Transient nature of the Earth's climate and the implications for the interpretation of benthic δ18Oδ18O records

From the marine benthic δ18Oδ18O records it is known that the Earth's climate has experienced significant variability over the past 40 million years. In general, a number of assumptions are often needed to disentangle the benthic δ18Oδ18O data into its temperature and ice-volume contributions. In this study, a transient (1-D ice-sheet) model is used which overcomes these shortcomings by relating temperature to the benthic δ18Oδ18O data, leading to a self-consistent and continuous record of δ18Oδ18O, temperature and sea level. The contribution of land ice to benthic δ18Oδ18O is examined with a set of sensitivity experiments, varying the mean δ18Oδ18O of the ice for different ice sheets. It is shown that the scaling factor of sea-water δ18Oδ18O (δw) to sea level is not constant over the long time scales. However, our sensitivity experiments do show that over the long time scale, the general assumed 1.0–1.1‰ per 100 m of sea level is in reasonable agreement with our model results. Moreover, the sea-level (ice volume) response to temperature (ΔS/ΔT) is shown to vary through time, with the largest response found when variations in Antarctic ice volume dominates the sea-level variations during the Oligocene to Early Miocene. Furthermore, this response is quite large compared to the individual response of ice sheets during the Plio-Pleistocene, whereas the combined ΔS/ΔT in this period is comparable to that during the Oligocene. Accordingly, the transient behaviour is very important for the interpretation of data records, with respect to climate sensitivity and climate change during the past 40 million years.

Research Highlights
► The scaling factor of δw to sea level is not constant over long time scales.
► A constant δw to sea-level ratio notably affects interpretation of benthic δ18O.
► The Antarctic ice sheet is shown to be the most sensitive to temperature changes.
► Transient behaviour is very important for the interpretation of data records.