Deep Oceans Are Cooling Amidst A Sea of Modeling Uncertainty: New Research on Ocean Heat Content

by Jim Steele 

JULY 21, 2014

Two of the world’s premiere ocean scientists from Harvard and MIT have addressed the data limitations that currently prevent the oceanographic community from resolving the differences among various estimates of changing ocean heat content (in print but available here).3 They point out where future data is most needed so these ambiguities do not persist into the next several decades of change. As a by-product of that analysis they 1) determined the deepest oceans are cooling, 2) estimated a much slower rate of ocean warming, 3) highlighted where the greatest uncertainties existed due to the ever changing locations of heating and cooling, and 4) specified concerns with previous methods used to construct changes in ocean heat content, such as Balmaseda and Trenberth’s re-analysis (see below).13 They concluded, “Direct determination of changes in oceanic heat content over the last 20 years are not in conflict with estimates of the radiative forcing, but the uncertainties remain too large to rationalize e.g., the apparent “pause" in warming.”




Wunsch and Heimbach (2014) humbly admit that their “results differ in detail and in numerical values from other estimates, but the determining whether any are “correct" is probably not possible with the existing data sets.”

They estimate the changing states of the ocean by synthesizing diverse data sets using models developed by the consortium for Estimating the Circulation and Climate of the Ocean, ECCO.   The ECCO “state estimates” have eliminated deficiencies of previous models and they claim, “unlike most “data assimilation" products, [ECCO] satisfies the model equations without any artificial sources or sinks or forces. The state estimate is from the free running, but adjusted, model and hence satisfies all of the governing model equations, including those for basic conservation of mass, heat, momentum, vorticity, etc. up to numerical accuracy.”

Their results (Figure 18. below) suggest a flattening or slight cooling in the upper 100 meters since 2004, in agreement with the -0.04 Watts/m2 cooling reported by Lyman (2014).6 The consensus of previous researchers has been that temperatures in the upper 300 meters have flattened or cooled since 2003,4 while Wunsch and Heimbach (2014) found the upper 700 meters still warmed up to 2009.

The deep layers contain twice as much heat as the upper 100 meters, and overall exhibit a clear cooling trend for the past 2 decades. Unlike the upper layers, which are dominated by the annual cycle of heating and cooling, they argue that deep ocean trends must be viewed as part of the ocean’s long term memory which is still responding to “meteorological forcing of decades to thousands of years ago”. If Balmaseda and Trenberth’s model of deep ocean warming was correct, any increase in ocean heat content must have occurred between 700 and 2000 meters, but the mechanisms that would warm that “middle layer” remains elusive.



The detected cooling of the deepest oceans is quite remarkable given geothermal warming from the ocean floor. Wunsch and Heimbach (2014) note, “As with other extant estimates, the present state estimate does not yet account for the geothermal flux at the sea floor whose mean values (Pollack et al., 1993) are of order 0.1 W/m2,” which is small but “not negligible compared to any vertical heat transfer into the abyss.3   (A note of interest is an increase in heat from the ocean floor has recently been associated with increased basal melt of Antarctica’s Thwaites glacier. ) Since heated waters rise, I find it reasonable to assume that, at least in part, any heating of the “middle layers” likely comes from heat that was stored in the deepest ocean decades to thousands of years ago.

Wunsch and Heimbach (2014) emphasize the many uncertainties involved in attributing the cause of changes in the overall heat content concluding, “As with many climate-related records, the unanswerable question here is whether these changes are truly secular, and/or a response to anthropogenic forcing, or whether they are instead fragments of a general red noise behavior seen over durations much too short to depict the long time-scales of Fig. 6, 7, or the result of sampling and measurement biases, or changes in the temporal data density.”

Given those uncertainties, they concluded that much less heat is being added to the oceans compared to claims in previous studies (seen in the table below).  It is interesting to note that compared to Hansen’s study that ended in 2003 before the observed warming pause, subsequent studies also suggest less heat is entering the oceans. Whether those declining trends are a result of improved methodologies, or due to a cooler sun, or both requires more observations. 



Years Examined


9Hansen 2005


0.86 +/- 0.12

5Lyman 2010


0.64 +/- 0.11

10von Schuckmann 2011


0.54 +/- 0.1

3Wunsch 2014


0.2 +/- 0.1


No climate model had predicted the dramatically rising temperatures in the deep oceans calculated by the Balmaseda/Trenberth re-analysis,13 and oceanographers suggest such a sharp rise is more likely an artifact of shifting measuring systems. Indeed the unusual warming correlates with the switch to the Argo observing system. Wunsch and Heimbach (2013)2 wrote, “clear warnings have appeared in the literature—that spurious trends and values are artifacts of changing observation systems (see, e.g., Elliott and Gaffen, 1991; Marshall et al., 2002; Thompson et al., 2008)—the reanalyses are rarely used appropriately, meaning with the recognition that they are subject to large errors.3

More specifically Wunsch and Heimbach (2014) warned, “Data assimilation schemes running over decades are usually labeled “reanalyses.” Unfortunately, these cannot be used for heat or other budgeting purposes because of their violation of the fundamental conservation laws; see Wunsch and Heimbach (2013) for discussion of this important point. The problem necessitates close examination of claimed abyssal warming accuracies of 0.01 W/m2 based on such methods (e.g., Balmaseda et al., 2013).” 3   

So who to believe?

