The design and implementation of a two and three-dimensional triangular irregular network based GIS

Abstract

It has long been realised in the GIS community that most 2D GISs are capable of handling 2D spatial data efficiently, but systems have had less success with 3D spatial data. This is reflected in the current GIS market place where systems which can handle 3D data are hardly available - due to several impediments in implementing suchsystems. This thesis attempts to address some of the impediments. The impediments which relate to spatial data especially data representation, data structuring and datamodelling using object-oriented (OO) techniques are the foci of this thesis. OO techniques are utilized because they offer several advantages over the traditional (i.e.structural) techniques in software development. In the aspect of spatial representation, several major representations are investigated, which then lead to identifying anappropriate representation both for 2D and 3D data, that is triangular irregular network (TIN) data structures. 2D data is represented by a 2D TIN, and 3D data is represented by a 3D TIN (also called a tetrahedral network or TEN). Several algorithms were developed for the construction of the data structures where procedures such as distance transformation (DT) and Voronoi tessellations were utilized. Besides standard Delaunay triangulations, constrained triangulations were also developed, thus the inclusion of real world objects in the spatial data modelling can be facilitated. Four classes of real world objects are identified (i.e., point, line, surface, and solid objects). For the purpose Abstract ii of spatial data modelling of the four types of objects, a formal data structure (FDS) is utilized. An OO database development is also investigated. This is done via a commercial system called POET OO DBMS where spatial data query and retrieval can be performed. Further, two application programs are developed, namely contouring and volume computation. All the developed algorithms and methods were tested using real data sets. To facilitate the output of the developed methods and algorithms, software called TinSoft, which is a Windows-based Multiple Document Interface software was developed in this research.