A one day short course for exposure analysis scientists dealing with spatial and space/time data
October 14, 2007
Space/time exposure assessment and the integration of data from multiple sources have received increasing interests in recent international conferences (e.g. ISEA06, ISEE06, IAMG06 and many others). Indeed, as exposure assessment experts we are increasingly asked to characterize exposure across space and time in order to better assess the exposure of susceptible populations, to provide more accurate exposure information for environmental epidemiology studies, and to improve health risk assessment under conditions of uncertainty.
Traditionally much of the information available for space/time exposure assessment has consisted of sparse monitoring data, and a general lack of adequate mechanistic models (hydrologic, atmospheric, toxicokinetic, etc.). As a result, the field of exposure assessment has recently seen an increased interest in using a Geostatistical (i.e. spatial statistics based) approach to obtain an efficient and accurate estimator for exposure. However; there is a critical need in exposure assessment for space/time Geostatistics, rather than just spatial Geostatistics.
More recently many scientists have recognized the usefulness of integrating secondary data from multiple sources. Secondary data may include satellite information, land use models, environmental sensors, confidential data, data obtained at different observation scales, empirical secondary data, GPS exposure data, physical models, etc. Such secondary data is very valuable, however it may have high data uncertainty, i.e. it should be treated as soft data. Hence the second critical need in exposure assessment is to integrate soft data from multiple sources.
The Bayesian Maximum Entropy (BME) approach appears to be a potential candidate for achieving this task: it is especially designed for managing simultaneously space/time data of various nature and quality ("hard" and "soft" data, continuous or categorical). It relies on a two-steps procedure that first involves a Maximum Entropy step (the ME part of BME) to objectively obtain a prior distribution in accordance with the general knowledge at hand, and an operational Bayesian conditionalization step that updates this prior probability distribution function (pdf) with respect to the specific data collected on the study site. BME provides a flexible framework that accounts for the wide variety of knowledge bases available, and leads in general to the best non linear space/time estimator. Traditional kriging methods are naturally obtained as a limiting case for linear estimation.
The BME approach thus appears as a kind of new unifying theory, opening new perspectives for space/time exposure assessment. Traditional simple, ordinary and universal kriging methods are derived as limiting linear cases, while the more general BME framework provides a powerful non-linear estimator for space/time Geostatistics using both hard data and a variety of soft data.
The course is intended as a large audience introduction to the concepts driving the BME approach for space/time exposure assessment. The basic concepts are illustrated through a series of conceptual examples and real case studies. The course combines lecture sessions and interactive practical sessions. Comparisons with traditional geostatistical methods are encouraged and open discussions are expected. Each participant receives at set of lecture notes. While BME is developed for the space-time domain, the examples presented in this course are both for the purely spatial case, as well as the space/time case.
The theoretical parts of the course include:
The practical part of the course consists of real case studies including:
The exposure assessment case studies that have been conducted using the BMElib library of comprehensive computer programs (written in Matlab®). The case studies provide examples at forefront of modern space/time analysis dealing with subsurface, soil, surface water, and air toxic contaminants. The participants are shown the benefits of using this integrated toolbox for exploratory analysis of the data, modeling of spatial and space/time variability, spatial and space/time analysis and estimation, as well as graphical presentation of maps.
The course is given in English.
This course is intended as a large audience introduction to researchers and professionals involved in the analysis of spatial and space/time data sets for exposure assessment. In order to fully benefit from the whole course, participants should be knowledgeable about linear regression theory, and the concept of spatial auto-correlation.
Introductory to intermediate
At least one graduate level course in statistics or bio statistics is required, and some knowledge of geostatistics or GIS is helpful but not required.
Marc Serre (
Assistant Professor, Department of Environmental Sciences
and Engineering
Room 115 Rosenau Hall,
Phone Office: +1 (919) 966 7014
email: marc_serre@unc.edu
Brief biography.
Dr. Marc L. Serre has been serving
as an assistant professor in the Department of Environmental Sciences and
Engineering at the
Dr. Serre has 16 years of experience in environmental engineering, space/time geostatistics, exposure mapping, health risk assessment, and spatial epidemiology. He has worked as an environmental engineer in the consulting industry, he has developed Geographic Information System (GIS) software for environmental applications in the Computer Aided Design (CAD) industry, and over the past decade he has been part of an active group developing the Bayesian Maximum Entropy (BME) method of modern spatiotemporal Geostatistics and he has applied this method for the spatiotemporal modeling of environmental and health processes.
Dr. Serre is currently involved in the application of the BME method to model the distribution of environmental pollutants across space and time in the groundwater, surface water and atmosphere. He is also using BME for the space/time modeling of infectious diseases, as well as the health impact of environmental contamination. This work has been presented in two books he has co-authored on temporal GIS and on modeling infectious diseases, as well as several journal papers and talks at international conferences or as invited speaker and during numerous workshops on the analysis of space/time data.
This one day
short course will be offered on October 14, 2007, in
The course is organized by the International Society for Exposure Analysis as part of the ISEA 2007 conference.
To register to this short course, go to the website of the ISEA 2007, and click on the “Short Courses” link.
For more information on the registration fee, go to the website of the ISEA 2007, and click on the “Short Courses” link.
Detailed program
A detailed program of the one day short course is available here