High-resolution summer temperature reconstruction from Lake Silvaplana based on in-situ reflectance spectroscopy

Annually laminated (varved) sediments of proglacial Lake Silvaplana (46 ̊27’N, 9 ̊48’E, 1791 m a.s.l., Engadine, eastern Swiss Alps) provide an excellent archive for quantitative high-resolution (seasonal – annual) reconstruction of high- and lowfrequency climate signals back to AD 1580. The chronology of the core is based on varve counting, Cs-137, Pb-210 and event stratigraphy. In this study we present a reconstruction based on in-situ reflectance spectroscopy. In situ reflectance spectroscopy is known as a cost- and time-effective non destructtive method for semi-quantitative analysis of pigments (e.g., chlorines and carotenoids) and of lithoclastic sediment fractions. Reflectance-dependent absorption (RDA) was measured with a Gretac Macbeth spectrolino at 2 mm resolution. The spectral coverage ranges from 380 nm to 730 nm at 10 nm band resolution. In proglacial Lake Silvaplana, 99% of the sediment is lithoclastic prior to AD 1950. Therefore, we concentrate on absorption features that are characteristic for lithoclastic sediment fractions. In Lake Silvaplana, two significant correlations that are stable in time were found between RDA typical for lithoclastics and meteorological data: (1) the time series R 570 /R 630 (ratio between RDA at 570 nm and 630 nm) of varves in Lake Silvaplana and May to October temperatures at nearby station of Sils correlate highly significantly (calibration period AD 1864 – 1951, r = 0.74, p < 0.01 for 5ptsmoothed series; RMSE is 0.28 ̊C, RE = 0.41 and CE = 0.38), and (2) the minimum reflectance within the 690nm band (min690) data correlate with May to October (calibration period AD 1864 – 1951, r = 0.68, p < 0.01 for 5pt-smoothed series; RMSE = 0.22 ̊C, RE = 0.5, CE = 0.31). Both proxy series (min690nm and R 570 /R 630 values) are internally highly consistent (r = 0.8, p < 0.001). In proglacial Lake Silvaplana the largest amount of sediment is transported by glacial meltwater. The melting season spans approximately from May to October, which gives us a good understanding of the geophysical processes explaining the correlations between lithoclastic proxies and the meteorological data. The reconstructions were extended back to AD 1580 and show a broad corresponddence with fully independent reconstructions from tree rings and documentary data.

Age modeling of young non-varved lake sediments: methods and limits. Examples from two lakes in Central Chile

High-resolution and highly precise age models for recent lake sediments (last 100–150 years) are essential for quantitative paleoclimate research. These are particularly important for sedimentological and geochemical proxies, where transfer functions cannot be established and calibration must be based upon the relation of sedimentary records to instrumental data. High-precision dating for the calibration period is most critical as it determines directly the quality of the calibration statistics. Here, as an example, we compare radionuclide age models obtained on two high-elevation glacial lakes in the Central Chilean Andes (Laguna Negra: 33°38′S/70°08′W, 2,680 m a.s.l. and Laguna El Ocho: 34°02′S/70°19′W, 3,250 m a.s.l.). We show the different numerical models that produce accurate age-depth chronologies based on 210Pb profiles, and we explain how to obtain reduced age-error bars at the bottom part of the profiles, i.e., typically around the end of the 19th century. In order to constrain the age models, we propose a method with five steps: (i) sampling at irregularly-spaced intervals for 226Ra, 210Pb and 137Cs depending on the stratigraphy and microfacies, (ii) a systematic comparison of numerical models for the calculation of 210Pb-based age models: constant flux constant sedimentation (CFCS), constant initial concentration (CIC), constant rate of supply (CRS) and sediment isotope tomography (SIT), (iii) numerical constraining of the CRS and SIT models with the 137Cs chronomarker of AD 1964 and, (iv) step-wise cross-validation with independent diagnostic environmental stratigraphic markers of known age (e.g., volcanic ash layer, historical flood and earthquakes). In both examples, we also use airborne pollutants such as spheroidal carbonaceous particles (reflecting the history of fossil fuel emissions), excess atmospheric Cu deposition (reflecting the production history of a large local Cu mine), and turbidites related to historical earthquakes. Our results show that the SIT model constrained with the 137Cs AD 1964 peak performs best over the entire chronological profile (last 100–150 years) and yields the smallest standard deviations for the sediment ages. Such precision is critical for the calibration statistics, and ultimately, for the quality of the quantitative paleoclimate reconstruction. The systematic comparison of CRS and SIT models also helps to validate the robustness of the chronologies in different sections of the profile. Although surprisingly poorly known and under-explored in paleolimnological research, the SIT model has a great potential in paleoclimatological reconstructions based on lake sediments