The second year of Center activities has constituted a building phase. The Center is now starting to build a reputation in the statistical community. During this year we have made considerable effort to give the work at the Center a national (and international) character. We have had several visitors, and expect that aspect of Center activities to keep growing.

At the Center offices on campus there is a lot of activity. More than a dozen graduate students are involved in Center work. Many of these students have offices in or near Center space. About half a dozen local researchers have (or sometimes share) desk space in Bagley Hall, and all visitors have desk space there. It is cramped, but full of vitality.

We first outline some of our outreach activities during the year, then describe the current research projects, outline some of the administrative and infrastructure details, and sketch an outlook towards the future of the Center.

A variety of activities have served to make the statistical community as well as the campus community aware of Center activities. Here we focus on some of the more tangible outreach activities such as seminars, web page, and workshops.

The Center seminar series continued with speakers including Center members and visiting faculty. The seminar series was organized as an official University course, and had about half a dozen registered students. The attendance was down from the previous year, with most seminars having 20-30 attending.

A complete list of seminars is given in Appendix A.

• to have faculty members organize the seminars with about half the seminars each quarter falling into a coherent theme
• to develop a Center newsletter
• to use seminars as a focal point for Center communication and interaction

• web-based tutorials in basic and advanced statistical methodology
• web-based demonstrations of Center-developed software
• web-based research discussion groups

Co-organizers: Tony Olsen, EPA; Loveday Conquest, U. of Wash./NRCSE

The workshop took place November 21-22, 1997, at the University of Washington Seattle campus. It was based upon the above theme proposed by Joe Sedransk. Additional participants included Phil Larsen (EPA Corvallis), Steve Rathbun (University of Georgia), Hans Schreuder (Rocky Mountain Forest and Range Experiment Station), Denis White (Oregon State University), Nick Chrisman (Univ. of Washington), Mark Kaiser (Iowa State), Jim Karr (Univ. of Washington), Adrian Raftery, Dale Zimmerman (Univ. of Iowa), Gary Oehlert (University of Minnesota), Mark Handcock (Penn State), Abdel El-Shaarawi (National Water Research Institute, Canada). The purpose of the workshop was to clarify issues and set a research agenda regarding combining multiple sources of information in environmental monitoring programs. Topics included the following: integrating probability samples and judgment samples to evaluate the conditions of the nation's aquatic resources; adoption of probability-based designs for combining data across time in water quality monitoring programs; integrating the USFS Forest Inventory and Analysis and NRCS Natural Resources Inventory to enrich knowledge of the nation's natural resources base; combining information across a multi-organization biodiversity research program; defining ecological integrity and measuring biological condition to assess ecological health; tolerance relations as a potential tool for regional monitoring in the absence of probability based samples; Bayesian synthesis methods for deterministic simulation models; combining environmental time series from multiple measurement systems. A host of attendant problems were discussed, including issues of missing data, scientific reasons for merging surveys as well as political burdens of merging; how best to do this; and how to best use concomitant information. Collaborative relationships were established for continuing this kind of research. Participants were also urged to apply for grants as NRCSE visitors.

As a follow-up project, the Center is funding a graduate student at Penn State University to work on improved understanding of stream and river systems in the United States by combining information from separate monitoring surveys, available contextual information on hydrologic systems, and remote sensing information. The project supervisor is Mark Handcock at Penn State (see section 3.1.1 below).

Co-organizers: Tony Olsen, EPA; Sarah Nusser, Iowa State University; Ray Czaplewski, USDA Forest Service Rocky Mountain Research Station; Loveday Conquest, U. of Washington/NRCSE

The conference, "Environmental Monitoring Surveys over Time", was held April 20-22, 1998, on the University of Washington campus in Seattle, Washington. The conference was organized and partially supported by the National Research Center for Statistics and the Environment (NRCSE); additional funding was provided by the Natural Resources Inventory and Analysis Institute of the US Dept. of Agriculture's (USDA) Natural Resources Conservation Service, and the Inventory and Monitoring Institute of the USDA Forest Service. Approximately 65 statisticians, biometricians, and environmental scientists exchanged state-of-the-science information in a series of 14 invited paper sessions. The objective of the conference was to provide a summary of the current state of statistical methodology for conducting longitudinal natural resource and environmental surveys; it was organized around design and analysis issues, social science issues pertinent to natural resources.

Invited paper sessions discussed current surveys for a number of natural resources, proposed modifications for surveys, and discussed promising approaches for future surveys. Sessions addressed terrestrial surveys, human population and institutional surveys, aquatic and avian surveys, remote sensing, watershed surveys, integrating different surveys, non-sampling errors, database construction and dissemination, and statistical estimation issues. Also included were perspectives from longitudinal surveys in other subject matter areas as a means of providing cross-fertilization between natural resource survey scientists and those involved with surveys of agricultural production, economic indicators, and human populations.

Program chair: Peter Guttorp, NRCSE

Steering committee chair: Francis Zwiers, Canadian Climate Center

Local organizer: Richard Lockhart, Simon Fraser University, Canada

The 7th International Meeting on Statistical Climatology took place at Whistler resort in British Columbia, Canada, May 25-29, 1998. About 150 participants from six continents participated.

The series of meetings on statistical climatology started as an ISI satellite meeting in Japan in 1979. Since 1983 the meetings have been held every three years. These meetings are unique in that they are not run by any scientific organization. Since 1987 the meetings have been organized by a free-standing steering committee, currently chaired by Francis Zwiers. Support is sought from a variety of organizations, and co-sponsorship is usually sought from national and international statistical and meteorological scientific societies. The late Allan Murphy of Oregon State University was instrumental in initiating and maintaining this series of talks, and the meeting was dedicated to his memory.

The program of the Whistler meeting was arranged with two plenary special invited sessions per day, in which prominent climatologists and statisticians gave in-depth presentation, followed by two invited discussants (one statistician and one climatologist). The format with invited discussants, while common in statistics, was a (popular) novelty to many of the climatologists. There were parallel invited and contributed sessions at other times of the day.

Among the climatological special invited papers were discussions of neural networks applied to remote sensing problems (Vladimir Krasnopolsky), and the North Atlantic counterpart to El Nino/Southern Oscillation (Tony Barnston).

The NRCSE support for this conference included maintaining the web site, automatic posting of abstracts, and editing of the abstract booklet and program. Among NRCSE participants were Jim Hughes, Chris Bretherton, Barnali Das and Peter Guttorp. The program can be found in Appendix C.

A number of Center members and graduate students have given presentations and organized sessions at national and international meetings of various scientific organizations. These include the Society of Toxicology annual meeting, the Novartis symposium on environmental statistics, the Interface conference, the Joint Statistical Meetings, and the 7th International Meeting on Statistical Climatology. A detailed list of presentations is given in Appendix C.

David Ford has made a substantial contribution to the White Paper on Model Assessment now submitted to the EPA Science Policy Council. The crucial contribution was to illustrate how different programs within EPA were all considering model assessment and developing approaches to it, but each had a different emphasis and/or used different terms. By defining different components of "uncertainty" it was possible to illustrate to the diverse members of the EPA team involved in preparing and critiquing the White Paper that an overall EPA policy could be developed that still permitted the necessary flexibility that the programs needed. The White Paper is being discussed by the Science Policy Council on 5 November, 1998.

Peter Guttorp participated on a site visit committee for the Superfund Remediation project at University of California at Davis. This resulted in a joint committee report to the project officers with recommendations for the upcoming renewal application to the EPA for funding.

Gerald Van Belle has been involved in several activities related to the Health Effects Institute. He is Chair of the Oversight Committee for the National Morbidity and Mortality Air Pollution Study. In addition he is a member of the research committee, and chaired a meeting of NMMAPS and APHEA researchers in London, England, July 15, 16, 1998. (APHEA=Air Pollution and Health–a European Approach)

Alison Cullen, School of Public Affairs, received the Outstanding Young Scientist award of the International Society of Exposure Analysis at their annual meeting in Boston, 1998.

Loveday Conquest, Fisheries, was chosen as the first director of the newly formed Teaching Academy at the University of Washington. The Academy consists of winners of the UW Distinguished Teaching Award.

June Morita, Statistics/Management Science, received the American Statistical Association Chapter Service Award at the national meeting in Dallas, 1998.

One of the important focuses of the Center is on educational outreach. Two of the main projects are development of multimedia modules for teaching environmental statistics and curriculum development on the scientific method, using environmental and ecological illustrations (section 3.1.2) We are assisting the EPA Region X office with a jointly funded graduate student intern for consulting help. We are also developing some web-based course material at a graduate level, and a course in Environmental Statistics using a case-based pedagogical approach. In addition, visitor Michael Phelan taught a course for graduate students in Environmental Statistics with emphasis on economic analysis for the Statistics department during Summer quarter of 1998. We hope to be able to teach such a course each summer.

We have continued our research links with the Washington State Department of Ecology, mainly in the area of air pollution (specifically ozone and car exhaust). These research projects are joint with the University of Washington Statistical Consulting Center.

We are continuing work on developing links with local industry. A natural link already exists to the Boeing Company through our visitor Dean Billheimer, who spends one day a week at the Center. We are working on furthering these links through a variety of joint projects with the Departments of Statistics, Applied Mathematics, and Mathematics. A recently funded NSF proposal (joint between the three departments) is aimed at vertical integration of education and research, and will provide vehicles for involving undergraduates, graduate students, and postdocs, in research group activities.

