Development of Models for Predicting the Distribution of Indoor Nitrogen Dioxide Concentrations

Extensive data on residential indoor and outdoor NO2 levels have been collected in a limited number of U.S. locations. To date, researchers have analyzed these data sets individually, but have not analyzed them in the aggregate. Results have not, therefore, been suitable for application in a nationwide exposure assessment. This paper presents an analysis of indoor and outdoor NO2 field measurements from five U.S. metropolitan areas for homes with gas-fueled ranges and discusses potential applications of the results. Using linear regression analysis, the relationship between indoor NO2 and various predictor variables was explored. Results indicated that ambient NO2 levels alone explain an estimated 37 percent of the variability in indoor NO2 levels, that the relationship between indoor and outdoor NO2 concentrations differs significantly from summer to winter months, and that homes with range pilot lights have indoor levels approximately 7 ppb greater than homes without pilot lights. A logistic regression model which predicts the distribution of indoor NO2 levels based on ambient NO2 concentrations was developed. Estimation and testing of the logistic model indicated good model performance. The model is particularly useful for addressing policy-oriented questions that involve the concept of "acceptable" threshold levels for human exposure to NO2.