MIDDLE THROUGH UPPER MIOCENE GEOCHEMICAL HISTORY OF THE WEDDELL SEA DETERMINED WITH AN XRF CORE SCANNER

Understanding the stability of Antarctic Ice Sheets is a key to understanding past climates and the response of ice-sheets to warming. Here we examine the middle through upper Miocene marine geochemical record in the Weddell Sea. Sediments were recovered at Ocean Drilling Program (ODP) Site 694 located in 4653 m of water in the Weddell Sea Abyssal Plain. Approximately 390 m of sediment was drilled, 109.5 m of which was recovered.

The Miocene sequence consists of six sedimentary facies: silt, laminated silt, diatom clayey mud/mudstone, diatom clay/claystone, diamictite, sandy mud and sand. These sedimentary facies are indicative of deposition via turbidity currents, ice-rafted detritus and hemipelagic settling.

XRF core scanning was chosen as an analytical method because it is nondestructive, relatively inexpensive, and provides high-resolution data. Data is expressed as elemental intensity. We used elemental ratios to identify areas of change. Potassium / Titanium (K/Ti) ratios were the most statistically significant among the ratios used. K/Ti ratios were overlain on core photos to show down-core changes. This ratio shows large differences between silts and muds. We are trying to determine if this relationship is grain-size or composition based.

We tried several methods to convert elemental intensity to concentration. The first conversion method used XRF sample measurements to generate oxide weight percent and compared these to scanner intensities. The samples were 1 to 2 cms long and wide. The sediment XRF analysis showed no correlation with core scanner intensities. This was followed by log-ratio transformation (compositional data analysis, after Weltje et al., 2008), which converted the intensities to element weight percent. These results also showed no correlation to the oxide weight percent. The elemental weight was then converted to oxide weight percent and normalized. We found that Fe₂O₃ was significantly higher and SiO₂ was significantly lower than the oxide weight percent of the XRF sample measurements. We are in the process of analyzing small samples located on the cores for specific K/Ti ratios. These samples were taken from the upper surface of the split cores because the XRF core-scanner penetrates only a few mm into the sample.