Detailed mapping of volcanoes provides key information for initial hazard assessments. By utilising remote sensing (RS) data a low-cost route to volcanic mapping in remote or inaccessible environments can be harnessed. However, despite the wealth of RS data, volcanic mapping still primarily relies on time consuming, manual analysis instead of automated mapping techniques.
This study presents a semi-automatic mapping technique of glaciovolcanoes based on their morphometric characteristics extracted from slope and profile curvature maps derived from a digital elevation model (20m/pixel). The detection of glaciovolcanic landforms was conducted in eCognition 8.8 software (Trimble), which provides a object-based image analysis modular programming environment, where hierarchical rule sets of customized algorithms is built . More than 600 rule sets were tested adjusting parameters and hierarchy in order to evaluate their sensitivity to the classification results. The accuracy of each classification result was evaluated by an error matrix, where the classification result was tested against a geological reference map. Overall very good results are obtained by simple segmentation and classification procedure on slope maps. The accuracy is most sensitive to the assigned slope values, but with slope values of 8-14, and scale parameters 3-18 the overall proportion of area correctly classified and correctly unclassified is > 90%. The classification results are exported as shapefiles and can easily be directly incorporated in classical mapping procedures in the GIS environment.
 Benz et al. (2004) ISPRS 58, 239-258.  Sæmundsson et al. (2010) Iceland GeoSurvey,1:100.000.