Scales and uncertainties in using models and GIS for volcano hazard prediction

Practical decision-making in civil protection based on predicting volcano hazards often involves using process models linked with Geographic Information Systems (GIS). Optimum use of these techniques for such decision-support requires careful and coordinated consideration of process, data and model scales and their related uncertainties. To avoid wasting resources and time on inappropriate data collection, improper model use, and resultant poor decision-making, there is a pressing need for a scientific and functional framework within which to examine implementation and use of geo-spatial assessment tools. To be useful for researchers and decision-makers, volcano hazard simulation approaches must consider the spatial and temporal variability in volcano processes and the data collected representing those. The successful application and implementation of a geo-spatial distributed volcano hazard model at variable scales requires explicit or implicit use of some form of scaling theory applied to the tasks of selection and transformation of appropriate data, and use of results. In general, there are five consecutive scaling steps that demonstrate how data and model scale, as well as the methods for information transformation between these, play key roles in controlling whether prediction results have been produced efficiently and are appropriate at the scale of interest for a civil protection manager's decision-making process. This new scaling theory can be used as a framework to construct practical procedures for applying GIS-Model-based volcano models that allow effective model application based on realistic data availability and environmental settings. (C) 2004 Elsevier B.V. All rights reserved.