We've Moved!
Visit SDSU’s new digital collections website at https://digitalcollections.sdsu.edu
Description
The United States -- Canada border is one of the longest borders in the world, spanning approximately 8900 kilometers (including Alaska). Currently, the United States Border Patrol utilizes various technologies for patrolling the border, but it is not practical to closely monitor the entire area. Knowledge of the physical geographic landscapes of the border can be useful in deciding which patrolling methods and technologies to implement, and a model that displays areas with greater need for monitoring than others would be useful in the decision making process of efficiently allocating monitoring resources. The focus of this study was to divide the border into homogeneous segments using geographic information and geocomputational methods and to assign porosity values to these segments. Porosity values in this research refer to the level of difficulty for a pedestrian to cross any given zone (i.e. the higher the porosity level, the easier it is for a pedestrian to cross that zone). Two unsupervised classification methods, the fuzzy K-means clustering and the Self-Organizing Map, were used to segment the border into homogeneous zones according to topographic attributes including land cover, elevation, and slope. Repeating the classification for different spatial resolutions of 1.5 km, 3 km, and 6 km showed that the sizes and locations of the segments vary according to spatial resolution and classification method. Once the segments were created, surveys were sent out to determine porosity values for each attribute, and scripts were developed in Python language to integrate Monte Carlo simulation in order to assign a porosity value for each segment. The porosity value pattern was almost the same throughout the border for all three spatial resolutions of the fuzzy K-means clustering segmentation. Such pattern was not visible between the different spatial resolutions of the Self-Organizing Map segmentations. This suggests that the fuzzy K-means clustering classification is less sensitive to changes in spatial resolution than the classification performed by the Self-Organizing Map. In addition, the results of the sensitivity analysis suggest that land cover may be the most influential but that all three attributes have significant impact in determining porosity values. The information obtained from this research provides insight into the geographical characteristics and porosity of the US -- Canada border. Having increased knowledge of the United States border will allow the nation to not only create effective border security policies but can also contribute to promoting smooth trade relations with Canada. Furthermore, the methods developed in this research can be used for other border studies, such as for tourism.