CBS Newsletter
Summer 1994
pg. 4

Developing a Methodology for Identifying High-Radon Areas

This two-part article is accompanied by further information from a link at the bottom of this page.

The second of two parts

Average concentrations of radon in U.S. homes carry an increased risk of lung cancer estimated to be on the order of one in a thousand (about 13,000 cases/year), a level that is considered excessive by some and acceptable by others. But almost everyone agrees that concentrations ten or a hundred times higher than this exceed acceptable risk levels in the indoor environment, where estimated risks of premature death due to various types of pollutants (other than radon) and accidents are also typically one in a thousand.

About 50,000 to 100,000 homes are estimated to have annual average concentrations exceeding 20 pCi/L (picocuries/liter) in the living space. At this level (approximately 20 times the national average), inhabitants receive annual radiation doses that exceed the occupational standard for underground miners. Fortunately, these higher-than-average-exposure homes are not spread uniformly throughout the national housing stock; they occur most frequently in concentrated areas.

Having a method of determining the distribution of indoor radon concentrations by geographic area would be extremely useful. With it, authorities could focus their monitoring and control efforts in regions containing most of the high-radon houses, identifying and fixing these houses faster than is likely with the current unfocused approach of seeking to monitor every home in the country. The objective of a focused strategy is to target for mitigation the homes with the highest exposures-and therefore the highest added risk-rather than spend considerably more money and effort trying to reduce the national average exposure. This strategy would also permit development of building codes tailored to specific regions and other approaches to constructing radon-resistant buildings.

Even in a small area, indoor concentrations are not uniform, but distributed over a wide range of values. The difference between a typical area and a "high-radon" area is that in the latter, the entire distribution of radon concentrations is shifted higher than in more typical areas. Even areas with average concentrations have a small probability of containing houses with unusually high radon levels, but a high-radon area has a substantially higher chance of having such houses. The radon research group in the Center's Indoor Environment Program has estimated that approximately 90% of the U.S. homes with concentrations of 20 pCi/L or greater are likely to be located in 10% of the area of the United States. Therefore, the probability that a house in one of these parts of the United States will have a concentration 20 pCi/L or greater is 80 times higher than the probability for a house in the rest of the country!

Identifying High-Radon Areas

The most straightforward way to identify high-radon areas is to monitor enough houses throughout the country to determine the local concentration distributions, even in small areas. This has not been done simply because of the high expense (although the cost is nonetheless small compared to the possible cost of today's unfocused radon measurement and control strategies, which could cost tens of billions of dollars). For example, monitoring 10 to 20 houses in each 4,000-person census tract in the U.S. would require a survey of one million homes, a daunting prospect if we are to follow proper monitoring protocols in a representative sampling of homes.

Another approach is to understand enough about the process of indoor radon entry and removal to create a computer simulation model that can predict indoor concentrations from information about the controlling parameters, such as soil permeability and weather-induced air-pressure changes. In spite of the last decade's advances in transport modeling, it is still not possible to model radon concentration reliably this way, in part because many of the physical parameters needed as input are not known.

Some groups have tried to derive a radon "potential" based on general knowledge about soil and house parameters that affect radon transport and entry. However, these efforts have been generally unsuccessful in predicting actual absolute indoor radon concentrations.

A Statistical Model Approach

A new method, under development by the Center scientists and collaborators at the U.S. Geological Survey, is to use available monitoring data jointly with data on physical parameters, not in a physical model, but in a statistical one. This is an extension of ordinary multiple regression analysis, where the result, in this case the indoor radon concentration, is determined by an array of parameters.

Estimated geometric mean radon concentration by county for Minnesota. Darker shades indicate higher indoor radon levels. Homes in white counties have estimated concentrations below 2.5 pCi/L (picocuries per liter); black counties are greater then 5.5 pCi/L.

Developers of this methodology initially used data from Minnesota, a state with higher-than-average indoor radon concentrations. The average concentrations vary substantially from one county to another, as shown by a sampling of homes in that state. A simple correlation analysis demonstrated that the variation in the county-average radium content of topsoil accounted for most of the difference in the county-average radon concentration, with lesser contributions from other soil characteristics. Although this is the type of behavior that might be expected in principle (since radon is produced by the decay of radium), this correlation emerged with surprising ease. Extensions of this approach to two other states, New York and Washington, also indicate a significant dependence on soil radium content, although the behavior is more complex. There appear to be other parameters that affect county-average radon concentrations in these states.

Work is in progress on extending the statistical approach-applying a geographic information system-to use the data more comprehensively. Also, the radon group is advancing the statistical method to take proper account of the varying (and often small) number of monitoring data that determine the county mean concentrations used in the analysis. For the relatively well-developed case of Minnesota, the group has found that radium content of the soil accounts for more than 70% of the variance in county mean concentrations. The correlation model yields an estimate of the county mean that is as accurate as an estimate based on the monitoring of 20 to 30 homes in the county. Thus, good estimates can be obtained even for the Minnesota counties in which there are only a few (or no) monitored homes.

The challenge now is to extend this approach to areas much smaller than counties, (e.g., to the census tract level). To accomplish this, the radon group is cooperating with the Minnesota Department of Health, which is conducting a new survey according to a Center/USGS design. Careful choice of the location of participating homes will provide a more robust test of the Center's statistical methodology. The radon group, including faculty from the statistics department of the University of California at Berkeley, believes that further investigation of this approach will provide better local concentration estimates. It hopes to develop techniques that can be applied across the nation to derive reliable estimates of local concentrations so that the high-radon homes can be identified efficiently and rapidly.

—Anthony Nero

Info icon

Phil Price
Ashok Gadgil

Ashok Gadgil's email address

Indoor Environment Program
(510) 486- 6591; (510) 486-6658 fax

Accomplishments

A summary of key scientific results from experiments using the basement structures.


Back to Part 1 of Radon article
EETD Newsletter Home Page
CBS Newsletter Home Page
Table of Contents for this Issue