Improved radon gas mapping finds nearly 25% of Americans living in highest risk areas

Researchers from Harvard T.H. Chan School of Public Health have assembled a national database with millions of multi-day indoor radon measurements from 2001 to 2021. Findings reveal that nearly 25% of the U.S. population may be exposed to radon concentrations exceeding 148 Bq/m³, a level associated with cancer risks.

Radioactive radon gas emerges from natural radioactive decay processes in underground uranium. Uranium-238 decays and eventually generates radon gas, which can migrate upwards, working its way into buildings.

Radon gas pollution is the second leading cause of lung cancer in the United States, claiming an estimated 21,000 lives per year (220,000 globally), with exposure linked to multiple other risks such as breast cancer, stroke and stomach cancer.

The US Environmental Protection Agency (EPA) recommends installing a radon mitigation system when the concentration in living space reaches or exceeds 148 Bq/ m³. Efforts to map radon levels to identify where communities are at most risk and require interventions have historically provided county-wide averages based on very little data.

Public health agencies have been recommending improvements in radon mapping because population growth, climate change, housing construction practices, and mitigation efforts have shifted the pattern of radon risk over time.

Regulations adopted in 35 states have mandated radon measurement and disclosure during property transactions, with tens of millions being recorded in the last few decades. The accumulated data provides an opportunity for revisiting the distribution of radon risk around the country.

In the study, “High-resolution national radon maps based on massive indoor measurements in the United States,” published in the Proceedings of the National Academy of Sciences, researchers compiled a national database with millions of radon measurement data points from 2001 to 2021, and used predictive modeling to map radon exposures across communities in the lower 48 states.

Approximately 4.48 million radon measurements, along with 186 radon-related contributing factors, were used to train a random forest algorithm to extrapolate zip-code-based community risk.

The predictive contributing factors associated with the measurements included geological parameters (uranium concentration in bedrock, soil composition), meteorological conditions (temperature, soil moisture, barometric pressure), socioeconomic indicators (median household income), and a key architectural feature regarding the size and age of homes and if the location has a basement.

This extensive set of predictors was essential for capturing the diverse influences on radon levels across different communities and environments as they correlate with exposure.

Validation methods confirmed that the random forest model effectively captured the variations in radon concentrations driven by the 186 contributing factors with a mean absolute error of 22.6 Bq/m³ for zip code-level observations.

Once validated, the trained model was utilized to extrapolate radon concentration estimates for zip-codes lacking any direct measurements by leveraging the known predictor values in those areas.

Average radon concentration across the contiguous United States was estimated at 53.3 Bq/m³, slightly higher than the EPA’s estimate of 48.1 Bq/m³. Regional variations were much more significant. Five radon zones were established based on predicted average concentrations.

Zone 1 (below 37 Bq/m³) includes the entire central valley of California, stretching up to the Pacific Northwest of the state, most of Texas, almost all of Louisiana and Mississippi, southern Arkansas, and the south half of the Eastern Seaboard. Louisiana and Texas had the lowest state averages, at 24.7 Bq/m³ and 29.7 Bq/m³

Zone 2 (37–74 Bq/m³) cuts across the country, keeping just south of Colorado and Kansas, as it crosses Oklahoma, northern Texas, New Mexico and Arizona, circling the Central Valley of California and covering the Pacific Northwest from Oregon to Washington.

Zone 3 (74–111 Bq/m³ dominates most of Colorado’s land area (but not the population centers) along with Utah, northern Nevada, up through Idaho, Montana, Wyoming, then western Pennsylvania, North Carolina, New York, New Jersey, and parts of Maryland

Zone 4 (111–148 Bq/m³) concentrates around some areas of Colorado and a large swath of territory running down from Canada into Wisconsin, parts of Illinois, parts of Iowa, and portions of Ohio.

Zone 5 (above 148 Bq/m³) covers much of South Dakota, North Dakota, Nebraska, parts of Iowa, eastern Pennsylvania and central Ohio. South Dakota had the highest population-weighted average radon concentration at 128.3 Bq/m³, followed by Nebraska (119.2 Bq/m³). Areas around the city of Newark in Ohio had the highest predicted average radon concentration in the nation at 246 Bq/m³.

On average, about 3.7 million people reside in Zone 5 year-round. Much of the average Zone 4 territory becomes Zone 5 during winter, putting the seasonal peak at nearly 10 million. One contributing factor to the seasonality is reduced ventilation while heating homes for winter, trapping and concentrating radon gas in basements and recirculated air.

Even within low-radon zones, certain homes or communities may experience higher radon concentrations depending on several factors that contribute to radon gas contact and emissions.

The primary driver of radon prevalence was uranium content in soil. High uranium geological basements (shales, granites, glacial deposits) correlated with high radon zones.

High-permeability soils, such as sandy or gravelly soils, allow radon gas to migrate upward more easily and into homes. If there is shallow bedrock, rainwater can pull the soluble uranium isotopes down to the bedrock, where it concentrates and then begins the journey up again as radon gas.

Much of the low radon map overlays well with traditional flood zones, in part because basements are rare in these areas, and homes are often built on raised foundations, allowing for additional ventilation.

Colder regions where basements are common tend to have higher radon levels as foundation (and basement wall) imperfections are prime points of entry for the gas. Radon infiltrates and then accumulates in the poorly ventilated basement spaces, often exceeding aboveground levels by a broad margin.

The study found that 83.8 million people, or 26.8% of the population, live in homes where radon concentrations can exceed 148 Bq/m³. Surprisingly, most of these residences are located in otherwise low-radon regions (Zones 1 and 2), highlighting the need for comprehensive testing nationwide.

Zone 2 alone accounts for 33.4 million residents exposed to high radon levels, with another 20 million in Zone 3. These scattered hot spots have a big impact on the number of people affected as they are in higher-density residential areas that have overlapping contributing factors.

Slight overall decreases in radon gas were observed in some previously high-radon zones, possibly reflecting stronger mitigation practices, radon awareness and evolving building codes.

The new model could be used to design community-specific building codes or increase homeowner mitigation efforts to prevent radon entry. It could also be used to assess residential exposure to radon in population health studies aimed at improving understanding of radon’s health effects.

© 2025 Science X Network