The most recent geologic interval characterized by warm temperatures similar to those projected for the end of this century occurred about 3.3-3.0 Ma, during the mid-Piacenzian Age of the Pliocene Epoch. Climate reconstructions of this warm period are integral to both understanding past warm climate equilibria and to predicting responses to today’s transient climate. The Arctic Ocean is of particular interest because in this region climate proxies are rare, and climate models struggle to predict climate sensitivity and the response of sea ice. In order to provide the first quantitative climate data from this region during this interval, sea surface temperatures (SST) were estimated from Ocean Drilling Program Sites 907 and 909 in the Nordic Seas and from Site 911 in the Arctic Ocean based on Mg/Ca of Neogloboquadrina pachyderma (sin) and alkenone unsaturation indices. Evidence of much warmer than modern conditions in the Arctic Ocean during the mid-Piacenzian with temperatures as high as 18 °C is presented. In addition, SST anomalies (mid-Piacenzian minus modern) increase with latitude across the North Atlantic and into the Arctic, extending and confirming a reduced mid-Piacenzian pole-to-equator temperature gradient. The agreement between proxies and with previously documented qualitative assessments of intense warming in this region corroborate a poleward transport of heat and an at least seasonally ice-free Arctic, conditions that may serve as a possible analog to future climate if the current rate of Arctic sea-ice reduction continues.

Arctic Ocean surface waters and those of the surrounding seas have been warming since 1965, increasingly since 1995, even more rapidly since 2000, with 2007 and 2008 marking the first two sequential years of extreme summer minimum sea ice coverage (Comiso et al., 2008; Steele et al., 2008; Stroeve et al., 2008). In addition, autumn surface air temperatures during these two years were greater than 5 °C higher than the central Arctic average (Wang & Overland, 2009). Continuation of this trend could lead to a dramatic change in the Arctic ice-ocean-atmosphere regime (Johannessen et al., 1999). In anticipation of continued warming, climate model scenarios for the near future commonly feature Arctic warmth and sea ice retreat yet struggle to predict climate sensitivity and the response of sea ice in these high latitudes. In fact, model simulations of sea ice retreat compare poorly to observations (Stroeve et al., 2007), some underestimating sea ice minima by at least 30 years (Wang & Overland, 2009). Future projections of sea-ice cover vary wildly with some models simulating seasonal ice-free conditions by 2070 while others project virtually no change over the same period of time (Boe et al., 2009).

One way to refine climate models and to improve projections is to attempt to recreate known warm climates of the past from climate proxy data (Robinson et al., 2008a). A model’s ability to accurately portray a past climate state, both in terms of magnitude and spatial variability, increases confidence in climate projections based on that model. Due to the high sensitivity displayed by polar regions during the current warming trend, accurate reconstructions of paleo-conditions in high latitude regions during past warm intervals are integral to reliable model results, but data are rare due to the shortage of paleoclimate proxies in high latitudes, and high resolution temporal correlation between regions is complicated.

The most recent geologic interval of global warmth comparable to climate projections for the end of this century was ~3.3-3.0 Ma (IPCC 2007), during the mid-Piacenzian Age of the Pliocene Epoch. During this time interval, the positions of the continents and the patterns of oceanic circulation were similar to modern, but mean global temperatures were 2-3 °C warmer, and sea level was about 25 m higher (Dowsett, 2007). It was also during the Piacenzian (between 3.6 and 2.4 Ma) that restricted local scale glaciations transitioned to extensive regional scale glaciations on the circum-Arctic continents (e.g., Fronval & Jansen, 1996; Mudelsee & Raymo, 2005). Paleoclimatologists interested in this warm interval as a possible analog to future warming, as well as other climate researchers intrigued by the transition between this warm period and the subsequent onset of Northern Hemisphere glaciation, recognize the potential of mid-Piacenzian climate reconstructions to reveal uncertainties regarding climate sensitivity. As a result, a wealth of paleoclimate data exists for this warm interval, but most is restricted to lower latitudes where traditional paleoclimate proxy methods (i.e., inferring  conditions from faunal assemblage data) work best.

The USGS Pliocene Research, Interpretation and Synoptic Mapping (PRISM) Project is charged with reconstructing global conditions during the ~3.3 to 3.0 Ma time interval (hereafter “the mid-Piacenzian”) in an effort to better understand past and possible future climate dynamics. PRISM reconstructions of sea-surface temperature (SST), based largely on planktic foraminifer assemblage data, indicate that temperature  differences between the mid-Piacenzian and modern increase with latitude in the North Atlantic (Dowsett et al., 1992). That is, mid-Piacenzian temperatures near the equator were similar to modern temperatures, but temperatures in the higher latitudes were several degrees warmer than at present. This reconstructed equator-to-pole gradient has been indeterminate at and above ~66° N, however, because temperature estimates from polar regions such as the Nordic Seas and Arctic Ocean have remained elusive due to the lack of geologic proxies yielding quantitative results as well as weak age control.