| Description |
xi, 93 leaves : illustrations, maps ; 28 cm |
| Summary |
"This work quantitatively evaluates the effectiveness of multi-spectral remote sensing data for geological mapping in arid regions. For this, Landsat Thematic Mapper (TM) data covering part of the Neoproterozoic Arabian Shield around the Makkah region in Saudi Arabia are used. The Makkah region is dominated by a variety of layered and intrusive rocks covered by unconsolidated sediments. The Landsat TM data have six spectral bands in the visible and near infrared (VNIR) and shortwave infrared (SWIR) with thirty meters spatial resolution. The Optimum Index Factor (OIF) has been computed to determine the best Red-Green-Blue (RGB) band combination emerging from the six spectral bands, the six corresponding Principal Components (PCs), and selected band-ratio images. Results of the OIF analysis showed that the RGB color combination 3-5-7 is the best among all twenty possible RGB color combinations obtained from the six Landsat TM bands. Also, the OIF analysis pointed to that the PC RGB color combination 2-4-5 have the highest spectral information among all twenty RGB color combinations obtained from the six PCs. As well, the modified band-ratio RGB color combination 5/7-5/4-3/1 has the highest OIF compared to other band-ratio Landsat TM images that have been previously used for lithological mapping in arid regions. Subsequently, in order to quantify the effectiveness of the Landsat TM data for lithological mapping, the Maximum Likelihood supervised classification is implemented. Results of the classification are evaluated in relation to previously published geological map using the Error Matrix and Kappa hat image classification accuracy assessment methods. These results show variation in accuracy between different lithological units, with an overall accuracy of 66.25% and Kappa hat of 56.98%. Part of this error is attributed to the presence of the unconsolidated sediments which are highly heterogeneous"--Abstract, leaf iii. |
|
