The Climate of the Pliocene: Simulating Earth's Last Great Warm Period copy

Mark A. Chandler

moof moof


Much of the research conducted by scientists at the Goddard Institute for Space Studies is aimed at developing tools for simulating future climate change. The ultimate objective is to help anticipate the impact that those changes will have on society and the environment. The development of computer models is central to our efforts, and global climate models (GCMs), in particular, are the primary tool we use to simulate the Earth's environment and the forces that affect it. Among those forces are many which are anthropogenic, or human-caused, including increased greenhouse gases and aerosols, ozone depletion, and deforestation.

Figure 1: During the Pliocene, global temperatures, particularly at high latitudes, are believed to have been significantly warmer than today. This figure shows the Pliocene surface air temperature increase compared to the present day as simulated by the NASA/GISS

In employing climate models, we must make every attempt possible to verify that they are capable of accurately portraying Earth's climate and its sensitivity to change. Validating a GCM's equilibrium capability is done by comparing simulations of the modern climate, often referred to as current climate control simulations, to observations. However, despite our interest in anthropogenic influences on future climate, testing the GCM's sensitivity to change generally relies on our ability to understand natural climate changes from the past. Predictions of future changes gain credibility when our models can accurately simulate changes that have actually occurred. Unfortunately, historical records of most climatological variables, such as temperature and precipitation, rarely exceed 100 years in length. Moreover, many of the changes anticipated in the future are likely to exceed historical precedents. Thus, we commonly step back further in time and examine the geologic record, which contains examples of global-scale climate change similar in magnitude to that predicted to occur during the 21st century.

Many past time periods have been simulated, both for the purpose of evaluating model capabilities and as a technique for studying the Earth's climatic evolution. Simulations of key periods during the last ice age commonly provide excellent climate change scenarios of large magnitude. If our interest, however, is in climates warmer than today, we must look back at least three million years, to the middle of the Pliocene epoch, to find a period in Earth history with global average temperatures more than a degree (Celsius) higher than the present.

A Warm Time in the Past

The Pliocene epoch covers the period from approximately 5 to 1.8 million years ago and, as such, spanned the period of time during which the Earth transitioned from relatively warm climates to the generally cooler climates of the Pleistocene. This transition included the emergence of the direct ancestors of humankind and contains the beginnings of cyclic Northern Hemisphere glaciation. The Pliocene epoch itself contains episodic climate fluctuations prior to the late Pliocene cooling, and our focus for study is a warm period in the middle Pliocene between 3.15 and 2.85 million years before present.

This middle Pliocene warming is, potentially, an analog of the future that may provide a means of gaining insight into the effects of global warming. Additionally, unlike many more ancient periods, which were also warmer than the present; the paleogeography of the Pliocene is similar to the present, many of the Pliocene plant and animal species are similar to those that remain today, and large numbers of ocean and land-based cores contain well-dated Pliocene sediments that are available for interpretation and mapping.

In our simulations of the middle Pliocene climate we use the GISS GCM and data generated and/or compiled by the PRISM (Pliocene Research, Interpretation, and Synoptic Mapping) project, part of the U.S. Geological Survey 's Global Change Research Program . PRISM focuses on documenting climates of the middle to late Pliocene, with a primary goal of providing the climate modeling community with improved quantitative global paleoenvironmental information . Our Pliocene modeling, in turn, helps test the consistency of different sets of paleo observations, each of which has its own uncertainties. Two examples of PRISM data that has been incorporated for use in the climate model are shown, below, in figures 2 and 3.

Figure 2: Pliocene sea surface temperatures. Differences from modern values values for two selected months. Units are C.

Figure 3: Pliocene and modern vegetation global albedo distribution.

GCM Simulations of the Middle Pliocene

Estimates of sea surface temperatures (SSTs), based on microfossils from deep ocean cores reveal a warm phase in the Pliocene between about 3.15 and 2.85 million years ago. Pollen records from land-based cores, although not as well-dated, also show evidence for a warmer climate at about this same time and further indicate that continental moisture levels were quite different from today. What caused the climate to be warmer is not known with certainty, but increased levels of greenhouse gases have been suggested (see next section). Also, previous sensitivity experiments using the GISS GCM imply that warmer climates, such as those of the Pliocene, can be simulated with increased ocean heat transport. Recent evidence from North Atlantic deep sea records indicates that the oceans may very well have played a major role in the warming seen in the Pliocene.

As a test of this hypothesis we applied Pliocene SSTs, together with an estimate of the terrestrial vegetation cover, as boundary conditions in a GISS GCM simulation (see figures 1 and 2). The GCM provides the method for investigating the atmospheric processes that might have maintained the warmer Pliocene climate while consistency between independent palynological estimates of climate and the simulation results help verify the GCM's sensitivity to altered conditions.

