Developing High Performance GIS Simulation Models on Geospatial Cyberinfrastructure: A Case Study of Population Change Models with Grid Computing and Cloud Computing Technologies.

摘要

Many geography phenomena and geospatial problems are complex and dynamic. Traditional Geographic Information System (GIS) models have limitations to represent the detailed dynamic processes of geospatial phenomena in the real world. This dissertation introduces a new GIS model framework which can simulate complex population change by using large amounts of real-world data sets and high performance computing resources. This research developed parallel GIS simulation models using cellular automata (CA) and agent-based modeling (ABM) that have been very popular in urban geography. The case study of the parallel GIS simulation models in this dissertation focused on the population change in the County of San Diego, California. In particular, for the parallel ABM, this research modified the original Schelling model, a popular residential segregation model, to process over 2.8 million resident agents. With the high performance computing capability, the GIS models could simulate the dynamics of population change efficiently and help us understand the formation of residential segregation in the future. In addition, the case study demonstrated performance improvement, technical feasibility, and implementation challenges of the parallelization of GIS simulation models in CA and ABM. While high-performance computing is essential for many scientific research disciplines and the domain of computational science, very few geospatial scientists have utilized high-performance computing for the analysis of geographic problems. One of the important reasons is the lack of accessible high-performance computing resources. A new user-friendly web portal framework was designed and implemented to integrate distributed computing resources and to provide high-performance computing capabilities for GIS models. Grid computing and cloud computing resources were integrated into the web portal framework with accessible web-based user interfaces. Usability test was conducted at the end of this study to evaluate the accessibility and functions of the new web portal prototype in high-performance computing environments. The test results indicated that many users did not prefer the detailed configurations of high-performance computing resources. This study suggested that the detailed settings of high-performance computing environments should be concealed in the web-based user interface designed for scientists or geographers, who could focus more on geographic problems.