Climate variability around the first Pliocene time slice

Existing data/model comparisons for the mid-Pliocene (Dowsett et al., 2013) have identified specific regions of concordance and discord between climate models and proxy data. One reason for site-specific disagreement is likely related to the time (warm peak) averaged nature of the mid-Pliocene ocean temperatures provided within existing proxy syntheses. To facilitate improved data/model comparisons in the future new proxy sea surface temperature reconstructions will focus on specific time slices within the Pliocene epoch. Haywood et al. (2013) identified an initial time slice for environmental reconstruction and climate modelling centred on an interglacial event at Marine Isotope Stage KM5c (3.205 Ma). Critically, this interval displays a very near to modern orbital configuration simplifying the interpretation of proxy data and the experimental design for climate models. Nevertheless, current limitations of chronology and correlation make it likely that new proxy records will be attributable to a time range around the time slice, and may not always represent the time slice specifically. This introduces an element of uncertainty through orbital forcing around the time slice which can be investigated and quantified within a numerical climate modelling framework. The Hadley Centre Coupled Climate Model Version 3 (HadCM3) has been used to perform a series of orbital forcing sensitivity tests around the identified time slice at MIS KM5c. Simulations every 2 kyr either side of the time slice to a range +/- 20 kyr have been completed. The model results indicate that +/- 20 kyr either side of the time slice, orbital forcing generates a less than 1°C change on global MAT. One exception to this relative stability in climate is seen in the North Atlantic (a region noted for disagreement in existing Pliocene data/model comparisons). Here, ocean surface temperature variations of up to 6°C are predicted. These model responses appear to be linked to changes in ocean circulation and the mode of deep water formation and thus the strength of the Atlantic Meridional Overturning Circulation over relatively short timescales (geologically). To place this predicted climate variability around MIS KM5c into context we have completed simulations 20 kyr either side of the 3.060 kyr PlioMAX peak, which is characterised by one of the lightest benthic oxygen isotope excursions evident in the entire PRISM time slab (Marine Isotope Stage K1; Raymo et al. 2004), and displays a radically different orbital forcing compared to present-day. The results show a 5°C change on global MAT, with some terrestrial areas showing changes of 10°C. Therefore, this larger climate variability at K1 would cause imperfect correlation to be much more harmful to data model comparisons than around the KM5c time slice. The results from this suite of simulations suggest that proxies producing MAT with imperfect correlation to the time slice up to 20,000 years before or after may still be representative of the conditions at the MIS KM5c time slice itself due to the subdued nature of orbital forcing at this time.