Case Studies in Environmental Statistics

This book offers a set of case studies exemplifying the broad range of statis­ tical science used in environmental studies and application. The case studies can be used for graduate courses in environmental statistics, as a resource for courses in statistics using genuine examples to illustrate statistical methodol­ ogy and theory, and for courses in environmental science. Not only are these studies valuable for teaching about an essential cross-disciplinary activity but they can also be used to spur new research along directions exposed in these examples. The studies reported here resulted from a program of research carried on by the National Institute of Statistical Sciences (NISS) during the years 1992- 1996. NISS was created in 1991 as an initiative of the national statistics or­ ganizations, with the mission to renew and focus efforts of statistical science on important cross-disciplinary problems. One of NISS' first projects was a cooperative research effort with the U.S. Environmental Protection Agency (EPA) on problems of great interest to environmental science and regulation, surely one of today's most important cross-disciplinary activities. With the support and encouragement of Gary Foley, Director of the (then) U.S. EPA Atmospheric Research and Exposure Assessment Laboratory, a project and a research team were assembled by NISS that pursued a program which produced a set of results and products from which this book was drawn.

1 Introduction: Problems in Environmental Monitoring and Assessment.- 1 Statistical Methods for Environmental Monitoring and Assessment.- 2 Outline of Case Studies.- 3 Sources of Data and Software.- Acknowledgments.- 2 Modeling Ozone in the Chicago Urban Area.- 1 Introduction.- 1.1 Health and Environmental Effects.- 1.2 Background.- 1.3 Overview of the Case Studies.- 2 Data Sources.- 2.1 Ozone Data.- 2.2 Meteorological Data.- 3 Trend Analysis and Adjustment.- 3.1 Parametric Modeling.- 3.2 Urban Ozone in Chicago.- 3.3 Comparison with Rural Locations.- 4 Trends from Semiparametric Models.- 4.1 Semiparametric Modeling.- 4.2 Application to Chicago Urban Ozone.- 5 Trends in Exceedances.- 5.1 Exceedance Modeling.- 5.2 Modeling Exceedance Probabilities for the Chicago Urban Area.- 5.3 Modeling Excess Ozone over a Threshold.- 5.4 Prediction of Extreme Ozone Levels.- 6 Summary.- Acknowledgments.- References.- 3 Regional and Temporal Models for Ozone Along the Gulf Coast.- 1 Introduction.- 1.1 Scientific Issues.- 1.2 Data and Dimension Reduction.- 2 Diurnal Variation in Ozone.- 2.1 Singular Value Decomposition.- 2.2 Urban Ozone in Houston.- 2.3 Conclusions.- 3 Meteorological Clusters and Ozone.- 3.1 Cluster Analysis.- 3.2 Nonparametric Regression.- 3.3 Urban Ozone in Houston.- 4 Regional Variation in Ozone.- 4.1 Rotated Principal Components.- 4.2 Gulf Coast States.- 5 Summary.- 6 Future Directions.- References.- 4 Design of Air-Quality Monitoring Networks.- 1 Introduction.- 1.1 Environmental Issues.- 1.2 Why Find Spatial Predictions for Ozone.- 1.3 Designs and Data Analysis.- 1.4 Chapter Outline.- 2 Data.- 2.1 Hourly Ozone and Related Daily Summaries.- 2.2 Handling Missing Data.- 2.3 Model Output.- 3 Spatial Models.- 3.1 Random Fields.- 3.2 Spatial Estimates.- 3.3 Design Evaluation.- 4 Thinning a Small Urban Network.- 4.1 Preliminary Results.- 4.2 Designs from Subset Selection.- 4.3 Results.