A methodological framework focused on integrating GIS and BBN data for probabilistic map algebra analysis
John Derek Morgan, M. W. Hutchins, Julie Fox, Katherine Rogers · 2012
The raster cell-by-cell comparison technique commonly referred to as map algebra has become a standard for Geographic Information Systems (GIS) spatial analysis and modelling (Bolstad 2012). Bayesian Belief Networks (BBNs) provide a graphical (and automated) way to display and interact with probabilities of related event information (Pearl 1988). Integrating spatial data with BBNs, at the resolution of pixels, allows for the opportunity to perform probabilistic map algebra (Taylor 2003; Ames & Anselmo 2008). Belief maps (or probability maps) can be created by transferring the results of probabilistic map algebra back into GIS. The methodological framework presented in this paper proposes a set of structured techniques for utilizing GIS and BBNs to inform spatial decision making and analysis.