TOWARDS A UNIVERSAL CALIBRATION MODEL TO PREDICT BIOGENIC SILICA AND ORGANIC CARBON PERCENTAGES IN LACUSTRINE SEDIMENT CORE SAMPLES

In many settings, biogenic silica (BSi) and total organic carbon (TOC) are widely used as proxies for temperature and/or environmental variations that are helpful in paleoclimate and paleoenvironmental reconstructions. Often, the methodology for analyzing these parameters in sediments can be expensive and time consuming (particularly for BSi). However, Fourier Transform Infrared (FTIR) Spectroscopy offers an efficient alternative where many samples can be run with minimal amount of sediment and time. This technique is advantageous in that it requires small volumes of sediment (~0.01g), minimal sample preparation (mixing sample with potassium bromide powder), and instrumental analysis times are relatively rapid (a few minutes per sample). FTIR Spectroscopy quantifies BSi and TOC using infrared radiation (IR) absorbance units—as opposed to percentages of BSi or TOC—which are difficult to compare across different studies and localities. Therefore, there is a need for a systematic way to convert the results from the FTIR Spectrometer into percentages. In this research project, we address this need by building a universal calibration model using partial least squares (PLS) regression that converts BSi absorbance to percentages. We developed this model using a PLS package in R and based our model on samples from Arctic lakes in Greenland and Norway. Our preliminary model uses a k-fold cross-validation method and utilizes three components. Ongoing work intends to improve on the model’s prediction accuracy, expand our calibration model to include TOC percentages, and incorporate more BSi samples from other locations. We aim for the model to be universal, where paleoclimatologists can use it on samples from various localities and compare their results. This model will prove a valuable tool in paleoclimate reconstruction by facilitating FTIR Spectroscopy on lake sediments.