NRCSE is developing as a national research center using a strategy with three components. First, guidelines for funding Center members to work on specific research projects (section 3.1) specify the importance of identified EPA contacts to ensure the relevance of the projects to the EPA mission. Second, the Center is emphasizing its growing visitors program with researchers from outside the University of Washington visiting the Center to set up joint research programs with one or more Center members (section 3.2). Third, Center members work on joint proposals with researchers in other insititutions nationwide. We have adopted a national perspective in our increasing activity building and maintaining contacts with other researchers in environmental statistics. In addition, the Center computing staff is continiuing to evaluate and implement tools for collaborative research at a distance.

Center members are entitled to submit proposals to the Center once a year, usually in March. A total of 11 proposals were submitted, all of which received at least partial funding. Some projects were also continuing from the previous year.

The current work at the Center can be divided into six areas:

1. Ecological impact
2. Education and outreach
3. Model assessment
4. Space-time modeling
5. Sampling and design
6. Standards and regulatory impact

EPA connection: Tony Olsen

Center researchers: Dean Billheimer, Jim Karr.

Research assistants: Mariabeth Silkey, Florentina Bunea.

This project deals with the statistical analysis of compositional data in space and time. Among the applications are realistic simulations of benthic population data for streams in order to derive statistical properties of measures of water quality such as the Index of Biotic Integrity (IBI). The statistical aspects of the project focuses on the Billheimer model of space-time compositional data, which the Master's thesis of Mariabeth Silkey (1998) used to assess trends and design aspects of the EMAP benthic monitoring project in Delaware Bay, and the use of graphical models to assess the components of the IBI. Recent work includes an assessment of the statistical variability of the IBI (Bunea and Guttorp, 1998, manuscript in preparation).

PI: Mark Handcock, Penn State University

Co-investigators: Joe Sedransk, Case Western

EPA researcher: Tony Olsen

Research Assistant: James McDermott, Penn State University

PI: Dennis Lettenmaier

EPA researcher: Iris Goodman

Research assistant: Laura Bowling

There is a perception in the Pacific Northwest that the frequency and severity of flooding has increased in the western Cascades due to forest harvest. Field studies have shown that substantial changes in snowmelt during rain-on-snow events can occur following the removal of forest cover due to differences in snow accumulation as a result of canopy interception changes, and enhanced latent and sensible heat transfer associated with increased wind at the snow surface. However, field studies are of necessity essentially snapshots; at the watershed scale, the effects of vegetation changes on any particular flood are complicated by variations in antecedent snow accumulation, spatial differences in temperature and precipitation during the storm, and the area-elevation distribution of the watershed.

From a statistical standpoint, retrospective assessment of the effects of logging on streamflow is a classical trend detection problem. An analysis of changes in annual maxima (AMS) and peaks-over-threshold (POT), uncorrected for climatic trends, was conducted for 26 Western Washington basins, ranging in size from 13.8 km² to 1560 km² using the non-parametric Mann-Kendall test. The basins were classified into three categories based on record length. Statistically significant increases in AMS or POT were found in 5 basins with short records (1960-1996), 4 basins with medium records (1945-1996) and 3 basins with long records (1930-1996). A short record length makes the trend analysis more sensitive to climate variability. Two techniques were used to correct for the potential influence of climatic trends: paired catchment analysis and analysis of model residuals.

Paired catchment analysis requires that adjacent, similar watersheds be identified that have had much different logging histories. Since both basins are driven by the same sequence of meteorological events, analysis of the discharge difference series should filter out systematic climate variations. Seven basin pairs were selected based on vegetation differences as predicted by Washington Department of Natural Resources canopy cover classifications. Significantly increasing trends in annual maxima were found for two of the basin pairs.

An alternative approach is to control for the effects of climate variability by analyzing the residuals of flood peaks predicted using a deterministic, spatially distributed hydrologic model with fixed vegetation (Wigmosta et al., 1996; Storck et al., 1995; 1996). The residual series (simulated less observed discharge) should filter out any systematic effects due to climate. An analysis of model residuals for the main stem Snoqualmie River detected a statistically significant increase in the smaller storms of the POT series.

These results are summarized in a journal article (Bowling et al., 1998) submitted to Water Resources Research.

PI: David Madigan

Center researchers: Peter Guttorp

Research staff: Erik Christiansen, Peter Sutherland

Research assistant: Tamre Cardoso

The main outreach tool this group maintains is the web page. In addition, the group has put seminars on the Web, investigated long distance collaborative computing tools, and developed long-range plans for the Center computing facilities. Work is in progress on a system for making Center-related software available to the community and adding web-based discussion tools to the web site. In particular, graduate student Tamre Cardoso has ported Doug Nychka's (NCAR) package FUNFITS, a collection of programs based in S-PLUS (on Unix) for curve and function fitting and spatial design, to run under S-PLUS version 4.0 for Windows. The Windows version is available as a self-extracting zip file. A current version is available on the NRCSE ftp server (ftp://nrcse-svr.stat.washington.edu).

Co-PIs: Alison Cullen and June Morita

Research assistant: Lynn Coriano

A graduate student at the School of Public Affairs, Lynn Coriano, wrote her MPA degree project on the topic "An Opportunity in Education: Promoting the Environment". She explored the approach to environmental education currently in practice in the US and in particular Washington state. In her conclusions she recommends that environmental education be infused into the classroom in all subject areas. She identified a lack of curriculum materials and teacher preparation as preventing fuller environmental education at the present. There are many opportunities for web based lessons and quantitative exercises with the environment as the major theme. As a practicum related to this degree project she developed 5 lesson plans for use with grades 6-8 in a unit titled "Quantitatively-Based Watershed Lesson Plans".

PI: June Morita

Research assistants: Kathryn Kelsey and Ashley Steel

This project, which is jointly funded by the Discuren Foundation and NRCSE, is developing and implementing a 10 week curriculum for middle and high school students about the process of scientific research–from hypotheses and research design to statistical analysis and presentation–using structured activities and long-term independent research projects. The curriculum materials are being piloted at two of Seattle's schools, and will be implemented in a variety of middle schools in Seattle school district in spring of 1999.

Co-PIs: Peter Guttorp, Paul Sampson

EPA contacts: Patricia Cirione, Dianne Ruthruff, EPA Region X.

Center researcher: Steve Millard (PSI)

Research assistant: Kris Ryding

The Center was approached during the summer of 1997 by personnel from the regional EPA office in Seattle about developing an introductory to intermediate series of lectures and computing exercises for office personnel. Due to timing problems, the original plan, which involved Center researchers Loveday Conquest and June Morita, could not be implemented. Instead, center affiliate Steve Millard (Probability, Statistics & Information) will be teaching the course. Kris Ryding, a QERM graduate student, has been hired jointly by the Center and the Region to serve as a statistical consultant at the regional office.

PI: David Ford

Research assistant: Tamre Cardoso

This project continues the Point process manual project from last year. We have conducted a web page search and located web pages that provide information on spatial statistics. We have prepared a draft web site that is not currently available to the public. This will be used and developed during Winter Quarter when "Spatial Processes in Ecology" will be taught as a graduate course at the University of Washington. Substantial additions will be made and the web site will be made public after then.

PI: Peter Guttorp

EPA connection: George Flatman

Research staff: Peter Sutherland

A prototype chapter of the Bayesian statistical methods tutorial has been developed. It has been made available for limited alpha-testing on the web. Work is in progress on the tools needed to implement the tutorial, and on methods for distribution.

PI: Dean Billheimer

Research staff: Peter Sutherland

We are developing software tools to aid the implementation of Markov Chain Monte Carlo (MCMC) algorithms for statistical inference. These tools build on the JAVA-based architecture developed in 1997. Our goal is a framework to easily implement hierarchical models for two types of applications: analysis of designed experiments and time series. To date, our work has focused on two activities. First is the standardization of programming conventions and computational methods. Existing algorithms are being redesigned to improve their ability to adapt to greater variety of data formats and to offer greater range of model structure. Additionally, methods of automating 'first pass' model tuning are being added. These methods allow a researcher to explore initial model parameters and view descriptive statistics of MCMC output. Our goal is to provide the researcher guidance in MCMC parameter selection.

Our second major activity has been in developing graphical representations of MCMC model output. Several projects at the center require similar high-dimensional visualization components. These components are being designed to integrate into the Bayesian tutorial and the graphical interface work being done for SPlus, as well as the MCMC model framework. To date, components for viewing histograms, ternary diagrams, and a general data translator to handle several standard data formats have been developed.

PI: June Morita

Center researcher: Peter Guttorp

In conjunction with the outreach activities of the Center, the goal of quantitative literacy for all citizens is important. The main target group for this project is school children and their teachers. The Center has offered minor support to June Morita for her work on activity-based mathematics education. The publication Morita (1999) is a result of this.

Co-PIs: Peter Guttorp and Jim Hughes

Center researcher: Chris Bretherton

Research assistant: Enrica Bellone

In assessment of global warming, much use is made of deterministic models of general atmospheric and oceanic circulation. These general circulation models generally are on too coarse a scale to produce realistic precipitation scenarios on local (or meso-) scales. We are developing stochastic models of precipitation that use atmospheric pressure and temperature data as input, and produce precipitation forecasts at observation stations or at unobserved sites as output.

