Post glacial mapping of holocene basaltic lavas in Iceland through the integrated proximal and remote sensing approach

In this paper we present an experimental methodology that originates from the need to apply common remote and proximal sensing techniques to areas were mainly basaltic rocks are present (Iceland). A systematic detailed mapping of Icelandic post glacial lavas is currently still missing: it is difficult to discriminate very similar basaltic units through classical geological field survey and petrology/chemical analyses. An interdisciplinary approach for characterizing typical features of geological units, such as morphology, vegetation cover distribution and other distinctive aspects, can be the key for understanding the geological asset of the area. Remotely sensed images, purposely processed, can allow to detect different volcanic unit especially if images interpretation can be supported by field spectro-radiometric data.For this reasons, we used a remote and proximal sensing original approach to map the recent (<8000 yBP) basaltic lava outcropping in the Northern part of Thjorsa valley (~20x20 km area). We collected field spectra with a portable spectroradiometer, slowly walking on the studied geological units. Then, we averaged the field-collected areal spectra and used them as reference for processing Landsat 8 multispectral images.Atmospheric correction, enhancing techniques and unsupervised classifications were applied in order to understand the lithological framework of the scene. Finally, a Spectral Angle Mapper classification was applied to obtain a 'distribution' map of the different geological units. The classification proved to be an important instrument for reviewing, updating but also producing a geological map for the area, as it has been possible to discriminate really subtle differences in basalts (whose spectral responses are very low and similar).