Techniques for the processing and interpretation of remotely sensed data have been widely used for various applications relating to surface and near surface environments (especially forest, crops, surface geological features, and oceans). This study applies these same techniques in an attempt to classify and quantify features which are subsurface in nature and have been gathered by current seismic techniques, and to see if the geology is evident in the texture of the seismic reflections. These techniques produce a series of images at various depths based on the reflections of the signals by the different rock layers. The strength of these reflections (their brightness values) can then be processed much in the same way as airborne radar type data, giving a more clear picture to interpreters with respect to the subsurface features. This study found that the selected features could be separated by the classifier very well with an average accuracy of 95% or greater for all feature classes with the training sites selected. The results for the test sites also achieved accuracy values of 85% or higher for the three test classifications. The separability of the signatures for the test and training sites were also found to be 1.99 or higher in average which is very good, the maximum separability being 2.0.
View full article as PDF (0.30 Mb)