TRJ Environmental

Statistical Analysis of Environmental Data

Distributional Fits to Air Quality Data. Ted Johnson has fit Weibull and lognormal distributions to numerous ambient data sets used in assessing population exposures to ozone and carbon monoxide. He developed goodness-of-fit indices that were used to select the best performing distribution with respect to varying regulatory goals. In recent work, he has used sophisticated statistical software to compare the performance of 37 candidate distributions fit to indoor emission rate data selected for the DIME database developed by TRJ Environmental for the American Chemistry Council.

Time Series Analysis. Ted Johnson developed a comprehensive time series model which uses Fourier Analysis and autocorrelation analysis to characterize the cyclical and autoregressive components of air quality data. This model has been used by EPA to estimate missing values in air quality data sets used in exposure assessment.

Identification of Significant Predictor Variables. Ted Johnson has conducted numerous analyses of data using Stepwise Linear Regression (SLR) and similar techniques to identify the factors which most affect the variables of interest. Recent SLR analyses have focused on identifying the variables that best predict (1) VOC and particulate concentrations in various microenvironments, (2) ozone concentrations near roadways, (3) human ventilation (respiration) rates, (4) energy expenditure rates, (5) window status in motor vehicles and residences, (6) air exchange rates, (7) UV-B exposure levels, (8) time spent in specific microenvironments, and (9) number of years spent in current residence.

Extreme Value Analysis. Ted Johnson has used extreme value theory to evaluate the reasonableness of large values observed in air quality data sets and to obtain robust estimates of the largest values likely to occur under hypothetical conditions. He is coauthor (with Dr. M. J. Symons) of the article “Extreme Values of Weibull and Lognormal Distributions Fitted to Air Quality Data.”

Trend Analysis. Ted Johnson managed a variety of statistical analyses in support of the Annual Trends Report prepared by Monitoring and Reports Branch (MRB) of EPA. These analyses included a statistical analysis of geographical patterns in trace metal concentrations which served as a basis for national cancer risk estimates prepared by MRB, an investigation of changes in precursor emissions and meteorology as possible explanations for recent trends in ozone levels in the Northeast, development of a linear model for characterizing regional trends in sulfur dioxide, and an evaluation of the relative precision of quantile and exceedance statistics as indicators of air quality. Mr. Johnson also analyzed the geographical and temporal relationships between particulate concentrations and emissions as part of a study evaluating an unexpected decrease in particulate levels from 1981 to 1982. In addition, he statistically analyzed the relationship between temperature and the Pollutant Standard Index in 12 cities.

Descriptive Statistics. TRJ personnel have calculated and prepared numerous tables listing descriptive statistics relating to data sets obtained from in-house studies and from studies conducted by other researchers. For example, Ted Johnson managed the data analysis efforts supporting the Total Exposure Assessment Methodology (TEAM) study conducted in Baltimore. The TEAM study collected and analyzed indoor and outdoor data for a variety of volatile organic compounds. Data analysis included quantifiable limits, percents measurable, descriptive statistics, inter-pollutant correlations, stepwise regression of pollutant concentration versus selected explanatory variables, and indoor/outdoor ratios.