A reconstruction of temperature, ice volume and atmospheric CO2 over the past 40 million years

Knowledge on past climate change largely emerges from sediment records drilled from the ocean floor and ice-core records from the Antarctic and Greenland ice sheets. From these records proxy data is obtained indicating changes in, for example, temperature, sea level and greenhouse gas concentrations. A key parameter that emerges from the sediment records is the oxygen isotope ratio, δ18O, from fossilised benthic foraminiferal shells. The benthic δ18O data serves as a proxy for changes in local deep-water temperatures and global ice volume. With an innovative modelling approach surface-air temperatures relative to present day are derived from changes in benthic δ18O. This temperature anomaly is used to derive the two contributions to the benthic δ18O data with ice-sheet models and a simple deep-water temperature coupling function. With multiple 1-D ice-sheet models a continuous record of temperature and sea level over the past 40 million years has been derived from the observed benthic δ18O. The obtained data is shown to compare well with observations and other modelling results. It is shown that both the relation of ice volume with temperature and ice volume with sea water δ18O are highly variable and not constant through time. Therefore the use of a global coverage of ice volume changes is important for interpreting the benthic δ18O records in a transient mode. Using the temperature reconstruction from the 1-D ice-sheet models, a variety of different CO2 proxies are tested for their coherence with the observed ice-core CO2 data. The most coherent records are selected to derive a self-consistent and continuous CO2 record over the past 20 million years. Moreover, the long-term climate sensitivity of temperature to CO2 changes is derived, which includes radiative forcing of CO2, short and fast feedbacks and a correction for other greenhouse gasses. The large sensitivity derived here implies that subtle changes in atmospheric CO2 could be related to the Mid-Pleistocene Transition and the initiation of NH glaciation. With comprehensive 3-D ice-sheet models a more in-depth analysis has been performed on ice volume changes over the past million years. For the first time the separate contributions of these four ice sheets have been calculated explicitly over this time interval. With respect to eustatic sea level, the large NH ice sheets are naturally responsible for the largest variability during the Plio-Pleistocene. Coherently, these ice sheets also provide the largest contribution to sea water δ18O. The Antarctic and Greenland ice sheets contribute about 10 % to changes in eustatic sea level. For sea water δ18O the contribution is even larger, between 10 and 20 %, mainly due to the relatively much lower δ18Oice values of the Antarctic ice sheet. With a simple linear forcing of temperature and/or sea level it is shown that ice volume on Antarctica is increased due to lowering of sea level, which is enhanced by decreasing temperatures. On the contrary, ice growth on Eurasia and North America starts when temperatures drop, due to a positive surface mass balance which is enhanced by lowering of sea level