In our experiments we have found both consistencies and inconsistencies between model and data-generated paleoclimate estimates. Temperature estimates show the greatest consistency, with both model and data indicating significantly warmer temperatures at high latitudes and diminished warming nearer to the equator (figure 4). The continental temperatures agree well with estimates from palynological studies, especially in the circum-North Atlantic region. This is not unexpected since that region is strongly influenced by the dramatically warmer North Atlantic SSTs. The GCM also yields temperature increases up to 10C along the Arctic coasts and shows greatest warming in the winter. Although the original temperature increase is driven by warmer SSTs, much of the continental interior warming is generated by an ice-albedo feedback, as reduced snow cover in the warmer climate reflects less solar radiation away from the surface during winter months (see figure 5). Further warming at high latitudes comes from the increased levels of atmospheric water vapor (a greenhouse gas) which results from the warm, ice-free ocean conditions (figure 5 again).

Despite the generally warmer climatic conditions, some areas show overall cooling. Notably, East Africa cools by 2 to 3C due to increased low-level cloud cover, which reflects large amounts of incoming solar radiation back to space. Very few paleo observations are available for some remote parts of Africa, but our simulation is consistent with the single palynological record that exists for that region.

Figure 4: Change in Northern Hemisphere surface air temperatures. Results of a Pliocene simulation minus a current climate "control" simulation. Units are C.

Figure 5: Feedback mechanisms in the Pliocene Northern Hemisphere. All values are the zonally averaged difference between the Pliocene and current climate

Estimates of hydrological values such as precipitation, soil moisture, and surface runoff show far less consistency between the simulation and data than do temperatures. This is not really a surprising result given that hydrologic processes are notoriously difficult to simulate using coarse-grid numerical models while terrestrial environments (what the data report) are usually quite heterogeneous.

The most common discrepancy seems to be an underestimation by the model of wetter conditions, as interpreted from pollen records, throughout the Northern Hemisphere. For example, the model predicts lower effective moisture (precipitation minus evaporation) in western North America, but geologic records indicate wetter conditions during the Pliocene. The root of the difference seems to lie in the northern summer season, where the model's ground hydrology responds to the warmer ground temperatures by drying out. Adding to the problem, the somewhat diminished intensity of the atmospheric circulation (a result of reduced latitudinal [i.e. equator-to-pole] temperature gradients) decreases the ability of the atmosphere to carry moisture evaporated from the ocean surface over the continents, where it could rain out and replenish the soil.

In the Arctic, Pliocene forests dominated where tundra exists today. In altering the specified vegetation cover to match this change, wetter soil moisture condtions were also assigned. Throughout the simulation, Pliocene Arctic soils remained wetter than the present day, fed by increased rainfall originating over the warmer Arctic ocean. The results indicate, at least, that these specified wet conditions are in equilibrium with the simulated climate.

What Caused the Middle Pliocene Warming?

Sea surface temperature patterns such as of the Pliocene (e.g., large warming at mid and high latitudes with stable tropical temperatures) are inconsistent with the warming caused by increased CO 2as we understand it from GCM doubled-CO 2experiments. Well-mixed greenhouse gases tend to warm the tropics substantially as water vapor evaporated from tropical and subtropical oceans provides a positive feedback to the low latitude warming. However, it is possible that some combination of CO 2increase and ocean heat transport change may have resulted in the warmer Pliocene surface temperatures since altered ocean circulation could increase the divergence of heat from the tropics.

Our simulations of the Pliocene climate used near-modern levels of atmospheric carbon dioxide (315 ppm) but required a nearly 30% change in the implied meridional ocean heat transports to maintain Pliocene conditions. This additional heat transport implies substantial changes in the ocean's thermohaline circulation , wind-driven circulation, or both. Evidence of such thermohaline circulation changes comes from carbon isotopic data from deep-sea microfossils, which show that the strength of North Atlantic deep water production was increased during the middle Pliocene. Wind-driven changes, however, are not yet supported by the wind velocities indicated by model simulations or by geologic evidence.

We also conducted several Pliocene simulations with varying levels of increased atmospheric carbon dioxide. Simulated surface energy fluxes were collected from those simulations and were used to calculate the ocean heat convergence/divergence at each grid cell. From the convergences we calculated the implied ocean heat transports which would have been necessary to maintain the specified SST distribution; in this case the SSTs are those derived from Pliocene paleo observations. Figure 6 shows the poleward heat transports from this series of Pliocene experiments. The plot reveals that CO 2levels must be four times current values, and perhaps higher, before ocean heat transports could be reduced to modern levels. At lower levels of atmospheric CO 2the ocean heat transports must remain higher than modern in order to maintain anything close to the observed Pliocene SSTs.

Estimates based on carbon isotope measurements (Raymo and others, 1992; 1996) indicate that Pliocene atmospheric CO 2levels were, at most, 100 ppm greater than today. Moreover, if we compare Pliocene and modern ocean heat transport distributions (Figure 5) we find that a poleward shift in the peak ocean heat convergence would have been necessary to balance the Pliocene SSTs regardless of the CO 2level. Thus, neither simulation results or data support the conclusion that Pliocene warming was caused entirely by a large increase in atmospheric CO 2content. We cannot rule out, however, that some combination of the altered CO 2and altered ocean heat transport caused the warmer climate of the middle Pliocene.

Figure 6: The implied ocean heat transports calculated by the climate model are dependent upon the amont of CO2 in the atmosphere.

