Global climate reconstruction across time and space using data assimilation

Abstract

Paleoclimate proxy data span seasonal to millennial time scales, and Earth's climate system has both high- and low-frequency components. Yet it is currently unclear how best to incorporate multiple time scales of proxy data into a single reconstruction framework and to also capture both high- and low-frequency components of reconstructed variables. In the first half of this dissertation we present a data assimilation algorithm that can explicitly incorporate proxy data at arbitrary time scales. Through a series of pseudoproxy experiments, we find that atmosphere--ocean states are most skilfully reconstructed by incorporating proxies across multiple time scales compared to using proxies at short (annual) or long ($\sim$ decadal) time scales alone. Additionally, reconstructions that incorporate long time-scale pseudoproxies improve the low-frequency components of the reconstructions relative to using only high-resolution pseudoproxies. We argue that this is because time averaging high-resolution observations improves their covariance relationship with the slowly-varying components of the coupled-climate system, which the data assimilation algorithm can exploit. These results are insensitive to the choice of climate model, despite the model variables having very different spectral characteristics. Our results also suggest that it may be possible to reconstruct features of the oceanic meridional overturning circulation based solely on atmospheric surface temperature proxies. Water isotope data from ice cores, particularly $\delta^{18}$O, has long been used as a paleoclimate proxy. In the past decade or so isotope-enabled climate models have allowed for a more rich understanding of the climate processes that produce the isotopic signals in precipitation. Such modeling-based studies tend to complicate simple temperature-isotope interpretations by pointing to the many non-local influences on isotopic signals at coring locations. Recent observational studies have also linked ice cores to non-local patterns of climate variability, particularly to mid-latitude atmospheric circulation patterns and to variations in tropical climate. However, the full spatial and temporal extent to which ice cores can robustly inform past climate is unknown. In the second half of this dissertation we estimate a realistic upper-bound on what ice cores can tell us about climate at annual and decadal time scales in a paleoclimate reconstruction context. We employ a similar data assimilation-based reconstruction technique that optimally combines isotopic proxy information with the dynamical constraints of climate models. Through several pseudo and real proxy experiments we assess the spatial and temporal extent to which ice cores can reconstruct the key variables of surface temperature, geopotential height at 500 hPa, and precipitation. We find local reconstruction skill to be the most robust across the reconstructions, particularly for temperature and geopotential heights. Non-local skill is also found for these variables in certain locations. For precipitation we find virtually no reconstruction skill outside of coastal and West Antarctica. These results are in agreement with long-held views that isotopes in ice cores have clear value as local climate proxies, particularly for temperature and atmospheric circulation. These results also show that in principle non-local climate information may also be reliably deduced from ice cores, though the spatial range of this information depends on how proxies are modelled within the reconstruction process, is non-uniform, and may not extend into plausible nearby locations: in particular, Greenland ice cores, by their nature, appear to be relatively uninformative for Europe and the British Isles at annual and decadal time scales.