Enrica Bellone, a graduate student in the department of Statistics and funded by the NRCSE, has been working under the supervision of NRCSE members Peter Guttorp and Jim Hughes to develop such models. In previous work Hughes et al (1998) developed a nonhomogeneous hidden Markov model for relating precipitation occurrences to atmospheric circulation. The present work extends these results to include precipitation amounts. Preliminary results were presented at the Sixth International Conference on Precipitation (Bellone et al, 1998, contributed poster) and a manuscript is in preparation. Related work has been carried out by NRCSE member Jim Hughes and colleagues from Australia's Commonwealth Scientific and Industrial Research Organization (CSIRO). Bryson Bates and Stephen Charles, both of the CSIRO Land and Water Division in Perth, Australia, visited the NRCSE in June of 1998 to work with Dr. Hughes on developing models for downscaling precipitation in western Australia. Preliminary results on this work were presented at the Sixth International Conference on Precipitation (Charles et al., 1998b, contributed) and two manuscripts have been submitted (Charles et al., 1998a, Charles et al., 1998b)

Co-PIs: Alison Cullen and Adrian Raftery

Center researcher: Chris Bretherton

EPA researcher: Allan Marcus

Research assistant: Samantha Bates

A Superfund clean-up is underway at the New Bedford Harbor site in Massachusetts, where marine sediments are contaminated with poly-chlorinated biphenyls (PCBs). Harbor dredging at the site and subsequent transport and deposition may result in human exposure via air, soil and ingestion of locally grown foods. Sampling of households and farms around the site in 1994 and 1995 yielded produce, air and soil samples which in turn provided measurements of PCB concentration in soil, outdoor air and root, leafy and vine plants. A probabilistic exposure assessment in which average annual exposure to local inhabitants is assessed, is underway at the site. This assessment requires distributions for PCB concentration in soil and in root, leafy and vine plants. The work has resulted in one paper: Vorhees et al., 1997. In the coming year, the Bayesian technique will be used to incorporate both prior knowledge and likelihoods in order to find posterior distributions of PCB concentration in leafy, root and vine plants.

Ongoing collaboration between Cullen and Bretherton involves improving the development of distributions of contaminant concentration representing annual averages from data in which dependency is present among daily measurements. In a second collaboration, a simple compartmental model for PCB concentration in soil (adapted by Cullen, 1992) serves as the basis for the application of the Bayesian Synthesis method of Raftery. This method incorporates both prior knowledge and likelihoods for the soil model inputs and a likelihood for the soil model output in order to find posterior distributions of PCB concentration in soil. It uses Monte Carlo simulation from the prior distributions for the model inputs and resamples these with weights determined by both the air and soil likelihoods. In this way, the posterior distribution of PCB concentration in soil incorporates both expert/prior knowledge of the model inputs and the observed data.

In December 1998 at the Society for Risk Assessment (SRA) Annual Meeting in Arizona, Bates will present a paper titled "A Bayesian Synthesis Approach to assessing exposure to PCBs in New Bedford Harbor," co-authored by Cullen and Raftery. Bates received a Student Travel Award from the SRA for this work. Cullen will present a paper entitled "Developing Distributions of Annual Average Concentration with Dependency among Daily Values," co-authored by Christopher Bretherton.

PI: Alison Cullen

Collaborator: Michael Brauer, UBC, Canada; Eleanora Fabianova, Eva Mikhalikova, Peter Miskovic, Frantiska Hruba, SUHE, Slovakia.

EPA researcher: John Vandenberg

Recent interest in the levels of and health effects associated with airborne particulate matter exposure have sparked studies in the US and worldwide. Working with local scientists we are examining new measurements of PM2.5 taken by personal monitors in occupational settings, both industrial and office type, and in the home, by researchers at the SUHE (Institute for Epidemiology and Hygiene) in Banska Bystrica, Slovakia. This work also involves Michael Brauer at UBC and John Vandenberg at HERL, EPA, and has received funding from the Joint Fund for US/Czech/Slovak Science and Technology. Regression analyses will be carried out in the coming year to identify factors influencing particulate matter exposure in Slovakia and to support the standard setting process.

The research team visited Seattle in August 1998 to discuss preliminary results and to plan next steps. At this meeting the group prepared a talk for the ISEA (International Society of Exposure Analysis) annual meeting in Boston. The talk entitled "US-Slovak Cooperation in Environmental Health Risk Assessment: Preliminary Estimates of Personal Exposure to Particles and NO2 in Banska Bystrica, Slovakia " was presented by Eva Mikhalikova. Further analysis was planned and a talk describing additional analyses will be presented by Frantiska Hruba at the Society of Risk Analysis Annual Meeting in Phoenix in December 1998.

At the request of EPA's Vandenberg, NCRSE hosted additional meetings between the visitors and researchers from UW involved in PM work including: Jane Koenig, Tim Larsen, Lianne Sheppard, Sally Liu, and Dave Kalman, as well as Tim Nyerges of UW Dept. of Geography's GIS in decision making group. During these sessions Vandenberg highlighted the interests and needs of EPA anticipated in this area.

PI: Paul Sampson

EPA connections: Sharon LeDuc, Brian Eder, Larry Cox

Center researchers: Peter Guttorp, Joel Reynolds, Wendy Meiring

Research assistants: Ruth Grossman, Doris Damian

This project aims at developing tools for model assessment, using model runs from the SARMAP air quality model for the San Joaquin Valley in Central California. The model assessment work focuses on fitting a nonstationary space-time covariance structure to observed data, and using this covariance to estimate (with specified uncertainty) the ozone levels in the grid squares for which the model produces output. We will pursue these ideas using RADM and MODELS-3, where longer runs of the model will enable us to also compare the covariance structure of the model output to the covariance structure inferred from the data. The project was presented at the Novartis symposium on environmental statistics in London, and the presentation is available as a technical report (Sampson and Guttorp, 1998).

Recent work (not reported in the symposium-based technical report) includes empirical modeling of temperature effects on the San Joaquin Valley ozone monitoring data for more precise spatial estimates or interpolations. In addition, the development of a Bayesian estimation methodology using Markov Chain Monte Carlo for the nonstationary spatial covariance model of Sampson and Guttorp is under development.

PI: Elaine Faustmann

EPA researchers: Woody Setzer and Chris Lay

Center researchers: Brian Leroux, Scott Bartell, Rafael Ponce

The last twenty years have seen the development of numerous models for predicting the kinetics of lead in the human body. These models are necessary because health effects have historically been linked to specific blood lead concentrations, while pollution control and industrial hygiene efforts are most easily directed at environmental (e.g. air, water, soil, and food) lead concentrations. Exposure models and toxicokinetic models provide the quantitative link between environmental concentrations and biomarkers such as blood lead concentration.

Toxicokinetic models are particularly useful when toxicant exposure patterns are temporally complex or when the distribution of a toxicant in the body is significantly affected by physiologic changes such as normal human growth, pregnancy, and aging. Childhood lead exposure kinetics appear to be influenced by both temporal complexity and normal physiologic growth. In addition, children are particularly susceptible to the neurotoxic effects of lead, which may include increased blood pressure, colic, encephalopathy, intelligence quotient deficit, and aggressive behavior . Because of these factors, and because children typically engage in mouthing behavior that results in higher lead exposures than for most adults, childhood lead toxicokinetic models play a prominent role in regulatory decisions and standards regarding environmental lead.

EPA requested that the NRCSE review childhood toxicokinetic lead models and suggest additional validation strategies. The agency is particularly interested in validation of its own model, the Integrated Exposure and Uptake Biokinetic lead model (IEUBK), which predicts blood lead concentrations for children ages 0 to 7 based on environmental lead concentrations.

One of the most controversial parameters in exposure models is the soil ingestion rate. Experimental estimates are usually determined from tracer studies, in which aluminum, silicon, titanium, and other rare earth elements are measured in the diet, urine, and feces. Steady state conditions are assumed, and mass balance approaches are used to estimate the rate of soil ingestion. Soil ingestion rate estimates derived in these studies vary by several orders of magnitude, appear to fluctuate daily for each monitored individual, and are highly dependent on the tracer and statistical model selected .

An alternative to the tracer study is the use of pollutant biomonitoring studies that include environmental measurements. We have obtained data from one such study, the Urban Soil Lead Abatement Demonstration Project (USLADP), in which children’s blood lead concentrations were monitored for two years following the replacement of contaminated yard soil with soil with lower lead content . A perturbation analysis is being conducted using a previously developed toxicokinetic model (Bartell, 1996) to estimate a soil ingestion rate for each child in the USLADP study. The model includes a probabilistic uncertainty analysis component that assesses the impacts of parameter uncertainty of each child’s predicted soil ingestion rate.

We have recently completed the model runs for this analysis, and are now compiling the results. We are preparing "Estimation of soil ingestion rates from observed blood lead loss following soil remediation" for submission to Environmental Health Perspectives by January 31, 1999. Results will also be presented as "Uncertainty and variability in childhood soil ingestion rates estimated from USLADP blood lead levels" at the International Society for Risk Analysis annual meeting in 1999.

2. A toxicodynamic model of 5-fluorouracil-induced developmental neurotoxicity in the fetal rat

The chemotherapeutic agent 5-fluorouracil (5-FU), and other fluoropyrimidines, are known teratogens in a number of species. Among the most prevalent developmental effects of fluoropyrimidine exposure are dose- and stage- dependent hindlimb effects. Shuey et al reported a sequential biochemical and cellular alterations following 5-FU exposure in the developing limbs (Shuey et al., 1994). These effects elicited by 5-FU were later integrated into an empirical model (Shuey et al., 1994).

