Abstract - A methodology is being developed for identifying "high-radon" areas by correlating actual indoor levels with local soil, housing, and meteorological data. In preliminary multiple regression analyses using "screening" indoor radon data from Minnesota, radium concentrations from aerial surveys, and information derived from a state soils map, indicate county geometric mean (GM) radon concentrations with an R squared of approximately 0.5. Furthermore, these data have even greater underlying predictive power, considering the substantial variability in GMs arising from the small numbers of homes monitored in most counties. This suggests that most of the variability of actual indoor radon concentrations from one area to another can be predicted quantitatively based on a correlation analysis between suitable indoor monitoring data and physical data on soils and other factors. This contrasts with methods for mapping the radon "potential" that provide indicators of indoor concentrations without quantifying their relationship to actual indoor levels.
Methodologies for Identifying High-Radon Areas: A Brief Review, Anthony V. Nero, Jr., in Indoor Air '93, vol. 4 (Eds. P. Kalliokoski, M. Jantunen, and O. Seppanen), Indoor Air '93 (publisher), Helsinki, 1993, pp. 419-424.
Abstract - Indoor radon concentrations are found to vary substantially among geographical areas, implying that most homes with high concentrations may "cluster" to a significant degree. Programs for finding and fixing homes with high levels may thus be aided by identifying "high-radon" areas. This has usually been accomplished by simply monitoring indoor levels or by using information on physical factors - such as radium content and permeability of soils, building characteristics, and local meteorology - to map the "radon potential," either using physical models or scoring procedures. A third approach is to develop a statistical exposure model, based on correlation of measured indoor concentrations with data on physical factors, providing a mapping of estimated "actual" indoor concentrations. A preliminary regression analysis between county GM indoor radon concentrations from Minnesota and associated physical data yields an R squared exceeding 0.5.
Bayesian Prediction of Mean Indoor Radon Concentrations in Minnesota Counties, P. N. Price, A. V. Nero, A. Gelman, Health Physics, in press 1996.
Abstract - Past efforts to identify areas having higher than average indoor radon concentrations by examining the statistical relationship between local mean concentrations and physical parameters such as the soil radium concentration have been hampered by the noise in local means caused by the small number of homes monitored in some or most areas. In the present paper, indoor radon data from a survey in Minnesota are analyzed in such a way as to minimize the effect of finite sample size within counties, in order to determine the true county-to-county variation of indoor radon concentrations in the state and the extent to which this variation is explained by the variation in surficial radium concentration among counties. The analysis uses hierarchical modeling, in which some parameters of interest (such as county geometric mean (GM) radon concentrations) are assumed to be drawn from a single population, for which the distributional parameters are estimated from the data. Extensions of this technique, known as a random effects regression and mixed effects regression, are used to determine the relationship between predictive variables and indoor radon concentrations; the results are used to refine the predictions of each county's radon levels, resulting in a great decrease in uncertainty. The true county-to-county variation of GM radon levels is found to be substantially less than the county-to-county variation of the observed GMs, much of which is due to the small sample size in each county. The variation in the logarithm of surficial radium content is shown to explain approximately 80% of the variation of the logarithm of GM radon concentration among counties. The influences of housing and measurement factors, such as whether the monitored home has a basement and whether the measurement was made in a basement, are also discussed. This approach offers a self-consistent statistical method for predicting the mean values of indoor radon concentrations or other geographically distributed environmental parameters.
Joint Analysis of Long- and Short-Term Radon Monitoring Data from the Northern U.S., P. N. Price and A. V. Nero, Environment International, in press 1996.
Abstract - We analyze data collected as part of two types of radon survey of U.S. homes---the National Residential Radon Survey (NRRS) and the EPA/State Residential Radon Surveys (SRRS)---to determine the distribution of annual-average living-area radon concentrations for ground-contact homes in the Northern U.S. A statistical model is used to link the short-term SRRS measurement in each home with the home's annual-average living-area radon concentration, although in no case are both a short- and long-term measurement available for the same home. We show that even though an individual short-term winter measurement from the SRRS is a poor predictor of the home's annual-average living-area radon concentration, an aggregation of such measurements can be used, after adjusting for bias, to characterize the distribution of annual-average living-area concentrations as determined by the NRRS. Different types of homes and different regions of the country require different adjustment equations. We present the adjustment equations, and use them to estimate parameters describing annual-average living-area concentration distributions. Model approximations and validation are briefly discussed. The methods presented here could be applied to calibrate other radon data sets.
The Regression Effect as a Cause of the Nonlinear Relationship between Short- and Long-Term Radon Concentration Measurements, P. N. Price, Health Physics 69, 111-114 (1995).
