Error Propagation Modeling in Raster GIS: Adding and Ratioing Operations

The purpose of this paper is to identify error properties arising when source maps that individually contain error are added or when the ratio of one map with respect to another is computed. The research approach to the problem combines mathematical analysis and simulation where source maps and error processes have been constructed with specified properties. Geman and Geman's corruption model is used to represent error in individual source maps. The paper reports spatial and aspatial error properties arising from adding and ratioing error-corrupted maps. These are identified as functions of the true characteristics of the individual source maps and the errors inherent within them; the relative contribution of these two components to the errors in maps is quantified by regression (for addition) and ANOVA (for ratioing). The paper considers the broader usefulness of this type of experimental analysis in using artificially constructed maps in geographic information science.