comparison identified in Table 7-1, risk equivalent, is an expression of the relative weight given to the toxicity of carcinogens and noncarcinogens within a given ranking scheme. If two sites were to be compared, each having only one contaminant—a carcinogen at one site and a noncarcinogen at the other site—what would be the health risk at each site when both sites delivered the same dose of equally-weighted chemicals to a human receptor? In Oregon, for example, if exposure conditions were such that one were to receive exactly the RfD of a noncarcinogen at one site and the same dose of a carcinogen at another, the lifetime excess health risk from the site with the carcinogen would be 0.12. For California, Ohio, and Washington, the cancer risk at the RfD is 0.003. At a risk of 0.12, there is 40 times more of a given carcinogen than at a risk of 0.003. Therefore, Oregon treats carcinogens less stringently compared to noncarcinogens than the other states because at a specific dose (the RfD) a greater amount of a carcinogen is given the same rank as a noncarcinogen.
The values for the migration parameters are either identical or fairly close among the four states. There is approximately a factor of ten difference in the values of the target parameters between California and Oregon. For a medium score in the Oregon ranking system, more people need to be served by contaminated drinking water wells and more acres of land irrigated. Because the Oregon system has higher middle values for toxicology and target parameters, and approximately the same values for migration, the overall site rank would be relatively more influenced by contaminant migration than for the other states.
Routes of Exposure
To calculate a final score for a site, distinct routes of exposure to humans or other non-human environmental receptors are considered for up to the four site attributes (under consideration).
TABLE 7-1 Comparison of Selected Site-Specific Values Used in Scoring
|
Middle Value of Range |
||||
Parameter |
Calif. |
Ohio |
Ore. |
Wash. |
HRS |
Acute health effects, LD50, mg/kga |
550 |
550 |
1,500 |
550 |
250 |
RfD, mg/kg-daya |
0.00055 |
0.00055 |
0.05 |
0.00055 |
0.027 |
Slope factor, mg/kg-daya |
5.5 |
5.5 |
2.5 |
5.5 |
0.5 |
Risk Equivalentb |
0.003 |
0.003 |
0.12 |
0.003 |
0.008 |
MCL, µg/La |
55 |
55 |
|
55 |
|
Quantity, tonsa |
50 |
260 |
110 |
125 |
|
|
0.5 |
0.5 |
0.5 |
|
|
100 |
100 |
100 |
100 |
50 |
|
Precipitation, monthly, inchesc |
10 |
|
20 |
20 |
15 |
Depth to groundwater, feetc |
100 |
100 |
100 |
100 |
138 |
Hydraulic conductivity, cm/secc |
10-5 |
10-5 |
|
10-5 |
10-5 |
0.5 |
0.5 |
1 |
0.5 |
0.75 |
|
Middle Value of Range |
||||
Parameter |
Calif. |
Ohio |
Ore. |
Wash. |
HRS |
Number of people served by drinking water wellsd |
100 |
|
1,200 |
|
3,000 |
Acres irrigatedd |
100 |
|
1,500 |
|
|
a Substance characteristics. b Does not appear in any ranking scheme, for analysis purposes only. It is an expression of the relative weight given to toxicity of carcinogens and noncarcinogens within a given ranking scheme. c Migration. d Targets. Sources: Data from California Department of Toxic Substances Control, 1991; Ohio Division of Emergency and Remedial Response, 1992; Oregon Department of Environmental Quality, 1991; Washington State Department of Ecology, 1992; Federal Register, 1990. |
The subscores for the attributes are combined to form a score for the route. Not all of the four states have considered the same routes for all of the attributes, but in most cases groundwater, surface water, air, and soil or direct contact have been evaluated. Table 7-2 shows the routes of exposure that have been considered by each state. In contrast to California, the other states consider surface water and air as different routes that depend on whether there is a human or environmental receptor. Ohio does not include a soil pathway or direct contact pathway because its model is not intended to deal with emergency conditions. Washington has a sediment pathway that is not present in the other state models.
TABLE 7-2 Routes of Exposure
California |
Ohio |
Oregon |
Washington |
Groundwater |
Groundwatera |
Groundwatera |
Groundwatera |
Surface water |
Surface watera Surface waterb |
Surface watera Surface waterb |
Surface watera Surface waterb |
Air |
Aira Airb |
Aira Airb |
Aira Airb |
Soil |
|
Direct contacta |
Sedimenta Sedimentb |
a Human. b Environment. Sources: Material from California Department of Toxic Substances Control, 1991; Ohio Division of Emergency and Remedial Response, 1992; Oregon Department of Environmental Quality, 1991; Washington State Department of Ecology, 1992. |
Algorithm Structure
Although there is some similarity among the states in the selection of site-specific values (Table 7-1) and routes of exposure (Table 7-2), their mathematical operations for combining the resulting subscores are quite variable. Table 7-3 shows an example of the different methods used to combine values to arrive at the substance characteristics subscore.
