Stephen L. Rathbun

Associate Professor of Statistics
Ph.D., Iowa State University, 1990

Summary of research interests

Dr. Rathbun's research focuses on the application and development of geospatial statistical methods in ecology and the environmental sciences. Emphasis is given to two types of geospatial data: geostatistical data and spatial point patterns.

Geostatistics involves data that may be collected at any sample of locations in a region of interest. Such data may include the concentrations of chemical contaminants in air, water, or soil samples, or counts of numbers of individual organisms or species at each site. Data from neighboring sites often take more similar values than data from disparate sites. This spatial dependence may be modeled through the variogram, a function of the distance between a pair of sites. Although the distance between a pair of sites is usually taken to be Euclidean distance (i.e., distance "as the crow flies"), distance "as the fish swims" may be more relevant for data collected in irregularly-shaped estuaries. Instruments used to assay chemical contaminants often have minimum detection limits (MDLs). Naïve approaches to handling such left-censored data can yield biased estimates of model parameters and biased predictions at unsampled sites. Dr. Rathbun's geostatistical research has focused on the use of alternative distance metrics and the geostatistical analysis of left-censored spatial data. Applications include data from estuaries, soil nemotodes, and the fresh water marshes of the Florida Everglades. Current research considers multivariate geostatistical methods for modeling the interactions among multiple environmental variables.

The spatial point pattern formed by the locations of individual organisms is the result of past patterns of birth, growth, and survivorship, and so, the investigation of spatial point patterns can lead to a better understanding of these important demographic processes. Dr. Rathbun's research has focused on the methods for modeling the effects of partially observed environmental variables (e.g., elevation, soil moisture, light, etc.) on spatial point patterns, and on the development of spatio-temporal point process models. Applications include the spatial point patterns of forest trees, turtle nests, and earthquake epicenters.

Representative publications

Rathbun, S.L., and Cressie, N. (1994). Space-time survival point processes: longleaf pines in southern Georgia. Journal of the American Statistical Association 89, 1164-1174.

Rathbun, S.L. (1996). Estimation of Poisson intensity using partially observed concomitant variables. Biometrics 52, 226-242.

Rathbun, S.L. (1996). Asymptotic properties of the maximum likelihood estimator for spatio-temporal point processes. Journal of Statistical Planning and Inference 51, 55-74.

Rathbun, S.L. (1998). Spatial modeling in irregularly shaped regions: Kriging estuaries. Environmetrics 9, 109-130.

Ettema, C.H., Rathbun, S.L., and Coleman, D.C. (2000). On spatiotemporal patchiness and coexistence of five species of Chronogaster (Nemotoda: Chronogasteridae) in a riprarian