Proposal for the Introduction of the Spatial Perspective in the Application of Global Sensitivity Analysis

In any model, Sensitivity Analysis (SA) is a fundamental process to improve the robustness and credibility of the results, as part of validation procedure. Generally, SA determined how the variation in the model output can be apportioned to different sources of variations, and how the given model depends upon the information fed into it. Many complex tech- niques of SA have been developed within the field of numerical modeling; however, they have limited applications for spatial models, as they do not consider variations in the spatial distributions of the variables included. In this research, a variation in the implementation of a Global Sensitivity Analysis (E-FAST) is proposed in order to include the spatial level. For this purpose the conventional tools available in a raster Geographical Information System (GIS) are used. The procedure has been tested in a simulation of urban growth for the Madrid Region (Spain) based on Multi-Criteria Evaluation (MCE) techniques. The results suggest that the inclusion of the spatial perspective in the application of the SA is necessary, because it can modify the factors that have a decisive influence on the results.