Evaluation of JERS-1 (FUYO-1) OPS and Landsat TM images for mapping of gneissic rocks in arid areas

Systematic defects in images produced by the JERS-1 Optical Sensor's short wave infrared (SWIR) channels have deterred wide geological applications of the global data that JERS-1 has gathered. Frequency-domain filtering to separate noise from scene-determined signals allows the advantages of the three JERS-1 SWIR channels, covering Al-OH, Mg-OH and C-O spectral features to be assessed in relation to the widely used Landsat Thematic Mapper (TM) data. For the arid, high-grade Barka Terrane in NW Eritrea, two data-optimizing methods permit consistent and comprehensive lithofacies discrimination. These are based on: (i) pairwise principal component analysis that employs groups of two channels oriented to hydroxyl, carbonate and ferric iron species' spectral features; (ii) selection of channels to display RGB images. Optimizing this approach is based on detailed analysis of local spectroradiometry. As a general mapping tool for high-grade rocks the Landsat TM outperforms the JERS-1 Optical Sensor (OPS). However, recovery of OPS channel 8 by frequency-domain filtering uniquely permits routine identification of marbles and, together with TM band 7, allows distinction between rock surfaces whose spectra are dominated by Al-OH-and MG-OH-bearing phyllosilicates.