Using Visible/Near-Infrared Spectroscopy to Identify Cryptotephra Layers

Friday, 19 December 2014
Molly C McCanta1, Bradley J Thomson2 and Elizabeth Fisher1, (1)Tufts University, Medford, MA, United States, (2)Boston University, Center for Remote Sensing, Boston, MA, United States
Continually accumulating marine sediments incorporate tephra layers within their depositional record that can be linked to individual explosive volcanic events. These layers can range from several meters in thickness, to discrete layers invisible to the naked eye (cryptotephra). Identification of cryptotephra layers is paramount for complete characterization of the eruptive record of a volcanic center, not just the largest eruptive events. However, cryptotephra recognition is hampered by their small volume in most drill cores.

A non-destructive method to distinguish tephra layers, particularly those of a high silica nature which may not be readily detectable with magnetic methods, is visible/near-infrared (Vis/NIR) spectroscopy. The Vis/NIR region of the light spectrum contains strong absorption features due to charge-transfer absorptions in transition metals (dominated by iron) and vibration and overtone bands due to hydroxyl and water (including near 1.4 µm, 1.9 µm, and 2.2-2.5 µm). The exact position and nature of these bands provide a means to identify various carbonate-, hydroxyl-, iron-, phyllosilicate-, sulfate-, and water-bearing minerals (e.g., Pieters and Englert, 1993).

We produced a series of mixtures of hemipelagic sediment and tephra which were used to identify band positions and features which strongly correlate with the presence of tephra (see figure). The addition of ~15-20 wt.% tephra to a sediment results in recognizable spectral changes. The mixture data was used to create a MATLAB program to run unknown sample analyses through. We then used an ASD FieldSpec to collect Vis/NIR data (0.39-2.5 µm) on the upper 10 m of core collected during IODP 340 (U1396C) off the coast of Montserrat at 0.5 cm resolution and applied our tephra recognition program to this data. We identified 29 potential cryptotephra layers in the 10 m analyzed. Dissolution techniques are being completed to corroborate the spectral data.