2014 GSA Annual Meeting in Vancouver, British Columbia (19–22 October 2014)

Paper No. 117-7
Presentation Time: 10:30 AM

UTILIZING HYPERSPECTRAL REFLECTANCE TO ANALYZE SAND COMPOSITION


SMITH, Molly Elizabeth and SELCH, Donna E., Geosciences, Florida Atlantic University, 777 Glades Rd., Boca Raton, FL 33431

Spectral signatures quickly and objectively aid the analysis of sand composition because specific wavelengths correspond with distinct grains. This provides a tool that gives additional objectivity to traditional microscopic methods, with the option to create a custom spectral library for Hyperspectral Remote Sensing (HRS) applications. Removal of salt (as a precipitated solid from sea water) is a useful practice for clearer microscopic viewing of sand because siliciclastic and calcitic grains are less-likely to be misidentified as precipitated solids. Though removal of precipitated solids aids in qualitative visual identification, it is problematic for studies requiring spectral reflectance data to match real-life conditions. Sand samples collected from the nearshore zone contain a precipitated solid component which is lost if laboratory preparations include rinsing and drying of samples for microscopic analysis.

Spectroradiometric techniques were used to assess the effects of precipitated solids in spectral signatures of sand. Sand samples of mixed siliciclastic-carbonate composition were collected from 15 locations across the Southeastern Florida coast. Spectral plots were generated from data collected by an ASD Spectroradiometer in a laboratory setting. Spectral data was collected from the samples before they were prepared for microscopic study, and again after sample preparation. Laboratory-prepared samples show negative slope at approximately 1500 nm and 2000 nm ranges on the generated plots. These wavelengths are indicative of grains that have either predominately carbonate (~1500 nm) or siliciclastic (~2000 nm) compositions, which agrees with composition determined in microscopic analysis. Particulates present in a sample will affect the spectral signature, thus particulate removal yields spectral plots not necessarily concurrent with plots generated from the raw, unprepared samples. For studies utilizing airborne HRS data, the order in which data is collected and prepared is important. To ensure a more precise match between the spectral library and the hyperspectral imagery, spectral data must be collected before the sample is prepared for microscopic analysis.