Efficient geographic information systems: Data structures, Boolean operations and concurrency control

Geographic Information Systems (GIS) are crucial to the ability of govern mental agencies and business to record, manage and analyze geographic data efficiently. They provide methods of analysis and simulation on geographic data that were previously infeasible using traditional hardcopy maps. Creation of realistic 3-D sceneries by overlaying satellite imagery over digital elevation models (DEM) was not possible using paper maps. Determination of suitable areas for construction that would have the fewest environmental impacts once required manual tracing of different map sets on mylar sheets; now it can be done in real time by GIS. Geographic information processing has significant space and time require ments. This thesis concentrates on techniques which can make existing GIS more efficient by considering these issues: Data Structure, Boolean Operations on Geographic Data, Concurrency Control. Geographic data span multiple dimensions and consist of geometric shapes such as points, lines, and areas, which cannot be efficiently handled using a traditional one-dimensional data structure. We therefore first survey spatial data structures for geographic data and then show how a spatial data structure called an R-tree can be used to augment the performance of many existing GIS. Boolean operations on geographic data are fundamental to the spatial anal ysis common in geographic data processing. They allow the user to analyze geographic data by using operators such as AND, OR, NOT on geographic ob jects. An example of a boolean operation query would be, Find all regions that have low elevation AND soil type clay. Boolean operations require signif icant time to process. We present a generalized solution that could significantly improve the time performance of evaluating complex boolean operation queries. Concurrency control on spatial data structures for geographic data processing is becoming more critical as the size and resolution of geographic databases increase. We present algorithms to enable concurrent access to R-tree spatial data structures so that efficient sharing of geographic data can occur in a multi user GIS environment