In this team project, we investigate the influence of the model used to store polygons on the runtimes of polygon overlay operations.

In the first model, only standard data types are used and the polygons are stored in specially designed relations in a relational database (DB2). To compute polygon intersections a single very complex query, formulated in core-SQL, is employed (see [TN10]). To improve the runtime an appropriate set of standard indexes (b-trees, hash) should be created.

The second modelling approach is a spatial extension of a relational database system, i.e. the data type polygon and the intersection operator of DB2 Spatial Extender are used. In addition, a spatial index should be applied.