Remote sensing based improvement of the geological map of the Neoproterozoic Ras Gharib segment in the Eastern Desert (NE-Egypt) using texture features

Geological mapping in the Eastern Desert is impeded by difficult accessibility. We improve the existing geological maps by including texture features in a classification scheme of ASTER and Landsat 8 data. We tested the improvement of support vector machine classification using band ratios, principal component analysis (PCA) and texture analysis in the Ras Gharib segment (NE Egypt). A very high classification overall accuracy of 99.85% was achieved. We demonstrate that the input of textures provide valuable additional data for lithological mapping. With the gained information, the existing geological map of the study area was improved distinctly in precision and resolution, but also in terms of correction of yet wrong or inaccurate locations and of lithological unit extents.