Clay mineralogy of soils has important environmental and agricultural implications. We used Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS) data to map clay minerals in the Charleston area. AVIRIS is an optical sensor that provides hyperspectral images in 224 contiguous bands from 0.4-2.5 microns. The sensors were flown at 20 km altitude producing images with 20 m pixel resolution. We spectrally subset the data to concentrate on the SWIR region of the electromagnetic spectrum (2.1-2.4 microns). We were particularly interested in spectral absorption features near 2.2 microns where the combination metal-OH- bend plus OH- stretch are diagnostic of phyllosilicate mineralogy.

Prior to image analysis, the AVIRIS data were corrected for atmospheric interference (mostly H2O vapor and CO2) using a Modtran 4+ based algorithm. Data were calibrated by multiplying a “difference spectrum” based on field spectra through the AVIRIS data. This resulted in a pronounced reduction in spectral noise. Further segregation of signal from noise required a two-cascaded Principal Components transformation, followed by selection of mapping endmembers after the purest pixels were identified. A Mixture Tuned Matched Filter algorithm allowed us to identify areas consisting of one endmember, only.

Two distinct soil endmembers were identified and mapped over an area of approximately 660 km2. Kaolinite is the dominant clay in both endmembers based on the presence of absorption features at 2.208 and 2.165 microns. One of the endmembers displays a very weak 2.165 feature, suggesting the presence of subordinate montmorillonite. Field spectra are being used to validate results.