High-resolution grain size distribution of sediment core with hyperspectral imaging

Abstract

The study of sediment cores allows for the reconstruction of past climate and environment through physical-chemical analysis. Nevertheless, this interpretation suffers from many drawbacks that can be overcome with the newest technologies. Hyperspectral imaging is one of these and allows a fast, high resolution, and non-destructive analysis of sediment cores. In this study, we use visible and near-infrared hyperspectral imaging to predict particle size fractions and distribution (PSD) at a resolution of 200 μm on a previously well-studied sediment core taken from Lake Bourget (Western Alps, France). These predictions agree with previous studies on this core. Then, the PSD was used to estimate sedimentary deposit sources using the PSD unmixing algorithm AnalySize. It permitted estimation of the contribution of five sources (micrite, small and large bio-induced calcite crystals, diatom frustules, detrital particles), which had previously been characterized. The spatial dimension allowed for laminae to be discretized and counted, in agreement with the age-depth model previously established. We then evaluated the particle size and spectral signatures of each of these annual laminae, hence characterizing their physico-chemical composition. These high-resolution data also allowed for estimation of the accumulation rate (cm/year) of each of the main sources in the laminated unit and inferring the trophic status and the presence of instantaneous events of the lake.