Christian Terkelsen Holme

A thesis for the degree of Doctor of Philosophy defended February 2019.

The PhD School of Science, Faculty of Science, Physics of Ice Climate and Earth, Niels Bohr Institute, University of Copenhagen

Supervisor:
Bo M. Vinther

Co-supervisor:
Vasileios Gkinis

Stable water isotope variability as a proxy of past temperatures - Using polar ice cores as a climate archive

The stable water isotope signal (d 18O and dD) of polar ice cores can be employed as a proxy of the local temperature history. However, it has proven to be challenging to quantify the magnitude of past temperature change solely from the measured isotopic composition. Thus, in a pursuit of improving the accuracy of temperature reconstructions, this PhD dissertation investigates some of the paleoclimatic interpretations that can be made from ice core d18O and dD variability. The work has led to two published papers and one submitted manuscript, all presented in this thesis.

The first study presents temperature reconstruction techniques that only depend on the magnitude of diffusive smoothing on the d18O and dD signals. The objective is to evaluate the performance of such diffusion-based reconstruction techniques in terms of accuracy and precision. By utilizing the methods on both synthetic generated data and ice core data from Greenland and Antarctica, the study finds that single diffusion approaches have higher precision (1.1C) than methods that rely on the differential diffused signal of d18O and dD (1.9C).

The second study examines how to estimate the diffusion length from power spectra of newer, continuously measured water isotope data sets that have lower instrument noise levels. Such power spectra reveal a deviation from the conventional power spectral structure, which complicates the currently used diffusion estimation approaches. By performing tests with synthetic data, the results show that noise and system smoothing occurring through the continuous flow analysis system can explain the observed power spectra. This led to the proposal of two modified techniques that can be used to estimate the diffusion length of continuously measured water isotopes.

The third study analyzes the d18O variability of three ice cores drilled on Renland, East Greenland. The objective is to examine their common signal and how it correlates with regional temperatures back in time. While it is shown that the data can be merged into robust seasonal stacks, the linear relation between d18O and regional temperature changes with time. The unstable covariation coincides with a fluctuating amount of southward exported sea ice along the East Greenland coast. Although, an explanation for the varying d18O-temperature relation remains unknown, the study demonstrates that it can be complicated to perform regressionbased temperature reconstructions for some ice core drill sites.

Ultimately, preliminary work on a high resolution diffusion length profile from the West Antarctic Ice Sheet Divide is presented. The results indicate remarkably high diffusion during the transition from glacial to interglacial. An investigation is initiated to examine whether the elevated diffusion reflects warmer temperatures than previously inferred or if it is a consequence of perplexing post-depositional processes. While this work is unfinished, it emphasizes some of the challenges ahead of the water isotope diffusion community.