The missing heat

by Nick Foukal

For the past 15 years, the slope of the global-mean surface air temperature (GMST) record has plateaued, a change that almost none of the IPCC global climate models (GCMs) predicted. More recently, a couple of papers have come out pointing to the North Atlantic (specifically the subpolar gyre) as the primary location for the missing heat of our climate system [Chen and Tung, 2014; Drijfhout et al., 2014]. The argument goes that because anthropogenic forcing (i.e. greenhouse gases) has increased over this period, yet GMST has not continued to increase, then the oceans must be storing the excess buy diflucan online heat. The North Atlantic is one of the few ‘deep water formation’ sites, thus it is a good candidate to absorb the heat. But let’s quickly review some background information and then the evidence for the North Atlantic.


– Although almost none of the IPCC models predicted this slowdown in global warming, the SAT has never deviated outside of the ensemble range of the models, and 2014 (the warmest year on record) falls directly on the slope of warming from 1970-2000. Some climate scientists thus consider the hiatus as nothing more than noise about a warming trend, while others view it as proof that our GCMs are flawed.

– GMST is one indicator of whether the Earth is warming, but it is not sufficient by itself. Specifically, GMST is a measure of the atmospheric temperature at 2 m above the Earth’s surface, averaged over the entire globe. But the Earth contains heat that is stored outside of that narrow band. The oceans store a majority of Earth’s heat (because of the high heat capacity of water), and other aspects of the climate system such as ice and the upper atmosphere also contain heat that is not directly measured in GMST.

– Despite many reports to the contrary, the existence and sign of a radiative imbalance at the top of the atmosphere is not completely understood. There are two ways to calculate this imbalance, one is by satellites, and the other is by the increase in ocean heat content (OHC). The former is not precise enough to calculate the sign of the imbalance, and the latter uses somewhat circular reasoning: we know the oceans are warming, so there must be more net radiation coming in, but the GMST isn’t increasing, so the ocean’s must be storing the excess heat. Despite these concerns, most estimates claim that about 342 W/m2 of solar radiation enter our atmosphere and about 341.5 W/m2 leave. And though the strength of the imbalance is debatable, there is a far greater chance that the net flux is directed into the atmosphere than out.

– Observations of the deep ocean are sparse, especially as you go back in time. The Chen and Tung paper shows data coverage on a 5° x 5° grid at monthly resolution (where a single measurement within that box during that month represents 100% coverage), and most of the ocean basins before 2000 (when the Argo project started) have less than 10% coverage. Due to the lack of historical data, it is very difficult to figure out if the oceans are currently taking up more heat than they did during periods when the GMST was increasing rapidly (1970-2000). The good news is that the Atlantic has the most data coverage, and Argo is really helping us get better coverage on a global level in the last decade (at least down to 2000 m depth).

The importance of the North Atlantic

The North Atlantic is one of the few locations in the ocean where ‘deep water’ is formed, thus it connects the surface conditions to the deep ocean. In the North Atlantic, deep convection occurs during winter, mixing surface waters down to depths of greater than 1500 m [Yashaev and Loder, 2009]. Furthermore, the deep limb of the Atlantic Meridional Overturning Circulation (AMOC) may be able to distribute this heat throughout the Atlantic. The deep ocean is where the excess heat of the hiatus period is thought to be stored because the upper ocean temperatures do not seem to be warming. For example, we have very good data on sea-surface temperature, but its global-mean record is largely similar to GMST and also plateaus in the last 15 years. Similarly, many papers have shown the dependence of the GMST record on the El Niño-Southern Oscillation (ENSO) phase [e.g. Kosaka and Xie 2013]. But Chen and Tung [2014] correctly point out that ENSO varies on time scales shorter than our current hiatus (2-7 yr vs. 15 yr), and does not store heat deep enough (~300 m) to account for the hiatus in SAT. Furthermore, Douville et al. [2015] suggest that the GCMs overestimate the sensitivity of the GMST on ENSO variability and under-represent the role of the North Atlantic.

In contrast, the observational record points toward the North Atlantic as the more important ocean basin. Chen and Tung use EOF analysis of global OHC data from 1970-2012 for the upper layer (above 300 m) and then another for 300 m to 1500 m to show where the deep OHC is stored. In the upper layer (0-300 m), the Pacific is by far the dominant region, and the canonical zonal-dipole ENSO pattern across the equatorial Pacific is apparent. But in the deeper layers, the role of the Pacific is almost non-existent, whereas the Atlantic becomes very important. Perhaps the most important aspect of these results is the time series (PCs) that goes along with the EOFs. In the shallow EOF, the leading PC has no discernable trend and varies on time scales shorter than about 7 years. In contrast, the leading PC of the deep EOF analysis shows a distinct increase in OHC around 2000, when the hiatus started. Thus the authors claim that the North Atlantic is absorbing more heat in its deeper layers (300-1500 m) during the GMST hiatus than it was between 1970-2000. As mentioned above, there should be large warnings with this result that data coverage of OHC back to 1970 is sparse at best.

Rather than point to one basin or another, Drijfhout et al. [2014] quantify the role of each basin and claim that the North Atlantic is responsible for about 40% of the missing heat, the Pacific about 30%, and the Southern and Indian Ocean combined constitute the remaining 30%. The authors use multiple data sets and find that the oceans have taken up 0.7 W/m2 more heat from 2001-2009 than they did between 1992-2000. Mechanistically, the authors propose that the North Atlantic absorbed more heat in the latter decade because of a slowdown of the AMOC. This decrease in the AMOC would have decreased the upper limb heat transport and diminished the heat fluxes from the ocean to the atmosphere over the subpolar regions. Thus by losing less heat to the atmosphere than it usually does, the North Atlantic actually stored more heat during the hiatus than during previous periods.

How do these results fit into OSNAP? Well first off, the moorings, floats and gliders will help increase our coverage of OHC. And in terms of questions answered by OSNAP, one aspect that I am working on in particular is how quickly the upper limb of the AMOC affects upper ocean temperatures. We have seen a cooling of the subtropical gyre when the AMOC at the RAPID line decreases [Cunningham et al., 2013], but what about the subpolar gyre? Are temperature anomalies advected from one gyre to the other? Another student in our lab, Sijia Zou, is studying the connection between the deep convection in the Labrador Sea to the strength of the AMOC – when deep convection occurs, where does the water go and how quickly does it move? Answering these questions (and many more!) will help us constrain the North Atlantic’s role in global climate and possibly its connection to the recent global warming hiatus.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.