Because ocean heat is stored asymmetrically and that heat is shifting 24/7, any limited sampling scheme will be riddled with large biases and uncertainties. In Figure 12 below Wunsch and Heimbach (2014) map the uneven densities of regionally stored heat. Apparently associated with its greater salinity, most of the central North Atlantic stores twice as much heat as any part of the Pacific and Indian Oceans. Regions where there are steep heat gradients require a greater sampling effort to avoid misleading results. They warned, “The relatively large heat content of the Atlantic Ocean could, if redistributed, produce large changes elsewhere in the system and which, if not uniformly observed, show artificial changes in the global average.” 3




Furthermore, due to the constant time-varying heat transport, regions of warming are usually compensated by regions of cooling as illustrated in their Figure 15. It offers a wonderful visualization of the current state of those natural ocean oscillations by comparing changes in heat content between1992 and 2011. Those patterns of heat re-distributions involve enormous amounts of heat and that make detection of changes in heat content that are many magnitudes smaller extremely difficult. Again any uneven sampling regime in time or space, would result in “artificial changes in the global average”.

Figure 15 shows the most recent effects of La Nina and the negative Pacific Decadal Oscillation. The eastern Pacific has cooled, while simultaneously the intensifying trade winds have swept more warm water into the western Pacific causing it to warm. Likewise heat stored in the mid?Atlantic has likely been transported northward as that region has cooled while simultaneously the sub?polar seas have warmed. This northward change in heat content is in agreement with earlier discussions about cycles of warm water intrusions that effect Arctic sea ice, confounded climate models of the Arctic and controls the distribution of marine organisms.

Most interesting is the observed cooling throughout the upper 700 meters of the Arctic (Figure 15 below). There have been 2 competing explanations for the unusually warm Arctic air temperature that heavily weights the global average. CO2 driven hypotheses argue global warming has reduced polar sea ice that previously reflected sunlight, and now the exposed dark waters are absorbing more heat and raising water and air temperatures. But clearly a cooling upper Arctic Ocean suggests any absorbed heat is insignificant. Despite greater inflows of warm Atlantic water, declining heat content of the upper 700 meters supports the competing hypothesis that warmer Arctic air temperatures are, at least in part, the result of increased ventilation of heat that was previously trapped by a thick insulating ice cover.7 That second hypothesis is also in agreement with extensive observations that Arctic air temperatures had been cooling in the 80s and 90s. Warming occurred after subfreezing winds, re?directed by the Arctic Oscillation, drove thick multi-year ice out from the Arctic.11

Regional cooling is also detected along the storm track from the Caribbean and along eastern USA. This evidence contradicts speculation that hurricanes in the Atlantic will or have become more severe due to increasing ocean temperatures. This also confirms earlier analyses of blogger Bob Tisdale and others that Superstorm Sandy was not caused by warmer oceans.



In order to support their contention that the deep ocean has been dramatically absorbing heat, Balmaseda/Trenberth must provide a mechanism and the regional observations where heat has been carried from the surface to those depths. But few are to be found. Warming at great depths and simultaneous cooling of the surface is antithetical to climate models predictions. Models had predicted global warming would store heat first in the upper layer and stratify that layer. Diffusion would require hundreds to thousands of years, so it is not the mechanism. Trenberth, Rahmstorf, and others have argued the winds could drive heat below the surface. Indeed winds can drive heat downward in a layer that oceanographers call the “mixed-layer,” but the depth where wind mixing occurs is restricted to a layer roughly 10-200 meters thick over most of the tropical and mid-latitude belts. And those depths have been cooling slightly.

The only other possible mechanism that could reasonably explain heat transfer to the deep ocean was that the winds could tilt the thermocline. The thermocline delineates a rapid transition between the ocean’s warm upper layer and cold lower layer. As illustrated above in Figure 15, during a La Nina warm waters pile up in the western Pacific and deepens the thermocline. But the tilting Pacific thermocline typically does not dip below the 700 meters, if ever.8

Unfortunately the analysis by Wunsch and Heimbach (2014) does not report on changes in the layer between 700 meters and 2000 meters. However based on changes in heat content below 2000 meters (their Figure 16 below), deeper layers of the Pacific are practically devoid of any deep warming.



