Point Pattern Analysis of San Diego Search and Rescue Event Data
Mark P. Kumler, Ph.D.
Sally Westmoreland, Ph.D.
Lauren Scott, Ph.D.
The US Coast Guard, Sector San Diego, performs search and rescue operations seaward of the southwest coast of California, as far north as San Mateo Point, and as far south as the Cabo San Lucas, Baja California. Summary tracking data since 1998 has been used to evaluate station statistics, but never for spatial analysis. Spatial analysis may reveal patterns in nominal search and rescue events that may help Sector San Diego advance its life-saving mission. Cluster analysis of event data is problematic: point events may lack a weight or value which can be used. Preliminary analysis using point densities and nearest neighbor analysis reveals obvious clusters. However, first-order effects may hide the presence of weaker spatial influences. To identify weaker effects, a quadrat grid is used, and the ratio of analysis events to total events provides a weight value. By evaluating multiple iterations of Moran’s I statistic at various distance bands, we can identify distance bands where the clustering is most evident. Using the peak clustering distances obtained from Moran's I, Getis-Ord’s Gi* statistic is used to locate clusters where certain events make up a ratio significantly higher or lower than the mean ratio. In this way, areas where unusual proportions of events occur can be determined.
Ford, J. D. (2005). Point Pattern Analysis of San Diego Search and Rescue Event Data (Master's thesis, University of Redlands). Retrieved from http://inspire.redlands.edu/gis_gradproj/55
Full text is available at the University of Redlands