We compile 160-kyr long non-polar mono-specific stacks of δ¹³C and of δ¹⁸O from the planktic foraminifera G. ruber and or T. sacculifer and compare them with carbon cycle simulations using the BICYCLE-SE model. Simulations are forced with  atmospheric CO₂ and δ¹³CO₂. In our stacks and our model-based interpretation we can not detect a carbonate ion effect, the species-specific isotopic fractionation during hard shell formation as function of carbonate chemistry in the surounding sea water.

Stable isotope ratios of foraminiferal shells from deep-sea sediments preserve key information on the variability of ocean circulation and ice volume. We present the first global atlas of harmonized raw downcore oxygen and carbon isotope ratios of various planktonic and benthic foraminiferal species. The atlas is a foundation for the analyses of the history of Earth system components, for finding future coring sites, and for teaching marine stratigraphy and paleoceanography.

Maps and fields of near-sea-surface temperature differences between the past and present can be used to visualize and quantify climate changes and perform simulations with climate models. We used a statistical method to map sparse and scattered data for the Last Glacial Maximum time period (23 000 to 19 000 years before present) to a regular grid. The estimated global and tropical cooling would imply an equilibrium climate sensitivity in the lower to middle part of the currently accepted range.

We present a technique for ocean state estimation based on the combination of a simple data assimilation method with a state reduction approach. The technique proves to be very efficient and successful in reducing the model-data misfit and reconstructing a target ocean circulation from synthetic observations. In an application to Last Glacial Maximum proxy data the model-data misfit is greatly reduced but some misfit remains. Two different ocean states are found with similar model-data misfit.