Establishing a chronology for lacustrine sediments using a multiple dating approach—A case study from the Frickenhauser See, central Germany

In order to establish a reliable chronology for lacustrine sediments of the Frickenhauser See (central Germany) different dating methods have been applied. A total of 17 AMS ¹⁴C dates, all within the last 2000 years, were supplemented with ¹³⁷Cs/²¹⁰Pb dating and varve counting of the uppermost sediments (131 years). The age–depth model for the Frickenhauser See has to cope with highly variable sedimentation rates and overlapping probability distributions of calibrated ¹⁴C dates. The uncertainty of calibrated ¹⁴C dates could be considerably reduced by including the stratigraphic relationship of the dated samples, the age information derived from short-lived isotopes and varve counting as well as an upper and lower limit of realistic sedimentation rates as ‘a priori’ information in the calibration procedure. Sets of possible age combinations obtained by repeated sampling from the modified probability distributions were used to calculate continuous age–depth relationships based on monotonic smoothing splines. The obtained age–depth model for the sediment record of the Frickenhauser See represents the average of over 16,000 such model runs and suggests a drastic increase in sedimentation rates from around 1–2 mm a⁻¹ (200–1000 AD) to over 25 mm a⁻¹ for the period between 1100 and 1300 AD. From then on, sedimentation rates exhibit relatively stable values around 3–9 mm a⁻¹. ‘Conventional’ age–depth models such as general polynomial regression or cubic splines either do not include the obtained age-information in a satisfying manner (the model being too “stiff”) or exhibit “swings” causing age-reversals in the model. Although the age–depth relationships obtained for monotonic smoothing splines and mixed-effect regression are generally very similar, they differ in their respective sedimentation rates as well as in their uncertainties. Mixed-effect regression resulted in much higher sedimentation rates of more than 37 mm a⁻¹. These results suggest that monotonic smoothing splines give better control of the age–depth model characteristics and are well suited in situations, where the integrity of ¹⁴C dates is high, i.e. the dated material represents the age of the respective layer.