Spatial Database Quality and the Potential Uncertainty Sources

Abstract

One of the most significant features of geo-information systems is the creation of geospatial analyses. The analyses are based on fundamental geospatial data which model the landscape in the certain territory of interest. The analyses themselves are often described by a mathematical apparatus which uses a wide range of branches of mathematics, especially vector analysis, differential geometry, statistics, probability, fuzzy logic, etc. The classical mathematical description of analysis is clear and precisely defined. Complex geospatial analyses, however, work above geospatial data that do not have to be homogeneous from the point of view of quality. With respect to the capacity and technological possibilities of the data supplier, the input data can have different level of geometric and thematic accuracy, their thematic attributes can remain unfulfilled or the data can be obsolete to a certain extent. Also the location of objects (e.g. forested areas, soil, water area, etc.) can be uncertain concerning the impossibility to define them accurately (e.g. areas of different soil kinds are blended together) or change with time (coast line of watercourse that changes depending on rainfall). The stated imprecision and uncertainty then influence the result of the complete geospatial analysis. This influence gets bigger with the number of input data objects.

The aim of the presented paper is to find a relation between the quality of input data and the reliability of the geospatial analysis result. The authors approach is based on mathematical models of analyses, models of vagueness and uncertainty, as well as from models for quality evaluation. In the research were used real data from the territory of the Czech Republic - current as well as historical - and specific methods of space evaluation used in decision-making processes of commanders and staff of the army.