A biologically-based dose response model for developmental toxicants has been developed by Leroux et al. (1996). Unlike other empirical models, this model simulates developmental outcomes based on stochastic probability distribution of crucial developmental events such as cell differentiation, cell cycle and cell death. The pattern of malformation rate is predicted as a function of critical number of committed cells in a both dose- and time-dependent fashion. Because this model incorporates events that are common targets of many developmental toxicants, the potential application of this model to simulate the toxicity of other developmental toxicants is implicated.

The goal of current studies are to apply the Leroux model to simulate the developmental toxicity of 5-FU and methylmercury . Evaluation of the model parameters for 5-FU toxicity will be largely based on experimental data reported by Shuey et al. (1994) and other recent work from Dr. Setzer's group.

The framework of the model for the developmental toxicity of 5-FU is based on the original Leroux model. Digit agenesis at term will be used as the developmental toxicity endpoint, and will be modeled in a function of cell numbers of committed Y cell on Day 17. By use of in vivo data from the literature and the in vitro laboratory studies (Shuey et al., 1994), four types of model parameters were evaluated: (1) cell numbers during gestational Day 14 to Day 17; (2) cell transformation rate; (3) cell death rate; (4) cell cycling rate.

Recent experimental findings pertaining to the current modeling work were requested via email from Dr. Setzer's group.

Dr. Setzer's group forwarded their recent results on the effect of 5-FU exposure on cell cycle distribution. For example, their studies demonstrated an increased percentage of cells in the S phase, accompanied by a decrease in the G1 phase following 5-FU treatment. This information, however, did not indicate the alteration of cell cycle kinetics, or cell cycling rate by 5-FU. To obtain the cell cycling rate of control and 5-FU exposed limb cells, a critical model parameter of Leroux model, several assumptions have to be made as discussed in the following sessions.

Although their recent results included a complete measurement of the total protein and DNA, they were measurements form the whole embryos. To calculate cell numbers in the developing limb, protein values of embryonic limb from their previous publication (Shuey et al., 1994) were used.

Studies are currently underway in Dr. Setzer's group to determine the pharmocokinetics behavior of 5-FU. A simple PK model has been developed. Once their studies are complete, information on actual concentrations available to fetus during gestational Day 14 to Day 17 will be useful in further refining the current model.

Previous work conducted in this laboratory has led to the development of a toxicodynamic model of methylmercury-induced developmental neurotoxicity. Methylmercury is a naturally occurring organometal that is of concern because of the large population exposed through fish consumption and because epidemiological studies implicate even low levels, such as those expected among subsistence fish consumers, with adverse neurobehavioral development. Because the toxicodynamic model that has been developed is biologically based, it may be generally applied to agents that cause developmental toxicity through interference with cell proliferation. Such models could also allow cross-species extrapolations based on the incorporation of species-specific rates in model variable parameters; future applications of the pharmacodynamic model to other developmental neurotoxicants such as 5-fluorouracil should allow us to explore these issues. Such biologically-based models can thus reduce uncertainty, identify research needs, and improve estimates of developmental risks to humans from environmental exposures.

During the last year we have performed simulation testing of model variables to assess model sensitivity. We have examined all model assumptions and variable parameter estimates to determine their impact on model predictions. This analysis demonstrated sensitivity of the model to both the rates of cell division and the assumption of the critical number of cells for proper midbrain functioning. Example results from this sensitivity analysis are presented in the appendix. Of interest are refining estimates of the number of starting cells in the GD 12 rat midbrain, estimation of the fraction of cells dividing over GD12-17 (currently assumed to be a constant 55%), refined estimates of cell division over GD12-17, and improved methods for identification of differentiating cells.

To complement the existing toxicodynamic model, Mr. Lewandowski and others have developed a toxicokinetic model to predict maternal and fetal disposition of methylmercury during gestation; this toxickokinetic model is linked with the existing toxicodynamic model (see paper in appendix for diagrams and detailed description). The idea underlying the development of such a biologically based toxicokinetic-toxicodynamic model would be that one could relate a whole body dose, that is delivered by various routes of exposure, to an observable effect on the developing fetus.

The linking of the toxicokinetic and toxicodynamic elements of the model occurs using recursive model solutions over short intervals of time. For any given methylmercury dosing pattern, the fetal brain methylmercury concentrations predicted by the toxicokinetic model during each time step is used by the toxicodynamic model in propagating the effects of exposure on cell death, differentiation or division. Because maternal and fetal physiology change as gestation proceeds, parameters for model variables are adjusted to account for fetal growth and gestational stage. This is an important development to the model because many of the toxicokinetic and toxicodynamic properties used to parameterize model variable change as gestation proceeds. This modeling effort serves as the basis for Mr. Lewandowski’s dissertation research. Preliminary results of these efforts have been presented at the Society of Toxicology Annual Meeting, 1998 and the IUTOX Congress, 1998.

PI: David Ford

EPA researchers: Sharon LeDuc, Bill Benjey

Center researcher: Joel Reynolds

Research assistant: Marianne Turley

During the past year our principal task has been to develop software, and an associated manual, for Pareto Optimization of ecological models. There have been three component tasks: (i) development of an improved evolutionary search software for the Pareto front, (ii) development of an appropriate ecological computer model for use in conjunction with the manual, (iii) writing the manual itself.

(i) Development of improved evolutionary search software for the Pareto front: POMAC_Evolve.

An evolutionary computation optimization program, POMAC_Evolve, was developed and coded. While some of the data structures and general program outline were adapted from earlier prototype code (Reynolds, 1997), deficiencies were found which necessitated the selection of a new search algorithm and subsequent complete code redevelopment and writing.

The algorithm used in the prototype code was found to be susceptible to 'genetic drift' - the stochastic search became unduly focused on a small region of the parameter space. As the goal of POMAC is to survey the full Pareto Front rather than just find a restricted region of the Front, it was essential to revise the fitness assignment to each parametrization in order to avoid restricting the optimization search too quickly to a small region. The technique of fitness sharing (or niche sharing), which basically downweights the fitness measure of a parametrization by a function of how many other parametrizations are 'nearby' in the objective or criteria space, was incorporated into the new code in order to overcome this difficulty.

POMAC_Evolve searches out a model's Pareto Optimal Front using a non-dominated sorting algorithm (Srinivas and Deb, 1995), the most promising evolutionary computation multi-objective optimization algorithm appearing in our review of the current literature. The code is running and undergoing debugging and initial algorithm investigation.

(ii) Development of an appropriate ecological computer model for use in conjunction with the POMAC software and manual.

It is important that the POMAC manual be instructive, not only in how to use the technique but for its scientific value. This can best be done by applying the technique to an existing model but one that is simple enough for readers to understand but sufficiently complex that problems with the more usual types of model assessment can be illustrated.

During the past year, the plant competition model from Ford and Diggle (1981) has been translated from FORTRAN to C to be the working program for testing the POMAC program. This is a frequently cited model in the ecology literature. Unusually for ecological models parameter values were obtained using Simplex optimization against an integrated measurement of the outcome of competition. The model code is included with the manual as a sample for users. It has been well annotated so that users unfamiliar with the C language may be able to discern the structure and flow of the program.

(iii) The POMAC manual.

The target audience for the POMAC software is ecological and environmental modelers many of whom have had little or no instruction in optimization. The manual starts with a definition of the problem of optimization and illustrates some important features of how it can be used, e.g., that a single assessment criterion must be selected, and then proceeds to illustrate why considering multiple assessment criteria can be valuable.

PI: David Ford

Research assistant: Zoe Edelstein, University of Chicago undergraduate.

An important problem in the assessment of ecological and environmental models is that of repetition of the complete process of model fitting to new data sets. Where second data sets are available they are not treated as replicates but typically the question is asked: "Using the same set of parameters as obtained from fitting to the model to the first data set (i.e., obtained during calibration) does the model fit the second set?" This is referred to as validation. That such a procedure is not "validation", in the sense that a "successful" fit renders the model to be "true" is now generally accepted. But the question remains of what value a "second" data set is. A similar problem is faced when modelers take a single data set and break it in two and "fit" the model to one segment and seek to "test" the model against the second.

We have available a process model of plant competition that is fit to experimentally obtained data. The model has many characteristics of typically ecological and environmental models: it is stochastic and it predicts changes in the system modeled over time. Over the summer we conducted glasshouse experiments to provide two further instances of the data so that we now have four. We can now fit the model to obtain four sets of fitted parameters. We intend to treat these parameter sets as members of an ensemble and explore how such ensembles can be treated, both statistically and in interpretation of the system being modeled. This approach, of considering repetitions of the data as each producing a set of fitted parameters, brings a different perspective to the concept of model assessment that we will develop in future. The situation is similar to fitting a time series model to repeated realizations of a time series, though with an important difference. In time series from ecological systems the variation is often such that the order of the model changes, not necessarily by a great deal but sufficient to complicate comparison between model fits. In process models the structure of the model remains constant.

PI: Lianne Sheppard

Center researchers: Peter Guttorp, Paul Sampson

Research assistant: Doris Damian

A Bayesian approach to imputing air pollution exposure data is applied to monitoring data from Seattle. We are developing methods that allow for data missing at random due to temporary equipment failure and for data missing by design over time. Our focus is on methods that are computationally feasible for multiple years of daily observations from multiple monitoring stations.

During the year, progress has made assimilating the data set and developing a model for PM in the greater Seattle metropolitan area. We have also turned our attention to the health effects modeling aspect and discussed integrating the exposure modeling into that effort.

