Visualizations in Spatial Algorithm Development

Spatial algorithms as used in Geographic Information Systems (GIS) can be difficult to understand and use, both for the developers and users. Knowledge transfer between developers and GIS enabled application users is often inadequate, incomprehensible or non-existent. Novel approaches for spatially indexing and searching data involve trade-offs; all have their limitations and advantages. Effectively communicating these trade-offs is a challenge. Both the limitations and strong points of any algorithms used in scientific applications must be explained to end-users in an easily understood and digestible manner. Written documentation is only one way of describing an algorithm. Images, animations, and interactive demos have long been used to aid in understanding spatial algorithms but their adoption and use could be increased. This session demonstrates how to include interactive visualizations from a project's inception and outlines the possibility of using these visualizations not only as eventual documentation, but also as verification criteria for spatial algorithm development. Interactive tools, such as Google Earth, can be used to create and visualize inputs to spatial algorithms and validate results. During development, a developer can benefit from constant feedback and the ability to quickly test changes and new code. This session will also demonstrate methods of documenting spatial algorithms for end users. The use of literate programming tools such as docco, http:// jashkenas.github.com/docco/, and spatial visualizations document the code and aid in producing documentation for scientists and developers.