We established the statistical basis to estimate the space-time structure of climate variability from instrumental and paleo-data. We analyzed a set of replicate sediment cores, purchased the planned equipment, and published multiple research papers. We assembled a team of several postdocs, PhD students, and student helpers to cover research components in all work packages. Additionally, we gave several invited talks at international conferences and seminars and organized workshops on climate variability. Furthermore, we developed new methods to estimate the time-scale-dependent spatial structure and applied them to climate model and instrumental data. We analyzed replicate proxy records cores to better distinguish local from large-scale climate variability and characterize the time-uncertainty in proxy records. Since the estimated signal-to-noise ratios from some proxy types were low, we further enhanced our understanding of and refined new proxies for climate variability reconstructions. This endeavor included developing methods to reconstruct fast climate variability from sediment cores (Mass Spectral Imaging), gaining an in-depth comprehension of the climate signal recorded in ice cores, and developing methods to overcome the high noise level and uncertainties in the transfer function. To utilize our understanding for interpreting and correcting climate variability estimates, we implemented this knowledge in proxy system models and developed a theoretical framework to handle time-scale-dependent proxy uncertainties.
These advancements provide a foundation for estimating the space-time structure from annual to millennial time-scales by combining instrumental observations and various paleo-climate archives. We reconstructed local and global temperature variability, demonstrating that regional climate variations persisted for longer time scales than climate models simulating past climate states are able to reproduce. This suggests an underestimation of regional variability on multidecadal and longer time scales, as well as a bias in climate projections and attribution studies. We proposed that one reason for this discrepancy is the likely underestimation of marine temperature variability in model simulations, which drives the centennial-millennial variability on land. We conducted the first quantitative reconstruction of changes in temperature variability between the Last Glacial Maximum and the Holocene based on a global network of marine and terrestrial temperature proxies in order to study the mean state dependency of climate variability. Our findings revealed that the overall pattern of reduced variability could be explained by changes in the meridional temperature gradient, a mechanism that may indicate further decreases in temperature variability in a warmer future. Lastly, we estimated the signal-to-noise ratio of Holocene and Glacial temperature reconstructions. Due to the relatively low estimated signal-to-noise ratios, we decided to develop and optimize a method to reconstruct fast climate variability from sediment cores (Individual Foraminifera Analysis). This approach will enable us to refine our estimates of how climate variability may change in a warmer world.