PI: Paul Sampson

Washington Department of Ecology researchers: Chris Figueroa-Kaminsky, Clint Bowman

Center researchers: Peter Guttorp, Joel Reynolds

Research assistants: Barnali Das

The NRCSE/ Washington Department of Ecology jointly sponsored project to adjust meteorologically the surface ozone network observations for Western Washington over the last 20+ yrs and assess time trends has been completed. Besides the results of the analysis, of interest to researchers and agency managers in the region, the project produced a new methodology, canonical covariance analysis (based on using the singular value decomposition), for meteorological adjustment of surface ozone observations when presented with both a spatial network of ozone monitors and a spatial network of meteorological stations.

During the year the project has resulted in one presentation (by Barnali Das at 7IMSC) and one technical report (Reynolds et al., 1998) on the methodological developments and the subsequent results. The technical report will be submitted for publication, but the authors are awaiting the completion of a second application of the methodology to data from the Chicago metropolitan area (see the review paper description below).

Co-PIs: Joel Reynolds, Peter Guttorp, Paul Sampson, Mary Lou Thompson

EPA researcher: Larry Cox

Research assistants: Barnali Das, David Caccia

We are in the process of producing a review of the statistical literature on the meteorological adjustment of surface ozone observations. The review consists of a summary and an evaluation of the methodologies present in the literature, together with a comparison of four selected methods applied to the Chicago metropolitan ozone data network. This comparison is based largely on an application to 10 years of Chicago area ozone and meteorological data which have been considered in applications of other methods in the literature. As a related project, Joel Reynolds and David Caccia are applying the Canonical Covariance Analysis technique developed for the NRCSE / Washington Department of Ecology ozone adjustment project (see above) to the Chicago area observations. A draft of the review has been produced, though the final version awaits completion of the Chicago area analysis.

PI: Mary Lou Thompson

EPA researchers: John Warren, David Farrar, Timothy Barry

Center researcher: Bruce Peterson (Terastat)

Research assistant: Kerrie Nelson

The statistical practices of chemists are designed both to minimize the probability of mis-identifying a sample compound and the probability of falsely reporting a detectable concentration. In environmental assessment, trace amounts of contaminants of concern are thus often reported by the laboratory as "non-detects" or "trace", in which case the data may be multiply left-censored. We consider the observations on each individual as being a nonhomogeneous Markov chain with three states: "non-detect", "trace" and "detect". Given the presence of "detect", the distribution of the observed measurements is modeled by some appropriate parametric form. This allows estimation of the parameters of the "detects" distribution and the proportion of censored values as a function of covariates (such as time, rural vs. urban etc.).

There has been considerable interest in this topic from colleagues at the EPA and we have had input from David Farrar and Timothy Barry regarding our research plan for the coming year. Kerrie Nelson, a graduate student in Statistics, has joined the project as an RA. The study generating the time series data that motivated the project is still in progress and we are focusing in the meanwhile on developing and implementing regression techniques for double censored data to a study of children's exposure to pesticides that was carried out in Eastern Washington. We are currently considering a maximum likelihood approach and an extension of the Buckley-James methodology to double censoring.

PI: Peter Guttorp

Center researcher: Paul Sampson

Research assistant: Barnali Das

When dealing with global stochastic processes (such as temperature), most work to date has (implicitly or occasionally explicitly) been focusing on covariance structures that are isotropic, or rotation invariant. In many situations such an assumption is not reasonable, if only for the fact that Earth has a rotational direction. This project involves finding methods to simulate nonstationary processes on the globe as well as estimating covariances from real data. The work to date has focused on the simulation part. In order to start work at a realistic estimation problem, we have obtained some data from Scott Doney of NCAR.

PI: Loveday Conquest

EPA researcher: Barry Nussbaum

Center researcher: David Marker (Westat)

Research assistant: Nicolle Mode

Ranked set sampling (RSS, McIntyre 1952) is a two-phase sampling procedure that reduces the number of samples required from a more exacting and expensive measurement by using expert knowledge or other frugal measurement to select the sampled values. The current project is an extension of work funded by NRCSE for 1997-98 which was completed last summer. The past RSS proposal consisted of three phases [1] literature review of previous work, [2] theoretical examination of RSS, and [3] application to a real data set. Based upon the first phase and in collaboration with statistical scientist Barry Nussbaum at EPA, we focused upon cost analysis of RSS for normal and skewed data (with and without errors in ranking) including a real data set of stream habitat. A paper based on this work (Conquest et al., 1998) has been submitted to the journal Environmetrics for publication, and a seminar summarizing our work was given as part of the NRCSE seminar series on May 19, 1998. The current research will compare theory and simulation to real ecological data and give guidelines for researchers and managers to use when designing a study. Specifically, we will investigate the cost-effectiveness of ranked set sampling on real ecological data sets, investigate a modified RSS estimator for use with ecological data, and give specific information for researchers that want to use RSS in the field.

Co-PIs: Peter Guttorp, Paul Sampson, Mary Lou Thompson

EPA researcher: Larry Cox

Research Assistant: David Caccia

The debate surrounding the change in ozone standards illustrates many of the difficulties in translating scientific studies into practical policy decisions. This projects studies ways of setting standards that makes use of the information available in a way that takes proper account of uncertainties in knowledge and understanding of the process, in measurement of the pollutants, and in enforcement rules. While the initial work has focused on ozone and particulate matter data, the intent is to produce a methodology that can be applied to a variety of environmental concerns. Products of this project include presentations at the Novartis workshop on Environmental Statistics in London (Larry Cox), at the Joint Statistical Meetings in Dallas (Mary Lou Thompson), and at the Newton Institute workshop on Environmental Statistic and Technology (Larry Cox). A paper has been produced for the proceedings of the Novartis workshop (Cox et al., 1998).

We are also considering questions of statistical monitoring network design and their relevance to siting of air quality monitors in the context of air quality standards. Presentations on the background issues were presented by Paul D. Sampson at the Joint Statistiscal Meetings in Dallas and in the NRCSE seminar series this past fall. In addition, new approaches to simulated annealing calculations for optimal monitor siting are being developed by David Caccia.

PI: Michael J. Phelan

Environmental health policy decisions are characterized by irreversibility and uncertainty of an economic, ecological and biomedical nature. Economic analysis of problems of this kind fall within the framework of the theory of irreversible investments as applied to the sunk costs and sunk benefits of environmental regulation. The proposed research describes an application of the basic theory to problems associated with the regulation of particulate matter in environmental health policy.

The first line of investigation models the social costs of regulation in light of current scientific, medical and economic understanding of problems associated with particulate matter. A particular emphasis is given to representation of health effects. All such models involve however some uncertain parameters, so a second line of investigation integrates modern practices of stochastic inference with sequential policy designs. An important goal is to characterize fully the role of uncertainty and information on the design and implementation of policy, particularly learning strategies designed to address key uncertainties. The analytical tools will be implemented numerically using MATHEMATICA.

The research has produced a technical report, Phelan (1998).

Much of the cross-disciplinary work at the Center is done in working groups that meet at regularly scheduled times (weekly or less frequently, depending on the group). The precipitation group, the health effects group, the standards group and the ozone groups have regularly scheduled meetings which are announced on the web page and open to any interested members of the campus community.

It is the stated intent of the Center to have a vigorous and stimulating visitors program. The visitors during the second year included

• Wendy Meiring, NCAR (work on software implementation; with Sampson and Guttorp)
• Dean Billheimer, Boeing (work on Markov chain Monte Carlo and on biological monitoring; with Guttorp)
• Phillippe Jamet and Jasha Osterbaan, Ecole des Mines de Paris, France (deterministic models; environmental characterization; with Sampson)
• Steve Smith and Bryson Bates, CSIRO, Perth, Australia (work on precipitation modeling; with Hughes)
• Peter Green, University of Bristol, UK (Markov chain Monte Carlo; with Guttorp and Hughes)
• Chris Glasby, University of Edinburgh, UK (image warping; with Sampson)
• Michael Phelan, Chapman University (environmental economics; joint with Statistics department)
• Noel Cressie, Iowa State University, and Merlise Clyde, Duke University (particulate matter air pollution; with Sheppard, Guttorp, Van Belle, Thompson and Sampson)
• Per Sørensen, Technical University of Denmark (remote sensing image processing; with Sampson and Ford)
• Eva Mikhalikova, Peter Miskovic, Frantiska Hruba, SUHE, Slovakia (particulate matter monitoring; with Cullen)

We are continuing to pursue invitations to prominent researchers to visit the Center for short or extended periods of time. During the academic year 1998-9 Merlise Clyde from Duke University will be on a sabbatical leave at NRCSE to pursue research interests in Bayesian model selection and particulate matter air pollution, and Di Cook from Iowa State University will be visiting for six weeks during Autumn 1998, working on high-dimensional dynamic graphics with environmental applications.

The following is a partial list of Center funded research products during the year.

Billheimer, D., Guttorp, P. and Fagan, W. F. (1998): Statistical Analysis and Interpretation of Discrete Compositional Data. Submitted to Journal of the Royal Statistical Society.

Bowling, L., Storck, P., and Lettenmaier, D. P. (1998): Assessment of the hydrologic effects of logging in the Pacific Northwest. Submitted to Water Resources Research, 1998.

Charles SP, Bates BC, Whetton PH, Hughes JP (1998a) Validation of downscaling models for changed climate conditions: case study of southwestern Australia. Submitted to Climate Research.

