Abstract

Visible and near-infrared (VNIR) derivative spectroscopy of diffuse spectral reflectance (DSR) data can be a potent method to extract lithological information from sediment cores. However, synthesis of multiple DSR data sets collected with different instruments from sediment obtained at different times could subject the DSR measurements to errors arising from disparity in depositional environment, sample processing methods, and storage conditions. Here we apply a quotient normalization technique to set these data in a common reference frame as a first step in our VNIR derivative spectroscopic analysis. The effectiveness of the quotient normalization technique is illustrated with samples from the Bering and west Arctic Sea shelves; this is because of access to sediment samples collected by several coring expeditions at different times in the past over a range of locations.

DSR measurements were obtained from eleven groups of core samples processed under two conditions. Under one of the conditions, they were quotient normalized, while in the other, they were not. Lithological proxies in these cores, from both conditions, were extracted using varimax-rotated, principal-component analysis (VPCA). These lithologies are chlorite + muscovite, goethite + phycoerythrin + phycocyanin, smectite, calcite + dolomite, and illite + chlorophyll a. These lithological proxies were then plotted spatially using GIS kriging software.

The spatial distributions of the VPCA extracted lithologies, under the two conditions, were compared with lithologies obtained by previous workers in the study site. The lithologies after quotient normalization were found to be more consistent with previously published results for clay mineralogy determined by X-ray diffraction (XRD). The quotient normalized data also showed a lower variance in the interpolated data sets relative to the second processing condition, which is the norm, used for comparison. We conclude that the quotient normalization technique is an effective scaling tool for minimizing errors from combining DSR data sets collected during multiple coring expeditions.