TABLE 7-3 Substance Characteristics (SC) Subscore
California |
Toxicity + Solubility +Waste quantity = SC |
Ohio, Oregon, Washington |
(Toxicity x Containment) + Waste quantity = SC |
Sources: Material from California Department of Toxic Substances Control, 1991; Ohio Division of Emergency and Remedial Response, 1992; Oregon Department of Environmental Quality, 1991; Washington State Department of Ecology, 1992. |
Solubility could be considered roughly equivalent to the extent of containment on site, which, together with toxicity and waste quantity, forms the basis for the substance characteristics score. California's model has only additions among the various attribute scores and routes. Toxicity is added to solubility, whereas the Ohio, Oregon, and Washington models multiply toxicity by containment before the result is added to waste quantity. All of the states ultimately arrive at subscores for all the routes, which are then combined into a total score for the site. California averages each of the four routes. Since no subscore for a route can exceed 100 points, the final score is between 0 and 100. Ohio uses the "root mean 4th power method." This is the fourth root of the mean of the route scores raised to the fourth power. Oregon takes the maximum route score, adds it to the mean of the other five
routes, and adds ten bonus points if the site is in a sensitive environment. Ohio and Oregon have normalization procedures for generating final total scores between 0 and 100. Washington ranks subscores from 1 to 5 and combines ranks so that the highest pathway ranks are given more weight. The human health and environmental pathway ranks are combined in a matrix that weights human health more heavily. The final site score is between 1 and 5.
OTHER NUMERIC RANKING SYSTEMS
The Michigan ranking system provides a subscore for each of the following: environmental contamination, mobility, sensitive environment, population, toxicity, and waste quantity (see Chapter 1). These attributes closely parallel those considered in other state models and the EPA HRS. The method of scoring the attribute, however, is different from the other systems. For the most part, a series of narrative descriptions is associated with each attribute. For example, environmental contamination is scored by choosing from among 31 described conditions each of which has a point, as illustrated in the following example:
One point shall be scored for surface water if a surface water body or wetland is located within 1/2 mile of the site, three points shall be scored for groundwater if a sheen is visible on an exposed groundwater surface, nine points shall be scored for surface water if the department of public health has issued a fish advisory for a water body and the cause of such an advisory can be attributed, in part, to the site (Environmental Response Division, 1990).
The use of variables, such as coefficient of aqueous migration, monthly precipitation, depth to groundwater, and hydraulic conductivity, is not found in this model. The final score is simply the sum of all the subscores.
STATES WITH RANKING BY BROAD CATEGORY
New York, Montana, and Missouri differentiate all sites into one of three categories of priority (Missouri Department of Natural Resources, 1991; Montana Department of Health and Environmental Sciences, 1991; New York State Department of Environmental Conservation, 1992). Each category is described by one to seven characteristics of the site that are predictive of its ultimate capacity to cause harm to human health or the environment. For example, New York, in its Category One (High Priority), requires a determination of "probable release to groundwater which is a drinking water supply." A Category Two site must demonstrate "minimal potential for release to groundwater that is a drinking water source," and Category Three "minimal potential for release to groundwater which is not a drinking water source." Other characteristics of each category include release to air, release to surface water, and effect upon a sensitive environment. Demonstrating that the criteria for any one of up to seven site characteristics per category are met is sufficient to rank the site in that category.
These ranking schemes provide an outline of what end points resulting from the toxicity, targets, and environmental fate of contaminants should be considered in determining the priority of cleanup. They differ from the HRS-like systems in that there is no mathematical combination of factors that lead to a score. Any one of a number of potential effects, if documented, could lead to a maximum score. This approach leaves the analyst much more flexibility in deciding which of the potential effects to pursue in more detail. For New York, this represents a change from a previously used numeric scoring system, but perhaps the state considers the added flexibility to more than compensate for the loss of quantitative information.
DISCUSSION
The states have considerable collective experience with the problems of setting priorities for hazardous-waste site cleanup, and the need for and purpose of their ranking systems are clear. In general, the purpose and objectives of the states' ranking models are stated within the documentation accompanying the model. Often-stated objectives are that the models are intended to be scientifically defensible and easy to use, require minimal and inexpensive data, and provide results to help establish priorities for effective use of funds.
For many of the states considered, there is evidence of very thoughtful development of site ranking models such as parameters for location of fisheries, containment structures, population densities, and sources of drinking water. However, how the relationships between the model parameters were developed and what strategies were useful for combining the parameters is not always clear and often not documented, thus the models tend to lack credibility. There are many different scoring approaches among the states that use essentially the same type of data. It was not within the committee's scope of work to establish the reason for these differences or whether any of them offer a clear advantage over any others.
As was discussed in earlier chapters for the federal agencies' models, similar questions of appropriateness of the logic for combining various scores within the ranking methods apply to the state approaches. The documentation provided to the committee did not indicate the reasons—derived from first principles—for choosing the particular score combination approaches. The committee is unaware of comparisons of any state ranking system with other approaches such as risk assessment for the purpose of validation.
The extent and completeness of documentation varied consider-
ably among the states' models. In some instances, another state was cited as the source of the major elements of an approach. For other states, very detailed data and explanations were provided in support of the range of choices for particular site variables. It appears that all the state models could be used readily by persons who have a minimum of formal scientific training and are provided with the necessary data.
States not using formal ranking models often tend to develop less data-intensive methods that rely on the judgment of professionals in the state agencies to integrate information into site rankings. A derailed state-by-state survey, beyond the scope of this study, would be needed to ascertain the relative utility of each of these somewhat different designs of ranking or rating sites.