Automated Analysis of Spatial Grids: Motivation and Challenges

Abstract

Geography is often the only feasible way to tie together disparate data sets into something that can be analyzed together. Geographic information systems (GIS) are software systems that are capable of storing and carrying out spatial operations – operations that make use of geographic coordinates – on spatial grids. However, a GIS is typically employed interactively. Remote sensing offers substantial benefits in observing the environment, but the resulting spatial grids can be difficult to analyze interactively and routinely. If your objective is to deal with dynamic data, or large amounts of data, human interaction does not scale and you might want to consider analyzing the spatial data automatically. Creating an automated algorithm is difficult because interactive processing can build on the amazing capabilities of the human visual system, whereas automated processing has to explicitly encode every relationship. Another challenge with creating automated algorithms to analyze spatial grids is that low-level image processing operations are rarely sufficient, so domain knowledge needs to be incorporated. Hence, in order to create an automated algorithm to operate on geospatial data, it is often necessary to write it – off-the-shelf, general-purpose solutions will rarely suffice. The goal of this book is to give you the ability to do just that.