- 5 Adding Rural Stations to Northern Illinois.- 5.1 Space-Filling Designs.- 5.2 Results for Rural Illinois.- 6 Modifying Regional Networks.- 6.1 Results for the Larger Midwest Network.- 7 Scientific Contributions and Discussion.- 7.1 Future Directions.- References.- 5 Estimating Trends in the Atmospheric Deposition of Pollutants.- 1 Introduction.- 2 Monitoring Data.- 2.1 Case Study I.- 2.2 Case Study II.- 2.3 Additional Ongoing Monitoring.- 3 Case Studies.- 3.1 Gamma Model for Trend Estimation.- 3.2 Network Ability to Detect and Quantify Trends.- 4 Future Research.- Acknowledgment.- References.- 6 Airborne Particles and Mortality.- 1 Introduction.- 2 Statistical Studies of Particles and Mortality.- 3 An Example: Data from Birmingham, Alabama.- 3.1 Summary of Available Data.- 3.2 Statistical Modeling Strategy.- 4 Results for Birmingham.- 4.1 Linear Least Squares and Poisson Regression.- 4.2 Nonlinear Effects.- 4.3 Nonparametric Regression.- 5 Comparisons with Other Cities.- 5.1 Seasonal Parametric and Semiparametric Models.- 5.2 Results: Chicago.- 5.3 Results: Salt Lake County.- 5.4 Direct Comparisons Between Chicago and Birmingham.- 6 Conclusions: Accidental Association or Causal Connection.- References.- 7 Categorical Exposure-Response Regression Analysis of Toxicology Experiments.- 1 Introduction.- 1.1 Critical Exposure-Response Information and Modeling Approaches.- 1.2 Issues in Exposure-Response Risk Assessment.- 2 The Tetrachloroethylene Database.- 2.1 Severity Scoring.- 2.2 Censoring.- 3 Statistical Models for Exposure-Response Relationships.- 3.1 Haber’s Law.- 3.2 Homogeneous Logistic Model.- 3.3 Stratified Regression Model.- 3.4 Marginal Modeling Approach.- 3.5 Other Issues.- 4 Computing Software: CatReg.- 5 Application to Tetrachloro ethylene Data.- 6 Conclusions.- 7 Future Directions.- Acknowledgments.- References.- 8 Workshop: Statistical Methods for Combining Environmental Information.- 1 The NISS-USEPA Workshop Series.- 2 Combining Environmental Information.- 3 Combining Environmental Epidemiology Information.- 3.1 Passive Smoking.- 3.2 Nitrogen Dioxide Exposure.- 4 Combining Environmental Assessment Information.- 4.1 A Benthic Index for the Chesapeake Bay.- 4.2 Hazardous Waste Site Characterization.- 4.3 Estimating Snow Water Equivalent.- 5 Combining Environmental Monitoring Data.- 5.1 Combining P-Samples.- 5.2 Combining P- and NP-Samples.- 5.3 Combining NP-Samples.- 5.4 Combining NP-Samples Exhibiting More Than Purposive Structure.- 6 Future Directions.- References.- A Appendix A: FUNFITS, Data Analysis and Statistical Tools for Estimating Functions Douglas Nychka, Perry D. Haaland, Michael A. O’Connell, Stephen Ellner.- 1 Introduction.- 2 What’s So Special About FUNFITS?.- 2.1 An Example.- 3 A Basic Model for Regression.- 4 Thin-Plate Splines: tps.- 4.1 Determining the Smoothing Parameter.- 4.2 Approximate Splines for Large Data Sets.- 4.3 Standard Errors.- 5 Spatial Process Models: krig.- 5.1 Specifying the Covariance Function.- 5.2 Some Examples of Spatial Process Estimates.- Acknowledgments.- References.- B Appendix B: DI, A Design Interface for Constructing and Analyzing Spatial Designs Nancy Saltzman, Douglas Nychka.- 1 Introduction.- 2 An Example.- 3 How DI Works.- 3.1 Network Objects.- 3.2 The Design Editor.- 3.3 User Modifications.- C Appendix C: Workshops Sponsored Through the EPA/NISS Cooperative Agreement.- D Appendix D: Participating Scientists in the Cooperative Agreement.