Ocean-atmosphere-ice sheet interactions in the polar north, 50-150 ka BP: Implications for climate system processes

Preliminary results from drive-responce analysis by convergent cross mapping (CCM) on 800 000 year long time series of past climate variability (global ice volume, sea surface temperatures and solar insolation) show that IT analysis is a promising technique that can give insight into how information propagate through complex chaotic systems such as the Earths climate. we use CCM, a model-free causality test, to quantify the dynamical link between global climate and the locally varying summer energy (integrated summer insolation) received from the sun. We find that the influence of summer energy is most prominent north of ca 50 degrees North where the large ice sheets formed during glacial stages. This finding is consistent with the prediction of the Milankovitch theory. However, we also find that the results is consistent with insolation at mid southern latitudes also forcing global climate. These results represent the first step in the work to track how the forcing from insolation propagates through the climate system. Results from a synchronization of paleoclimate proxy records from sediment cores in the Norwegian Sea and Chinese speleothem records show that there is a very tight link between climate and ocean circulation in the Norwegian Sea and the intensity of the Asian Monsoon. This makes it possible to transfer the precise radiometric chronologies of the spelothem records to the marine sediments where independent time control is difficult to obtain. This chronological framework enables further analysis of the information transfer between climate records from the Norwegian Sea/North Atlantic and globally important climate parameters such as variability in Earth?s orbit, Atmospheric CO2 concentration, low latitude hydrology and global sea-level. We explore the drive-response relationships in these coupled systems with the information transfer (IT) and the convergent cross mapping (CCM) analytical techniques. These methods employ conceptually different approaches to detect the relative strength and directionality of potentially chaotic and nonlinearly coupled systems. IT is a non-parametric measure of information transfer between data records based on transfer entropy, while CCM relies on delay reconstructions using Takens? theorem. This approach enables us to address how the climate system processes interact and how this interaction is affected by external forcing from for example greenhouse gases and orbital variability.

Understanding how Earth's climate responds to present and past changes in climate forcing is hampered by the complexity of the global climate system, involving nonlinearity, teleconnections and feedback mechanisms. This makes it difficult to disentangle the drive and response relationships between climate system variables. Knowledge of how climate system processes interact with each other and how this interaction is affected by external forcing is essential in order to understand the causes for abrupt c limate change, and how effects from forcing of for example greenhouse gases and orbital variability propagate through the climate system. The INTERACT project seeks to elucidate such interactions between the oceans, the atmosphere and ice sheets occurring in the North Atlantic region and their relation with global climate over a 100 ka time interval that covers both glacial and interglacial climate states. We will achieve this by linking new and previously published high resolution climate proxy records f rom the Nordic Seas/North Atlantic region that records variability in atmospheric, oceanic and ice sheets, with key global climate parameters such as CO2, orbital variability, sea-level, low latitude hydrology and Southern Hemisphere climate/bipolar see-s aw. The drive and response relationships between the climatic factors and processes will be addressed by a novel data analytical technique, Information Transfer analysis, which is especially developed for non-parametric drive and response attribution in c haotic non-linear systems.