You are here

Dylan Keon's Ph.D. 2012

Automated Web-based Analysis and Visualization of Spatiotemporal Data

Doctor of Philosophy (Ph.D.), Geography, Oregon State University, Fall 2012
Graduate Certificate in Geographic Information Science

Graduate committee: D. Wright, A. Nolin, C. Daly, M. Bailey, R. Colwell

Dylan Keon
College of Earth, Ocean, and Atmospheric Sciences, Oregon State Univ
Corvallis, OR 97331-5503 | NACSE

Most data are associated with a place, and many are also associated with a moment in time, a time interval, or another linked temporal component. Spatiotemporal data (i.e., data with elements of both space and time) can be used to assess movement or change over time in a particular location, an approach that is useful across many disciplines. However, spatiotemporal data structures can be quite complex, and the datasets very large. Although GIS software programs are capable of processing and analyzing spatial information, most contain no (or minimal) features for handling temporal information and have limited capability to deal with large, complex multidimensional spatiotemporal data. A related problem is how to best represent spatiotemporal data to support efficient processing, analysis, and visualization.

In the era of “big data,” efficient methods for analyzing and visualizing large quantities of spatiotemporal data have become increasingly necessary. Automated processing approaches, when made scalable and generalizable, can result in much greater efficiency in spatiotemporal data analysis. The growing popularity of web services and server-side processing methods can be leveraged to create systems for processing spatiotemporal data on the server, with delivery of output products to the client. In many cases, the client can be a standard web browser, providing a common platform from which users can interact with complex server-side processing systems to produce specific output data and visualizations. The rise of complex JavaScript libraries for creating interactive client-side tools has enabled the development of rich internet applications that provide interactive data exploration capabilities and an enhanced user experience within the web browser.

This dissertation examines the automated web-based analysis and visualization of spatiotemporal data in the context of three distinct projects. Although particular methods were developed to solve the stated problems for each project, in most cases those methods can be generalized to other disciplines or computational domains where similar problem sets exist. Chapter 2 (to submitted for publication in the journal Transactions in GIS) describes methods of dynamically selecting and preparing data for tsunami modeling, and processing the resulting time-series output data. Chapter 3 (to be submitted for publication in the International Journal of Geographical Information Science) describes simulation modeling of potential human evacuation response to a modeled tsunami inundation event, with methods for the web-based definition of a simulation scenario and animated, interactive visualization of the simulation output. Chapter 4 (to be submitted for publication in the journal Computers & Geosciences) describes methods for web-based calculation and visualization of climate grid statistics over varying spatial and temporal scales, including methods for fast automated server-side grid processing.

Download Dissertation (8.0 Mb PDF file)
Also available in the ScholarsArchive@OSU permanent collection

Dissertation Defense (6.8 Mb PDF file)
Dissertation Defense Media: Project 1 Time Series Visualization | Project 2 Tsunami Evacuation Simulation | Project 3 GridStats Demo
Software Code on GitHub

Go to DJL to Voyages...