Abstract - The relationship between 4-day charcoal canister radon measurements and year- long alpha-track detector measurements in 983 New Jersey homes has been recently examined by others. The ratio of canister measurement to long-term measurement for the homes in the survey, a common parameter of interest, was found to increase as the canister measurement increased. The examination presented considerable discussion of the variation of the ratios as functions of various parameters. Although we did not examine the raw data used in the study, it appears that many of the results (and perhaps those in other papers) are consistent with a simple model in which both the long-term and rescaled short-term measurements provide measurements with error of the annual-average radon concentration in the home with no nonlinearity or other functional dependence on radon concentration. We provide an example and discussion of this result, which is caused by the widely known but frequently misunderstood phenomenon called "regression toward the mean," or simply the "regression effect". This does not invalidate the work of others; we merely wish to bring attention to the fact that the results in these papers may have a very simple explanation.
Predictions and Maps of County Mean Indoor Radon Concentrations in the Mid- Atlantic States, P. N. Price, submitted to
Abstract - We use measured surface radium content, geologic province information, information on the fraction of homes with basements and with living-area basements, and measurements from the EPA/State Residential Radon Surveys, in a Bayesian mixed effects regression to predict the distributions of short-term winter and annual living-area average radon concentrations by county in the mid-Atlantic states. Predicted county geometric means are subject to standard errors of about 15% to 30% for typical counties, with the uncertainty in a given county depending on the number of radon measurements in the county and the amount of information about the geologic province that contains the county. After controlling for soil radium concentration and the effect of measuring in a basement versus the first floor, typical geologic provinces are found to be associated with elevation or depression of indoor radon concentrations by about 30\% on average, with some provinces having effects that are considerably more extreme.
Interpreting Maps in Small Area Estimation, Andrew Gelman and Phillip N. Price, submitted to
Abstract - Maps are frequently used to display spatial distributions of parameters of interest, such as cancer rates or average pollutant concentrations by county. Plotting observed rates can have serious drawbacks when sample sizes vary by area, since very high (and low) observed rates are found disproportionately in poorly-sampled areas. Unfortunately, adjusting the observed rates to account for the effects of small-sample noise can introduce an opposite effect, in which the highest adjusted rates tend to be found disproportionately in well-sampled areas. In either case, the maps can be difficult to interpret because the display of spatial variation in the underlying parameters of interest is confounded with spatial variation in sample sizes. We illustrate the results with normal and Poisson models and propose the solution of displaying several maps indicating multiple draws from the posterior distribution of the parameters of interest. We derive results based on simple models with no age- adjustment and no covariates, but the basic findings hold generally.
Meteorological Database for the United States, M.G. Apte, A. V. Nero, and K.L. Revzan, Indoor Air 8, 61-67(1998)
Abstract - A meteorological database has been developed to aid in the prediction of indoor radon concentrations in the United States. The database contains predicted typical monthly meteorological statistics at the county level derived from hourly meteorological data from 208 (234 for precipitation) geographically distinct monitoring stations. Interpolation and extrapolation techniques were used to predict statistics for counties not containing a meteorological monitoring site. The LBNL database includes statistics for meteorological variables including dry-bulb temperature, dew-point temperature, arometric pressure, wind speed, wind direction, hours of precipitation, precipitation, and derived infiltration degree- days. The database consists of individual files of derived statistics for each weather variable and is potentially useful for indoor radon modeling as well as for other purposes. Each file contains data values for all 12 months and an aggregation of the 12 months up to a yearly statistic for all county centroids. A test was conducted to assess the quality of interpolated values. Examples showing the use of the database for mapping infiltration degree-days and an application of the database to a statistical correlation analysis attempting to find meteorological factors influencing indoor radon levels in the United States is discussed.
History of Environmental Restoration at LBL
Predicting New Hampshire Indoor Radon Concentrations from Geologic Information and Other Covariates,
Abstract - Generalized geologic province information and data on house construction were used to predict indoor radon concentrations in New Hampshire (NH). A mixed-effects regression model was used to predict the geometric mean (GM) short-term radon concentrations in 259 NH towns. Bayesian methods were used to avoid over-fitting and to minimize the effects of small sample variation within towns. Data from a random survey of short-term radon measurements, individual residence building characteristics, along with geologic unit information, and average surface radium concentration by town, were variables used in the model. Predicted town GM short-term indoor radon concentrations for detached houses with usable basements range from 34 Bq/m3 (1 pCi/l) to 558 Bq/m3 (15 pCi/l), with uncertainties of about 30%. A geologic province consisting of glacial deposits and marine sediments, was associated with significantly elevated radon levels, after adjustment for radium concentration, and building type. Validation and interpretation of results are discussed.
several other papers are presently in preparation