Testing the uncertainty of diffuse reflectance spectroscopy methods for estimating marine sediment components by full-resolution spectrophotometry and principle component analyses

Abstract

Estimating terrestrial and biogenic component variations in marine sediment cores provide essential information in paleoceanographic studies. The variations of terrestrial and biogenic components can be used to infer the changes of sediment provenances that are important for understanding what processes have governed the sediment compositional changes in the cores. One non-destructive technique, diffuse reflectance spectroscopy (DRS), is now commonly used to capture high-resolution information on sediment component changes. The DRS method is a quick and cost-effective procedure for sediment core analysis that requires no chemical analysis with complicated procedures. Recently, one DRS device for capturing visible (VIS) color reflectance data (CRD) (400–700nm) from core surfaces, the Minolta CM-2600d, has become a standard tool in all marine sediment core analysis. However, by capturing the visible wavelength of CRD only, there are uncertainties in the CRD near the ultraviolet (UV) and near-infrared (NIR) obtained by Minolta CM-2600d are poorly known. In this study, we used a full-resolution (FR) DRS device, the Analytical Spectral Device (ASD) LabSpec® Pro FR ultraviolet/visible/near-infrared (UV/VIS/NIR) spectrometer (250–2500nm) to compare with the CRD obtained from the Minolta CM-2600d, using the marine sediment core MD103264 taken from offshore southwestern Taiwan. We compared the differences in the VIS wavelength range of CRD between the two devices by analyzing the derivatives and a VARIMAX-rotated Principle Components Analysis (VPCA) of the CRD from 420 samples from core MD103264. Our results show significant uncertainties in CRD near the wavelengths of 400–450 and 650–700nm in the data from the Minolta CM-2600d. These uncertainties could result in biased or erroneous estimates for terrestrial and biogenic components of marine sediment cores. Our experiments indicate that the CRD obtained with any device with VIS wavelength range only should be interpreted with caution, particularly in the 400–450 and 650–700nm range.