ABSTRACT
Discovery of Optimal Boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306 21.5 Pattern Summarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 21.6 Application on Vegetation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 21.7 Application on Presidential Election Data Analysis . . . . . . . . . . . . . 312 21.8 Application on Biodiversity Analysis of Bird Species . . . . . . . . . . . 313 21.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Geospatial data identifies the geographic location and characteristics of natural, constructed, or socially-based features. A set of geographically coregistered geospatial datasets captures various aspects of an environmental process, involving variables that are highly coupled through a complex chain of mutual interactions and feedback loops. The analysis of relationships among different variables is challenging due to inherent nonlinearity and spatial variability of such systems. Recent advances in data collecting techniques (for example, satellite-based remote sensing) result in the “data rich” setting and provide an opportunity for more thorough analysis. However, the full benefit of these enormous quantities of data can only be realized by automating the process of extracting relevant information and summarizing it in a fashion
that is comprehensible and meaningful to a domain expert. In this chapter, we introduce a framework of discovery and summarization of empirical knowledge contained in spatial patterns observed in geospatial data using a fusion of techniques, including association analysis, reinforcement learning, and similarity measurement.