Charles SP, Bates BC, Hughes JP (1998b) A spatio-temporal model for downscaling precipitation occurrence and amounts. Submitted to Journal of Geophysical Research.

EF Melvin, JK Parrish, LL Conquest (1998): Novel Tools to Reduce Seabird Bycatch in Coastal Gillnet Fisheries. Submitted to Conservation Biology.

Lawrence H. Cox, Peter Guttorp, Paul D. Sampson, David C. Caccia and Mary Lou Thompson (1998) A Preliminary Statistical Examination of the Effects of Uncertainty and Variability on Environmental Regulatory Criteria for Ozone . To appear in a volume on Environmental Statistics published by the Novartis Foundation.

Cullen, A. and Frey, H. (1998): Probabilistic Techniques in Exposure Assessment: A Handbook for Dealing With Variability and Uncertainty in Models and Inputs, Publishing.

Hughes JP, Guttorp P, Charles SP (1998) A nonhomogeneous hidden Markov model for precipitation. J. Royal Stat. Soc., Series C, in press.

L. Kirkland, D. Hoffmeyer, H. Allender, L. Zaragoza, J. LaVeck, T. Barnwell, J. Fowle, J. Rowe and D Ford (1988): Science Policy Council Model Acceptance and Peer Review White Paper Working Group. The Nature and Scope of Issues on Adoption of Model Acceptability Criteria. EPA DRAFT.

Lewandowski, TA, Bartell, SM, Pierce, CH, Ponce, RA, and Faustman, EM (1998) Toxicokinetic and toxicodynamic modeling of the effects of methylmercury on the fetal rat [abs.]. The Toxicologist, 42(1-S), No. 683.

Lewandowski, TA, Bartell, SM, Pierce, CH, Ponce, RA, and Faustman, EM (1998) Effect of tissue binding uncertainty on a PBTK model of methylmercury in the fetal rat [abs.]. Toxicological Letters, 95(Suppl. 1), No. P2F148.

Mode, N., Conquest, L. and Marker D. (1998): Ranked set sampling for ecological research: Accounting for the total cost of sampling. Submitted to Environmetrics.

Morita, J.M. (1999): Capture and Re-Capture Your Students’ Interest in Statistics. Submitted to Mathematics Teaching in the Middle School.

Greg Nothstein, Dave Kalman, Gerald van Belle, Gardner Brown, Tim Larson (1988): An Evaluation of Public Willingness to Pay for Improvements in Visibility and Air Quality in the Pacific Northwest.

Phelan, M. (1998): Environmental health policy decisions: the role of uncertainty in economic analysis. NRCSE Technical Report NRCSE-TRS No. 12.

Reynolds, Joel H., B. Das, P. D. Sampson, P. Guttorp (1998): Meteorological Adjustment of Western Washington and Northwest Oregon Surface Ozone Observations with Investigation of Trends. NRCSE Technical Report NRCSE-TRS No. 15.

Reynolds, J. H., E. D. Ford (1999): Multi-Criteria Assessment of Ecological Process Models. Ecology 80(2).

Reynolds, Joel H. (1998): Causal Systems in Ecology. Letter in reply to Science's Compass essay. Science, 15 May, 280: 988-989.

Sampson, P. D. and Guttorp, P. (1998): Operational Evaluation of Air Quality Models. To appear in a volume on Environmental Statistics published by the Novartis Foundation.

Steel, E. A., and Guttorp, P. (1998): Modeling juvenile chinook salmon migration using a simple Markov chain. Submitted to Journal of Agricultural, Biological and Environmental Statistics.

Sheppard L, Levy D, Norris G, Larson TV, Koenig JQ (1999). Effects of ambient air pollution on non-elderly asthma hospital admissions in Seattle, Washington, 1987-1994. Epidemiology, Jan 1999.

Vorhees , D.V., Cullen, A.C. and L.M. Altshul (1997) Exposure to Polychlorinated Biphenyls in Residential Indoor Air and Outdoor Air Near a Superfund Site, Environmental Science & Technology, 31:3612-3618.

The fairly simple administrative structure of the Center seems to be sufficient to allow Center business to be performed smoothly. No additional administrative staff is currently contemplated.

The following new members have been elected during the year:

Chris Bretherton, Applied Mathematics and Atmospheric Sciences.

Patrick Heagerty, Biostatistics.

June Morita, Statistics and Management Science.

Joel Reynolds, Statistics.

John M. Wallace, Atmospheric Sciences.

The advisory committee met with acting director, Paul Sampson, during the Joint Statistical Meetings in Dallas in August, 1998. Agenda items prepared for review at the meeting were: the 1997-98 annual report; the upcoming events at NRCSE, including the PM Workshop scheduled for October and a workshop with EPA researchers tentatively planned for February, 1999; plans for the next Executive Committee election; an update on the NRCSE Newsletter; plans for the revamped NRCSE weekly seminar series; and a review of the office space situation for NRCSE in Bagley Hall. Other specific topics of discussion included the relationship between NRCSE and NISS, and the relationships between NRCSE researchers and EPA personnel. In particular, Sampson explained the Center's intention to require in future member proposals explicit plans for Center researchers to visit and make presentations at EPA labs.

Paul Switzer reported on his attendance at the EPA statistician's meeting. Larry Cox was positive on plans for the next EPA workshop and encouraged setting dates as soon as possible. He emphasized the importance of journal publications in the assessment of the Center. He also supported publication of NRCSE seminars and videotaping of both the upcoming PM workshop and the planned EPA workshop.

On the subject of the current focus on particulate matter research, Larry Cox noted that NRCSE might be considered as a "sixth PM center" in addition to the 5 PM centers that are soon to be established. Whether to strive for this status, or to consider an explicit link with one of the 5 PM centers (once awarded) should be considered. There was also discussion of the Federal Advisory Committee Act and NRCSE's possible relationship with, or contributions to the sub-committee on environmental statistics. Cox suggested that NRCSE might write a "white paper" on particulate matter statistical issues based on the upcoming workshop with suggestions to the agency.

On other topics, it was noted that if the Center had plans to move beyond its current situation as a primarily EPA-based (or EPA-limited) center, this is the year to do the planning. Paul Switzer noted some of the environmental areas that are not major research foci, including groundwater and toxic waste [although there is some work in these areas on campus, especially toxic waste in connection with the CRESP program]. Abdel El Shaarawi suggested that the Center bring in people/researchers from other agencies.

The Center has been allocated permanent space on the fourth floor of Bagley Hall. This space is shared with the Program on the Environment and the graduate program in Quantitative Ecology and Resource Management. The intent of the Center is that all researchers (including research assistants) who so wish may have access to a desk at the Center, adequate computing equipment and support, and reasonable office support. In addition, most visitors would be housed at the Center, although long-term visitors working with a Center member who does not use Center facilities may be housed in the member’s department or laboratory.

The Center has recently hired its first postdoctoral research fellows: Eun Sug Park from Texas A&M, and Kevin Brand from Harvard University. Dr. Park will be pursuing problems in identification of polluters from monitoring profile data, while Dr. Brand will be working on problems associated with human health risk analysis.

One of our main themes during year 3 of the Center will be research on particulate matter (PM) air pollution. The implementation of the PM-2.5 standard in 2002 means that quality research on health effects, monitoring network design, the spatial distribution of PM-2.5 and its relation to other air pollutants can have profound impact on the setting of the standard. Among the activities planned for the year are:

• PM science workshop October 19-22, 1998
• PM focus of the presentations in the Autumn quarter seminar series
• A directed research program, with additional funding from the EPA, involving researchers from North Carolina, Ohio State, Iowa State, Duke, Washington, Chapman and the EPA. The first aim is a special issue of Environmetrics.
• Two long-term visitors with PM focus: Merlise Clyde from Duke, and Allan Marcus from the EPA.
• A panel discussion on research direction for PM work at the Joint Statistical Meetings in Baltimore next August.

We intend to have similar focus activities in future years.

The EPA will be judging the Center performance at the end of year 3. Consequently it is of utmost importance that Center researchers cooperate with EPA researchers, and that work done at the Center is viewed as relevant to the EPA mission. We are currently discussing a variety of ways of informing the EPA community of our work, and the implementation of such plans will receive high priority during the year.

In addition to the PM workshop mentioned above, we are considering running workshops in other topics, including the meteorological adjustment of air pollution data, and sampling techniques for environmental monitoring.

The Center is being considered as organizers for the 2001 Environmetrics conference. The planning work will probably start during the coming year.

We are anticipating a continuing increase in the interest from researchers to visit NRCSE. In addition we put considerable effort into inviting researchers whose work fits well in with or complements Center research work. The system whereby the Director can make quick decisions on short visits has worked well, and will continue.

We intend to experiment with CD production of interesting software, tutorials, and workshops. For example, we anticipate to be able to distribute virtually all of the presentations at the PM workshop on a CD using a format similar to that used on the seminar web page. The CD will be made available to the community at cost or free.

The executive committee has discussed the possibility of producing a series of position papers on important issues in environmental statistics. It will probably be necessary to contact a commercial publisher to produce this series. Another possibility is to maintain an electronic environmental statistics journal, perhaps in conjunction with ASA and TIES.

