Rapid Prediction of past Climate Condition from Lake Sediments by Near-Infrared (NIR) Spectroscopy

Abstract

This study explored the feasibility of rapid, nondestructive near-infrared (NIR) reflection spectroscopy for the prediction of conventional physical properties, carbon-nitrogen-sulfur (CNS) analysis, and concentration of inorganic components in sediment cores from a brackish lake. A long core sample, which consisted of well-preserved annually formed lamina from Lake Ogawara along the Pacific coast in Aomori Prefecture, northeastern Japan, was used to investigate the past environmental record. The core was previously analyzed for physical properties, CNS, and inorganic components. Calibration models were developed from NIR reflection spectra of 149 core samples. Partial least squares (PLS) analysis provided good regression models between measured and predicted values for water content, total nitrogen (TN), total organic carbon (TOC), total sulfur (TS), Al₂O₃, S/Al₂O₃, Fe₂O₃/Al₂O₃, Sc/Al₂O₃, Cu/Al₂O₃, and Zn/Al₂O₃ with coefficients of determination (r ² ) for cross-validation of 0.73, 0.89, 0.88, 0.73, 0.92, 0.81, 0.82, 0.75, 0.82, and 0.82, respectively. The variation of predicted component values as a function of depth showed the same trend as that of conventionally measured values. This study also showed the possibility of NIR spectroscopy as an on-site, rapid analytical tool for the identification of tephra (fragmental material produced by a volcanic eruption regardless of composition, fragment size, or emplacement mechanism), which is important for dating.