The one region transporting the greatest amount of heat into the deep oceans is the ice forming regions around Antarctica, especially the eastern Weddell Sea where annually sea ice has been expanding.12 Unlike the Arctic, the Antarctic is relatively insulated from intruding subtropical waters (discussed here) so any deep warming is mostly from heat descending from above with a small contribution from geothermal.

Counter?intuitively greater sea ice production can deliver relatively warmer subsurface water to the ocean abyss. When oceans freeze, the salt is ejected to form a dense brine with a temperature that always hovers at the freezing point. Typically this unmodified water is called shelf water. Dense shelf water readily sinks to the bottom of the polar seas. However in transit to the bottom, shelf water must pass through layers of variously modified Warm Deep Water or Antarctic Circumpolar Water. Turbulent mixing also entrains some of the warmer water down to the abyss. Warm Deep Water typically comprises 62% of the mixed water that finally reaches the bottom. Any altered dynamic (such as increasing sea ice production, or circulation effects that entrain a greater proportion of Warm Deep Water), can redistribute more heat to the abyss.14 Due to the Antarctic Oscillation the warmer waters carried by the Antarctic Circumpolar Current have been observed to undulate southward bringing those waters closer to ice forming regions. Shelf waters have generally cooled and there has been no detectable warming of the Warm Deep Water core, so this region’s deep ocean warming is likely just re-distributing heat and not adding to the ocean heat content.

So it remains unclear if and how Trenberth’s “missing heat” has sunk to the deep ocean. The depiction of a dramatic rise in deep ocean heat is highly questionable, even though alarmists have flaunted it as proof of Co2’s power. As Dr. Wunsch had warned earlier, “Convenient assumptions should not be turned prematurely into ‘facts,’ nor uncertainties and ambiguities suppressed.” … “Anyone can write a model: the challenge is to demonstrate its accuracy and precision... Otherwise, the scientific debate is controlled by the most articulate, colorful, or adamant players.” 1

To reiterate, “the uncertainties remain too large to rationalize e.g., the apparent “pause" in warming.”



Literature Cited

1. C. Wunsch, 2007. The Past and Future Ocean Circulation from a Contemporary Perspective, in AGU Monograph, 173, A. Schmittner, J. Chiang and S. Hemming, Eds., 53-74

2. Wunsch, C. and P. Heimbach (2013) Dynamically and Kinematically Consistent Global Ocean Circulation and Ice State Estimates. In Ocean Circulation and Climate, Vol. 103. http://dx.doi.org/10.1016/B978-0-12-391851-2.00021-0 

3. Wunsch, C., and P. Heimbach, (2014) Bidecadal Thermal Changes in the Abyssal Ocean, J. Phys. Oceanogr., http://dx.doi.org/10.1175/JPO-D-13-096.1

4. Xue,Y., et al., (2012) A Comparative Analysis of Upper-Ocean Heat Content Variability from an Ensemble of Operational Ocean Reanalyses. Journal of Climate, vol 25, 6905-6929.

5. Lyman, J. et al, (2010) Robust warming of the global upper ocean. Nature, vol. 465,334-337.

6. Lyman, J. and G. Johnson (2014) Estimating Global Ocean Heat Content Changes in the Upper 1800m since 1950 and the Influence of Climatology Choice*. Journal of Climate, vol 27.

7. Rigor, I.G., J.M. Wallace, and R.L. Colony (2002), Response of Sea Ice to the Arctic Oscillation, J. Climate, v. 15, no. 18, pp. 2648 – 2668.

8. Zhang, R. et al. (2007) Decadal change in the relationship between the oceanic entrainment temperature and thermocline depth in the far western tropical Pacific. Geophysical Research Letters, Vol. 34.

9. Hansen, J., and others, 2005: Earth's energy imbalance: confirrmation and implications. Science, vol. 308, 1431-1435.

10. von Schuckmann, K., and P.-Y. Le Traon, 2011: How well can we derive Global Ocean Indicators from Argo data?, Ocean Sci., 7, 783-791, doi:10.5194/os-7-783-2011.

11. Kahl, J., et al., (1993) Absence of evidence for greenhouse warming over the Arctic Ocean in the past 40 years. Nature, vol. 361, p. 335?337, doi:10.1038/361335a0

12. Parkinson, C.  and D. Cavalieri (2012) Antarctic sea ice variability and trends, 1979–2010. The Cryosphere, vol. 6, 871–880.

13. Balmaseda, M. A., K. E. Trenberth, and E. Kallen, 2013: Distinctive climate signals in reanalysis of global ocean heat content. Geophysical Research Letters, 40, 1754-1759.

14. Azaneau, M. et al. (2013) Trends in the deep Southern Ocean (1958–2010): Implications for Antarctic Bottom Water properties and volume export. Journal Of Geophysical Research: Oceans, Vol. 118