Spatial Data Models

Abstract

A geographic space must be partitioned into a finite number of discrete pieces to accommodate a finite and discrete computing environment. Spatial data models describe the design of the discretization. Vector and raster data models are two commonly used spatial data models. Vector data models partition the space according to the form of spatial features, such as points, lines, polygons, and their derivatives. Raster data models, on the other hand, partition the space into regularly arranged cells to approximate spatial features. The store and management of spatial data are often separated from the phenomena that the data represent. Choosing appropriate data models for an intended study is critical, because it could affect the subsequent selection of methods for modeling and analysis, as well as the outcome. Spatial phenomena can be perceived as spatial objects, spatial regions, and fields. The spatial scale, the presence of boundary, the spatial variation and dynamics of attributes and processes, and mobility distinguish between the three types. Both vector and raster data models can be used to represent spatial objects, spatial regions, or fields, but to various degrees. The appropriate choice of a spatial model depends on whether a spatial data model is conceptually compatible with the perceived phenomenon.