Spatial pattern comparison metrics for detecting environmental change

点击翻译按钮获取中文摘要

A recent report by the United Nations Initiative on Global Geospatial Information Management highlighted exponential growth in the creation of geographic data as a key trend that is expected to continue over the next 5-10 years. Coupled with this is increasing transformation of the natural world, through both natural and anthropogenic processes. There is therefore expanding need for human and technical capacity to manage, process, and analyze new sources of geospatial information about the earth's surface, such as high-resolution satellite imagery. A key task in handling geographic information is the ability to make inferences about spatial patterns, and increasingly, to compare spatial patterns. This might be in evaluating the fit of a spatially explicit model about the spread of an invasive species, or evaluating whether a landscape has lost significant connectivity after resource extraction has occured. While methods for detecting patterns in spatial data have blossomed, there are decidedly few tools available for comparing spatial patterns about the environment. Given the vast landscape of geographical information sources, there is a growing need for pattern comparison algorithms. ******I propose a research program that directly confronts this challenge through the discovery of metrics and algorithms that support the comparison of spatial patterns. I aim to develop these approaches for a variety of spatial representations, and include both static and dynamic spatial patterns. Streaming spatial data sources are dynamic in nature, and pose a significant research challenge for pattern analysis and interpretation. I will develop new methods robust to data velocity, volume, and nature. Case studies will be used to demonstrate and evaluate the metrics for environmental change detection. As part of my research program, I will train high quality personnel who can meet the industry's growing demand for spatial data scientists