October 7: James B. Hatfield, Hatfield and Dawson Broadcast and Communications Consulting Engineers. "Electromagnetic Fields and Human Health"

October 14: Tony Rossini, University of South Carolina, "ESS and Literate Programming: Computer Environments for Effective Statistical Programming and Data Analysis"

October 21: Jeffrey Richey, UW Oceanography, "PRISM and the NRCSE: A (Spatial) View to the 21^st Century in Puget Sound"

October 28: Doug Bright, Royal Roads University, Victoria, BC. "Environmental Science and Management in the Georgia Basin Coastal Zone: Tales of a Bottom Feeder"

November 4: James Hilden-Minton, National Institute of Statistical Sciences, "Multilevel Monitoring of Drinking Water Systems"

November 18: George Hampson, Woods Hole Oceanographic Institute. "Land Development vs. Environmental Health of a Small New England Island. Nantucket Harbor Study: Benthic Animal Communities and Habitat Quality"

November 25: William Lavely, UW International Studies and Sociology, "Infant Mortality in China: A Multilevel Model"

December 2: Marina Alberti, UW Urban Design and Planning, "Measuring Urban Sustainability"

December 9: Robin Matthews, Western Washington University, "Problems and Issues in Quantifying Ecological Risk"

January 13: Richard C. Pleus, Ph.D., Senior Toxicologist and Principal, Delta Toxicology/Intertox, "Foul odor or adverse health effect? Odor investigation of a Portland cement plant"

January 27: Dr. Robert Francis, Professor, UW School of Fisheries, "Climate and Large Marine Ecosystems: A Statistical View"

February 3: Dr. Thomas M. Leschine, Assoc. Prof, UW School of Marine Affairs, "Ranking and Rating Systems in Environmental Management: An Organizational Learning Perspective"

February 10: Dr. Russell P. Herwig, Research Assist. Prof., UW School of Fisheries, "Knock, Knock, Who’s There? Microbial Communities in Contaminated Puget Sound Sediments"

February 17: Thomas Lumley, Ph.C., UW Department of Biostatistics, "Marginal regression modelling of space and time data"

February 24: Dr. Joel Reynolds, UW Department of Statistics, "Pareto Optimal Model Assessment and Statistics: Thoughts from applying POMAC to a stochastic process model"

March 3: Dr. Charles Fowler, NOAA, "Sustainability: Empirical Examples and Management Implications"

March 10: Dr. Philippe Jamet, Ecole des Mines de Paris, "Macroscopic phenomenologies and implicit approaches in the quantification of mass transport in the geosphere"

March 24: Jasha Oosterbaan, Ecole des Mines de Paris, "Application of geochemical prospecting and exploratory data analysis methods in characterization at contaminated sites"

April 7: Bill Wright, Montgomery Watson Americas. "Application of Statistics and Probability to Environmental Problem Solving With An Emphasis on Risk Assessment: Case Histories and Reflections."

April 14: Manon Faucher, Environmental Adaptation Research Group, University of British Columbia. "The Climatology of Surface Marine Winds Near the Western Coast of Canada."

April 21: Elaine Faustman, Ph.D., Environmental Health, UW.

April 28: Steve Smith, Ph.D., National Marine Fisheries Service. "Factors Affecting Survival and Travel Time of Migrating Juvenile Salmonids in the Lower Snake River."

May 5: Annette Hoffman, Ph.D., Washington State Department of Fish and Wildlife. "A Statisticians Role in Resource Management."

May 12: Andrew Gelman, Department of Statistics, Columbia University, "Statistical issues in home radon mapping and remediation decisions"

May 19: Loveday Conquest, Ph.D., and Nicolle Mode, Quantitative Ecology and Resource Management (QERM), UW. "Ranked Set Sampling—What is it and is it any good?"

May 26: Samantha Bates, Department of Statistics, UW. "The Bayesian Synthesis Approach to Environmental Risk Assessment: Separating Uncertainty and Variability."

June 2: Chris Glasbey, Biomathematics and Statistics Scotland, Edinburgh, "Problems in image warping"

All presentations are invited talks unless specifically noted.

MONDAY, APRIL 20

Opening Remarks: 8:30–8:45 a.m.

Chair: Loveday Conquest

SESSION 1: 8:45–9:45 a.m.

Plenary Session

Chair: Loveday Conquest

SESSION 2: 10:00–11:45 a.m.

Terrestrial Surveys

Chair: Ray Czaplewski

SESSION 3: 1:00–2:15 p.m.

Perspectives from Human Population and Institutional Surveys

Chair: Sarah Nusser

SESSION 4: 2:30–3:45 p.m.

Aquatic and Avian Surveys

Chair: Tony Olsen

SESSION 5: 4:00–5:15 p.m.

Remote Sensing and Surveys

Chair: Ray Czaplewski

Joint Conference Dinner at Ivar’s Salmon House (optional)

TUESDAY, APRIL 21

SESSION 6: 8:00–9:45 a.m.

Design Issues In Aquatic and Watershed Surveys

Chair: Tony Olsen

SESSION 7: 10:00–11:45 a.m.

Annualized Modifications to Terrestrial Surveys

Chair: Ray Czaplewski

SESSION 8: 12:45–2:00 p.m.

Survey Integration Panel Discussion

Chair: Loveday Conquest

SESSION 9: 2:15–3:30 p.m.

Nonsampling Errors

Chair: Sarah Nusser

SESSION 10: 3:45–5:30 p.m.

Database Construction and Dissemination

Chair: Tony Olsen

WEDNESDAY APRIL 22

SESSION 11: 8:00–9:45 a.m.

Statistical Estimation: Approaches

Chair: Loveday Conquest

SESSION 12: 10:00–11:45 a.m.

Statistical Estimation: Applications

Chair: Sarah Nusser

SESSION 13: 1:00–2:15 p.m.

Statistical Estimation: Annualized Inventories

Chair: Tony Olsen

SESSION 14: 2:30–3:30 p.m.

7th International Meeting on Statistical Climatology

MONDAY, MAY 25

Organizing committee

8:45 Linda O. MEARNS, NCAR, USA

9:30 Discussants: Francis Zwiers, Canadian Centre for Climate Modelling and

Roger Davidson, University of Victoria, Canada

10:30 Robert L. WINKLER, Duke University, USA

10:50 Barbara G. BROWN, NCAR, USA

11:10 Richard W. KATZ, NCAR, USA

11:30 Seijo KRUIZINGA, Royal Meteorological Institute, The Netherlands

11:50 Francis ZWIERS, Canadian Centre for Climate Modelling and Analysis, Canada

12:10 Discussants: Dan Wilks, Cornell University, USA

Martin Ehrendorfer, University of Wien, Austria

12:30 General discussion

10:30 Nathaniel B. GUTTMAN, National Climatic Data Center, USA

10:42 Ramon A. QUINTANA-GOMEZ, Universidad Nacional Experimental de los

Llanos Ezequiel, Venezuela

10:54 Vicky SLONOSKY, University of East Anglia, UK

Multiple analysis of series for homogenization (MASH)

