CONICET | Buscador de Institutos y Recursos Humanos

Remote sensing data are particularly suitable for the geological mapping of inaccessible and/or large non-vegetated areas. Volcanic mapping in the Altiplano-Puna plateau of the Central Andes is a challenging task due to inaccessibility, extreme climatic conditions, high altitude above sea level and dangerous issues like minefields. In this paper we present a practical spectral mapping methodology for arid volcanic regions using optical imagery data, which it has been tested in the area of the Cerro Tuzgle volcano, a Quaternary composite volcano, because of its well-known stratigraphy. Some of the most common and well known processing techniques were used to multispectral data (Landsat 7 ETM+, Landsat 8 and ASTER). Then, supervised classification and an assessment of the classification were produced for every image created. The quantitative accuracy of the classification maps resulting from different data sets was assessed by comparing the classification with ground truth data (reference data) extracted from the geological map by means of a confusion matrix and related statistics. We found that supervised classification over Principal Component Analysis images was the best methodology for lithological mapping. Low costs, data availability and broad swath of the multispectral data make these images valuable for lithological mapping in arid regions. Even if this methodology is effective to composite volcano preliminary mapping, the correct reconstruction of the stratigraphy of a volcano always requires field survey.