BrGDGT (branched glycerol dialkyl glycerol tetraethers) are a molecular biomarker utilized for paleotemperature reconstructions. One issue, however, with utilizing brGDGTs is that the distribution differs in relation to sediment environments (i.e., peat, lake, soil). We utilize the probability estimate outputs from five machine learning algorithms and a new modern brGDGT database to track provenance change and apply these models to two downcore records utilizing pollen, non-pollen polymorphs, and XRF (X-ray fluorescence) to confirm the models’ accuracy.