11:18 Lucie A. VINCENT, Environment Canada

11:30 Won-Tae KWON, Meteorological Research Institute, Korea

11:42 Vincent MORON, IMGA-CNR, Italy

11:54 Xiaolan WANG, Canadian Centre for Climate Modelling and Analysis, Canada

12:06 Michael WEHNER, Lawrence Livermore Laboratory, USA

12:18 Elena YULAEVA, University of California at San Diego, USA

12:30 Christine ZIEHMANN, University of Potsdam, Germany

10:30 Judit BARTHOLY, Eotvos Lorand University, Hungary

10:42 Christine MARTINEU, EDF/DER/Environment Dept., France

10:54 Olga C. PENALBA, University of Buenos Aires, Argentina

11:06 Amir SHABBAR, Environment Canada

11:18 Risheng WANG, Environment Canada

11:30 Takmeng WONG, NASA/Langley Research Center, USA

Chair: Olivier Talagrand, Ecole Normale Supérieure, France

11:42 Ulrich ACHATZ, University of Rostock, Germany

11:54 Juergen GRIESER, J.W. Goethe-University, Germany

12:06 Robert JACOB, University of Wisconsin-Madison, USA

12:18 Frank KWASNIOK, Max-Planck Institute for Meteorology, Germany

14:15 Mark BERLINER, Ohio State University, USA

15:00 Discussants: Hans von Storch, GKSS, Germany

Richard Lockhart, Simon Fraser University, Canada

16:30 Clive ANDERSON, University of Sheffield, UK

17:00 Richard L. SMITH, University of North Carolina, USA

16:00 Grant BRANSTATOR, NCAR, USA

16:30 Klaus FRAEDRICH, University of Hamburg, Germany

17:00 Huug M. van den DOOL, NMC, USA

17:30 General discussion

16:00 Frederic ATGER (presented by Olivier TALAGRAND), ECMWF, UK

16:12 Harold E. BROOKS, NOAA/ERL/National Severe Storms Laboratory, USA

16:36 Ignacio LOZANO, University of Alcala, Spain

16:48 Benyang TANG, University of British Columbia, Canada

17:00 Laurence WILSON, Environment Canada

17:12 Juying CHEN, Chinese Academy of Meteorological Sciences, China

17:24 Nityanand SINGH, Indian Institute of Tropical Meteorology, India

17:36 Neelima A. SONTAKKE, Indian Institute of Tropical Meteorology

9:00 Douglas NYCHKA, NCAR, USA

9:45 Discussants: Richard W. Katz, NCAR, USA

Noel Cressie, Iowa State University, USA

10:45 Gerald R. NORTH, Texas A&M University, USA

11:15 Simon TETT, Hadley Centre for Climate Prediction and Research, UK

11:45 Myles R. ALLEN, Rutherford Appleton Laboratories, UK

12:15 General discussion

10:45 Aristita BUSUIOC, National Institute of Meteorology and Hydrology, Romania

10:57 Dimitrios GYALISTRAS, University of Berne, Switzerland

11:09 James P. HUGHES, University of Washington, USA

11:21 Julie M. JONES, Climatic Research Unit, University of East Anglia, UK

11:33 J. P. PALUTIKOF, Climatic Research Unit, University of East Anglia, UK

11:45 Budong QIAN, University of Lisbon, Portugal

11:57 D. S. WILKS, Cornell University, USA

Statistical downscaling of daily precipitation using stochastic weather models

10:45 Alan CHIANG, University of Wisconsin-Madison, USA

10:57 Stefan GÜSS, GKSS Research Centre, Germany

11:09 Hauke HEYEN, GKSS Institute of Hydrophysics, Germany

11:21 Jianping HUANG, Atmospheric Environment Service, Canada

11:33 Lynne SEYMOUR, University of Georgia, USA

11:45 Xuebin ZHANG, CBB, Atmospheric Environment Service, Canada

14:00 Peter J. GREEN, University of Bristol, UK

14:45 Discussants: Paul D. Sampson, University of Washington, USA

James P. Hughes, University of Washington, USA

15:45 Gabriele HEGERL, University of Washington, USA

16:45 Tim BARNETT, Scripps Institute, USA

17:15 General discussion

15:45 Connely BALDWIN, Utah State University, USA

15:57 Roger R. DAVIDSON, University of Victoria, Canada

16:09 Marjana GAJIC-CAPKA, Meteorological and Hydrological Service, Croatia

16:21 D. GELLENS, Royal Meteorological Institute of Belgium

16:33 B. C. HEWITSON, University of Cape Town, South Africa

16:45 Alberto Solana ORTEGA, SCIC, Spain

16:57 Heikki TUOMENVIRTA, Finnish Meteorological Institute, Finland

15:45 Ulrike BURKHARDT, Reading University, UK

15:57 Rosa Hilda COMPAGNUCCI, University of Buenos Aires, Argentina

16:09 Barnali DAS, University of Washington, USA

16:21 C. A. GLASBEY, BioSS, UK

16:33 Arthur M. GREENE, Columbia University, USA

16:45 Wendy MEIRING, NCAR, USA

16:57 Belen Rodriguez de FONSECA, Universidad Complutense de Madrid, Spain

17:09 W. C. THACKER, Atlantic Oceanographic and Meteorological Laboratory, USA

17:21 Michael TOBIS, Argonne National Laboratory, USA

New technologies for very large empirical decompositions of simulated climate

9:00 Noel CRESSIE, Iowa State University, USA

Ian Jolliffe, University of Aberdeen, UK

Invited Session: Variational Methods Sea Ballroom 10:45 - 12:30

10:45 Joseph J. TRIBBIA, NCAR, USA

11:15 Andrew F. BENNETT, Oregon State University, USA

11:45 Grace WAHBA, University of Wisconsin, USA

12:15 General discussion

10:45 P. ANTICO, University of Buenos Aires, Argentina

Monthly anomalies in geopotential height fields over southern South America

11:09 Albert R. BOEHM, Nichols Research Corporation, USA

11:21 Manon FAUCHER, Atmospheric Environment Service, Canada

On the climatology of surface marine winds near the western coast of Canada

11:33 George GRUZA, Institute for Global Climate and Ecology, Russia

11:45 Donald V. HANSEN, CIMAS/University of Miami, USA

11:57 J. I. JIMENEZ, University of Granada, Spain

12:09 T. A. KABANDA, University of Venda, South Africa

12:21 Matilde RUSTICUCCI, University of Buenos Aires, Argentina

Contibuted Session: Precipitation I McGuire B 10:45 - 11:45

10:45 Annick DOUGUEDROIT, Institute of Geography, France

10:57 Sultan HAMEED, Institute for Terrestrial and Planetary Atmosphere, USA

11:09 E. KLOPPER, South African Weather Bureau, South Africa

11:21 C. Rodriguez PUEBLA, Titular University, Spain

11:33 A. C. CHIPANSHI, University of Botasqwana

Contributed Session: Climate Change Detection McGuire B 11:45 - 12:45

11:45 Shaleen JAIN, Utah State University, USA

11:57 J. F. GONZALEZ-ROUCO, Universidad Complutense de Madrid , Spain

12:09 Hans VISSER and Jeljer HOEKSTRA, KEMA Sustainable, The Netherlands

12:21 Nina M. DATSENKO, Hydrometeorological Research Centre of Russia

12:33 Esther RANKOVA, Institute for Global Climate and Ecology, Russia

9:00 Vladimir KRASNOPOLSKY, NCEP, USA

9:45 Discussants: Eduardo Zorita, GKSS Research Centre, Germany

Doug Nychka, NCAR, USA

Invited Session: Neural Networks Sea Ballroom 10:45 - 12:30

10:45 D. M. TITTERINGTON, University of Glasgow, UK

11:05 Mark FRENCH, University of Kentucky, USA

11:25 Stephane COTE, HMR Inc., Canada

11:45 Louis KEINER, World Data Center, USA

12:05 Dimitris TSINTIKIDIS, University of California at San Diego, USA

10:45 Martin EHRENDORFER, University of Wien, Austria

11:15 Zoltan TOTH, National Weather Service, USA

11:45 Olivier TALAGRAND, Ecole Normale Supérieure, France

12:15 General discussion

10:45 Scott M. ROBESON, Indiana University, USA

10:57 H. Y. TSENG, ANWS OAA, R.O.C.

11:09 Ole Einar TVEITO, Norwegian Meteorological Institute, Norway

11:21 Brandon WHITCHER, University of Washington, USA

11:33 Igor ZVERYAEV, Russian Academy of Science, Russia

11:45 P. ALLERUP and H. MADSEN, Danish Meteorological Institute, Denmark

Special Invited Lecture Sea Ballroom 14:00 - 15:15

14:00 David R. BRILLINGER, University of California at Berkeley, USA

14:45 Discussants: Andy Solow, Woods Hole Oceanographic Institute, USA

Peter Guttorp, University of Washington, USA

Invited Session: Multivariate Methods Sea Ballroom 15:45 - 17:30

15:45 Christopher S. BRETHERTON, University of Washington, USA

16:45 Hans von STORCH, GKSS, Germany

17:15 General Discussion

Contributed Session: Neural Nets McGuire A 15:45 -16:21

15:45 William W. HSIEH, University of British Columbia, Canada

15:57 Adam H. MONAHAN, University of British Columbia, Canada

16:09 Andreas WALTER, J.W. Goethe University of Frankfurt, Germany

Contributed Session: Spatial Statistics McGuire A 16:21 - 17:33

16:21 Gerard BIAU, Université Montpellier II, France

16:33 Miroslaw MIETUS, Institute of Meteorology and Water Management, Poland

16:45 David NOTT, University of New South Wales, Australia

16:57 Sam SHEN, University of Alberta, Canada

17:09 Thomas M. SMITH, NOAA, USA

17:21 D. M. SONECHKIN, HMC, Russia

15:45 Ulrich CALLIES, GKSS Research Centre, Germany

15:57 Radan HUTH, Institute of Atmospheric Physics, Czech Republic

16:09 William H. KLEIN, Bethesda, USA

16:21 Ricardo M. TRIGO, University of East Anglia, UK

16:33 Robert L. WILBY, NCAR, USA

16:45 Eduardo ZORITA, GKSS Research Centre, Germany

16:57 Anatoly CHAVRO, Russian Academy of Sciences, Russia

Special Invited Lecture: Sea Ballroom 17:45 -19:00

17:45 Jin-Song von STORCH, University of Hamburg, Germany

18:30 Discussants: Myles Allen, Rutherford Laboratories, UK

David Brillinger, University of California at Berkeley, USA

8:45 Anthony G. BARNSTON, NOAA, USA

9:30 Discussants: Chris Bretherton, University of Washington, USA

Hans Wackernagel, Centre de Géostatistique, France

Invited Session: Short Term Forecasting Sea Ballroom 10:30 -12:15

10:30 Judit BARTHOLY, Eötvos Lorand University, Hungary

11:00 Neville NICHOLLS, Bureau of Meteorology, Australia

11:30 Andrew SOLOW, Woods Hole Oceanographic Institute, USA

12:00 General discussion

10:30 Gil H. ROSS, UK Meteorological Office, UK

10:50 Radan HUTH, Institute for Atmospheric Physics, Czech Republic

11:20 Kayo IDE, University of California at Los Angeles, USA

11:45 Ian JOLLIFFE, University of Aberdeen, UK

Cluster analysis: some recent developments and their relevance to climatology

10:30 Anne-Catherine FAVRE, Swiss Federal Institute of Technology, Switzerland

10:42 Paul NORTHROP, University College London, UK

10:54 Nora E. RUIZ, University of Buenos Aires, Argentina

Some aspects of 200 hPa circulation over Argentina in relation to precipitation

11:06 Martin WIDMANN, University of Washington, USA

11:18 D. S. WILKS, Cornell University, USA

Low-frequency characteristics of several stochastic daily precipitation models

11:30 Abebe YESHANEV, National Meteorological Services Agency, Ethiopia

11:42 Abdesselam ZAROUGUI, Universidad Complutense de Madrid, Spain

Regionalization of daily rainfall in Morocco.

10:54 S. D. DAHALE, Indian Institute of Tropical Meteorology, India