Lithologic characterization of desertic regions using remote sensing techniques: An example in the central coast of Baja California, Mexico

Abstract

The use of band combinations for false color composites, principal component analysis, photo interpretation, and supervised classification techniques of a Landsat Thematic Mapper (TM) image enabled the lithologic characterization of a vast desertic region along the east-central coast of Baja California, México (~100 x 15 km). This region has a diversity of metasedimentary rocks, granitic intrusives, felsic to intermediate volcanic rocks, and continental and marine sedimentary deposits.

The TM 742 band combination on a RGB false color composite was the most revealing for lithologic discrimination of the main Neogene volcanic, and Quaternary sedimentary units overlying the metamorphic and granitic basement. An enhanced contrast between lithologic units was produced by the selective principal component analysis using the second components for the TM band pairs 5-7, 4-7 and 2-7 on a RGB false color composite. Through the reassignment and generalization of a supervised classification, we obtained 26 lithologic classes; thirteen Cenozoic volcanic rocks and intrusives, three pre-Cretaceous metamorphic units, and ten Plio-Quaternary sedimentary classes. The latter differ among each other by the source rocks and relative age.

The resulting lithologic map of the supervised classification was evaluated using the confusion matrix and Kappa index (k). The map obtained a 79% confidence level, and k = 0.733, indicating that this map meets the evaluation criteria, and allow to us to recommend this methodology for geologic studies in desert areas.