Concluding Statements

Simulating past warm climates and identifying model/data contrasts for periods such as the Pliocene provide a test of the sensitivity of our primary tool for study future climate change: global climate models. At present, our results do not support the suggestion that Pliocene warming was caused by carbon dioxide increase since such changes are not consistent with the SST distributions derived from deep sea cores. There is evidence that changes in ocean circulation and the amount of heat oceans transport may be one potential cause of the warming.

Still, investigators have found evidence that minor increases in CO 2(up to 380 ppm) did occur in the Pliocene. This causes us to wonder whether it is possible that an, climate feedback, as of yet unknown, associated with small increases in CO 2, could lead to the larger changes seen in the ocean circulation? Certainly the evidence for higher levels of CO 2and stronger thermohaline circulation challenges recent results from coupled ocean- atmosphere models, which suggest that thermohaline circulation weakens as global temperature rises. Perhaps the Pliocene warming is uncharacteristic of next century's expected warming, perhaps the causes are different but the effects will be similar, and perhaps the Pliocene is a warning that unkown factors still exist that could exacerbate or mitigate the CO 2increase and global warming.

Successful comparisons, while increasing our confidence in the basic approach, probably occur coincidentally in some cases and such errors would be difficult to identify. Nevertheless, mismatches between data interpretations and model results offer undeniable evidence that either the model, data, or both are innacurrate for a specific region and climate variable. Understanding this allows us to focus resources and efforts on areas that are likely to afford the most gain. Moreover, subsequent iterations, based on new treatments of the data or GCM, test the veracity of previous conclusions.

The GISS Pliocene GCM simulation and the PRISM reconstructions are a first step in the interative process of data collection and analysis, model experimentation and analysis, and data/model comparison; the gridded, boundary condition data sets are continuously being refined, updated, and extended into areas with scarce data. Additional modeling and sensitivity experiments involving new data sets and updated GCM versions will soon begin. Close cooperation between modeling and data groups can achieve an overall better understanding of global climate models, data, data collection and simulation strategies, and the climate changes our society and planet could face relatively soon.


Barrett, P.J., C.J. Adams, W.C. McIntosh, C.C. Swisher, III, and G.S. Wilson 1992, Geochronological evidence supporting Antarctic deglaciation three million years ago, Nature 359 , 816-818.

Barron, J.A. 1992. Pliocene paleoclimatic interpretation of DSDP Site 580 (NW Pacific) using diatoms. Marine Micropaleontol. 20 , 23-44.

Chandler, M. 1993. GCM simulations of the pliocene climate: Feedbacks, ocean transports, and CO 2. In Pliocene Terrestrial Environments and Data/Model Comparisons (R. Thompson, Ed.). USGS Open File Report No. 94-23.

Chandler, M. 1994. Global climate model simulations of the Middle Pliocene, continued . In Pliocene High-latitude Climate Records (S. Ishman, Ed.). USGS Open File Report No. 94-588.

Chandler, M.A., D. Rind, and R.S. Thompson 1994. Joint investigations of the middle Pliocene climate II: GISS GCM Northern Hemisphere results .Palaeogeogr. Palaeoclimatol. Palaeoecol. 9, 197-219.

Crowley, T.J. 1991. Modeling Pliocene warmth. Quaternary Sci. Rev. 10 , 275-282.

Dowsett, H.J., and R.Z. Poore 1991. Pliocene sea surface temperatures of the North Atlantic Ocean at 3.0 Ma. Quaternary Sci. Rev. 10 , 189-204.

Dowsett, H.J., T.M. Cronin, R.Z. Poore, R.S. Thompson, R.C. Whatley, and A.M. Wood 1992. Micropaleontological evidence for increased meridional heat transport in the North Atlantic Ocean during the Pliocene. Science 258 , 1133-1135.

Dowsett, H.J., R.S. Thompson, J.A. Barron, T.M. Cronin, S.E. Ishman, R.Z. Poore, D.A. Willard, and T.R. Holtz, Jr. 1994. Joint investigations of the Middle Pliocene climate I: PRISM palaeoenvironmental reconstructions. Palaeogeogr. Palaeoclimatol. Palaeoecol. 9, 169-195.

Matthews, J.V., and L.E. Ovenden 1990. Late Tertiary plant macrofossils from localities in Arctic/Subarctic North America: A review of the data. ARCTIC 43 , 364-392.

Raymo, M.E., D. Hodell, and E. Jansen 1992. Response of deep ocean circulation to initiation of Northern Hemisphere glaciation. Paleoceanogr. 7, 645-672.

Raymo, M.E., B. Grant, M., Horowitz, and G.H. Rau 1996. Mid-Pliocene warmth: Stronger greenhouse and stronger conveyor. Marine Micropaleontol. 27 , 313-326.

Rind, D., and M.A. Chandler 1991. Increased ocean heat transports and warmer climate .J. Geophys. Res. 96 , 7437-7461.

Webb, P.N., and D.M. Harwood 1991. Late Cenozoic glacial history of the Ross Embayment, Antarctica. Quaternary Sci. Rev. 10 , 215-223.

Figure 7:

EdGCM Logo