Evaluation of Pixel and Sub-pixel Classifications of ASTER Data for the Spatial Distribution Detection of Clay Minerals (Case-study: Esteghlal Mine, Abadeh)

Esteghlal fireclay mine is located in the Central Iranian geological zone, 10 km to the north east of the Abadeh, northern Fars province. With annual production of over million tons of kaolinite and pyrophilite, Esteghlal is one of the largest sedimentary deposits in Iran. Sub-pixel Mixture Tuned Matched Filtering (MTMF) method and whole-pixel Spectral Angle Mapper (SAM) algorithm, with the objective of detecting spatial distribution of clay minerals were applied on the visible and nearinfrared and shortwave infrared (VNIR + SWIR) datasets of the ASTER sensor using the spectra of images as well as the spectra from USGS spectral library. In order to extracting the end-members, Sequential Maximum Angle Convex Cone (SMACC) and Pixel Purity Index (PPI) methods were implemented on the VNIR + SWIR datasets. Comparison of pixel and sub-pixel classification results on the basis of XRD and spectral analyses of field samples, showed that MTMF method is more reliable in determining the spatial distribution of clay minerals at the studied area. Sub-pixel frequency assessment of the ASTER data showed that the pixels with higher than 50 percent of relative fractions of kaolinite and pyrophilite outcropped in two different geomorphological units of the Esteghlal mine.