A computational approach to Quaternary lake-level reconstruction applied in the central Rocky Mountains, Wyoming, USA

Sediment-based reconstructions of late-Quaternary lake levels provide direct evidence of hydrologic responses to climate change, but many studies only provide approximate lake-elevation curves. Here, we demonstrate a new method for producing quantitative time series of lake elevation based on the facies and elevations of multiple cores collected from a lake's margin. The approach determines the facies represented in each core using diagnostic data, such as sand content, and then compares the results across cores to determine the elevation of the littoral zone over time. By applying the approach computationally, decisions are made systematically and iteratively using different facies classification schemes to evaluate the associated uncertainty. After evaluating our assumptions using ground-penetrating radar (GPR), we quantify past lake-elevation changes, precipitation minus evapotranspiration (ΔP−ET), and uncertainty in both at Lake of the Woods and Little Windy Hill Pond, Wyoming. The well-correlated (r = 0.802 ± 0.002) reconstructions indicate that water levels at both lakes fell at > 11,300, 8000-5500, and 4700-1600 cal yr BP when ΔP − ET decreased to − 50 to − 250 mm/yr. Differences between the reconstructions are typically small (10 ± 24 mm/yr since 7000 cal yr BP), and the similarity indicates that our reconstruction method can produce statistically comparable paleohydrologic datasets across networks of sites.