Determination of beach sand parameters using remotely sensed aircraft reflectance data

Abstract

An algorithm was developed which determines the mineralogy, moisture, and grain size of beach sands based on the hemispherical reflectance in 17 discrete spectral bands. The bands chosen range between 0.40 and 2.5 [mu]m, a wavelength range practical for existing multispectral remote sensing technology. The sand spectra on which the mineralogy, moisture, and grain-size algorithm (MOGS) is based were obtained from laboratory spectrophotometric measurements. Selected spectral bands are used in a vector-length-decision framework to determine the mineralogical class of the input sand. Multiple linear regressions are then used, within a given mineralogical class, to determine the moisture and grain size of the sand. The predictive results of the MOGS algorithm are very encouraging. When tested on 70 of the sand reflectance spectra from which it was derived, the correlation of actual to predicted moisture and grain size was 96% and 88%, respectively. The MOGS algorithm has been successfully tested using aircraft multispectral scanner data collected over the Lake Michigan shoreline. The algorithm correctly identified gross mineralogy and predicted grain size to within 0.09 mm of measured values. Some difficulties were encountered in predicting high beach-sand moistures, probably due to the increasing non-Lambertian nature of sand as the moisture content of the sand increased.