Comparison of sub-pixel and per-pixel classification of ASTER data for determining the abundance of clay and carbonate of Mishan marly Formation

Mishan Formation with grey marl and sandy lime composition is one of the most important raw material for cement production. Remote sensing is approved as a low-cost and rapidly developing method in exploration of mineral resources. The sub-pixel mixture tuned matched filtering (MTMF) and the whole-pixel spectral angle mapper (SAM) algorithms were used on 6 and 9 bands of the ASTER data from Mishan type section and a part of outcrops of this formation in Jamileh village, northern Boushehr province, to determine the abundance of clay (illite) and carbonate (calcite and dolomite) minerals in marl. The image end members were extracted using the Sequential Maximum Angle Convex Cone algorithm and the Iterative Spectral Unmixing; and the abundances of clay and carbonate of marl were detected using the extracted end members. Although the output results of algorithms enhanced similar districts for the dominant minerals of marl, the sub-pixel outputs determined pixels with high abundances of carbonate, comparing to illite. The results of sub-pixel processing on 6-bands data set highly corresponded field observations, spectral characteristics, and the XRD results of samples, making this method as an efficient approach for exploration plans.