Temporal GIS and Space/Time Geostatistics for the Environment and Public Health
(Short title: Temporal GIS and Geostatistics)
Fall 2018, 3 semester hours, Tuesday Thursday 09:30AM-10:45AM
Lecture Location: McGavran-Greenberg 2304
Computer Lab Location: Health Sciences Library computer lab rooms 307
Instructor Office hours (MHRC 1303): Any time upon request to Instructor
Instructor: Marc Serre
The course focuses on the development of environmental Geostatistics and its application in temporal Geographical Information Systems (TGIS). TGIS describe environmental, epidemiological, economic, and social phenomena distributed across space and time. The course introduces the arcGIS software to query and manipulate geographic data, it provides the concepts and mathematical framework of space/time Geostatistics necessary to map environmental contaminants across space and time, and it leads to a real-world TGIS project where students analyze their own data, following a comprehensive example using EPA and USGS freshwater contaminant data across an entire state (e.g. physical, microbial and biological environmental water quality contaminants in New Jersey and North Carolina), as well as an example using an infectious disease (e.g. STDs and HIV prevalence across North Carolina).
The course starts with a 4 to 5 weeks review of basic GIS consisting in intensive computer labs on the ESRI ArcGIS software. Prior knowledge of GIS is highly recommended, but not required. Lessons from these ArcGIS computer labs is tested in a homework where students research and display maps of their own space/time environmental data using basic ArcGIS functions (see Graph 1). In the remainder of the course we then switch to using the BMEGUI software developed especially for this course for the advanced visualization and spatiotemporal geostatistical analysis of space/time data. The use of BMEGUI is tested in the remaining homework of the course, and culminates in an individual final project.
The concepts and mathematical formulation of spatiotemporal Geostatistics are progressively introduced throughout the course. We start with the concept of space/time distance. We then rapidly review multivariate calculus (derivatives and integrals) and basic statistics (probability density function, or pdf, and expected value) of random variables. Multivariate calculus is a pre-requirement for this course, and prior introductory statistics or probability courses are recommended, but not required. Using this foundation in multivariate calculus and basic statistics, we then cover the theory of spatiotemporal Geostatistics, which include 1) bivariate pdf and conditional probabilities, 2) variability in space and time and covariance function, 3) spatial and spatiotemporal random fields and 4) spatiotemporal estimation and uncertainty assessment. The concepts of the Bayesian Maximum Entropy (BME) method is presented, which provides a powerful framework for space/time mapping, and leads to the classical kriging methods as special cases.
The application consists of a real-world mapping TGIS project. Using skills acquired in basic GIS (i.e. arcGIS), and in advanced TGIS (i.e. BMEGUI) each students research a space/time dataset of concern for society, s/he formulates the space/time mapping problem, and s/he uses concepts and mathematical tools together with the BME method of space/time Geostatistics to provide a realistic representation of the field over space and time.
George Christakos, Patrick Bogaert, and Marc Serre (2002) Temporal GIS: Advanced Functions for Field-Based Applications, Springer-Verlag, New York, N.Y., 250 p., CD ROM included
The prerequisite for this class is Calculus of Functions of One Variable I & II (MATH 231 & 232) and preferably a multivariate calculus course like MATH 233. An introductory course in Statistics or Probability is useful, but not required. Additionally, knowledge of GIS (from beginner to expert) is highly recommended, but not required.
Philosophy of Grading and Course Evaluation:
The students should learn the concepts, and not use the tools as a black box. They will be graded on solving conceptual problems rather than just applying the programs. The students are expected to promptly do their homework, a class project, and fill out the course evaluation at the end of the semester. The grading will be as follow
Student-defined project 50%
Filling out course evaluation: 1 point bonus
The homework and class project are open book. However each student is expected to do their homework and class project on their own, without outside help nor help from classmates. Classmates can form groups to study together and discuss problems together, but each student is expected to arrive at homework solutions on their own. A cluster of students providing the same erroneous solution should therefore be a rare event occurring by chance only. Likewise each project should be original compared to that of others.