Paper No. 6
Presentation Time: 9:00 AM-6:00 PM


WANG, Ting1, SURGE, Donna2 and LEES, Jonathan1, (1)Geological Sciences, University of North Carolina, 104 South Road, CB #3315, Chapel Hill, NC 27599, (2)Dept. of Geological Sciences, University of North Carolina at Chapel Hill, 104 South Road, Mitchell Hall, Chapel Hill, NC 27599,

Many sclerochronological studies have used growth patterns and isotopic ratios in mollusc shells to reconstruct past environmental and climate changes at seasonal time scales. However, uncertainties exist using this approach (e.g., time-averaging biases due to changes in growth rates throughout the year, unconstrained noise in the environmental signal, etc.). To address and better quantify these uncertainties, we developed a statistical tool to evaluate the errors of temperatures calculated from high-resolution isotopic time series of mollusc shells.

An earlier study developed an approach that resolved best-fit sinusoids along oxygen isotope (δ18O) time series (Wilkson and Ivany, 2002). It assumes annual variation in δ18O values across accretionary growth increments represents sinusoidal variation in temperature and/or the composition of ambient water. The output determines quantitatively mean annual δ18O, seasonal range in δ18O, and sine period reflecting growth rate. However, it does not estimate the errors of seasonal extremes (i.e., the most positive/negative isotope ratio in response to the coldest/warmest temperature). We implemented an evaluation system and modified the previous approach by introducing additional statistical computation and analysis, such as window size calculation for sinusoidal smoothing, JackKnife error estimation, Discrete Fourier Transform to initialize parameters for sinusoidal function, and the Nelder-Mead method to optimize parameters for sinusoidal function. The final product of this study is ClamR, an R package that can automatically calculate and plot the error of a temperature or δ18O time series in addition to the errors of annual averages and seasonal ranges. ClamR improves paleoclimate interpretation by providing errors for all the major temperature variables (annual averages, seasonal ranges and seasonal extremes) and making comparisons between paleoclimate and modern climate records more comprehensive and justified.