My Recent Discussion with Gavin Schmidt On Real Climate

Over the last few weeks, I have been engaged to a series of exchanges on the weblogs Skeptical Science and Real Climate. In this post I want to summarize the science issues that were discussed on Real Climate particularly with Gavin Schmidt. I will post separately on my interactions on Skeptical Science in the next few days.

In terms of Real Climate, in their post

Global warming and ocean heat content

I posted

Response To Gavin Schmidt’s Post Of October 3 2011 “Global Warming And Ocean Heat Content”

The comments and replies by Gavin Schmidt (and others) on the Real Climate post subsequent to my above weblog post are informative. I have reproduced several below.

The main conclusion I have reached from the series of comments/replies  are:

1. The ocean heat content change is the robust methodology to diagnose global warming and cooling.

2. To provide the most robust measure of the heating and cooling, the heat at levels below 700m needs to be considered.

3. The odd sudden increase in heating in about 2004, whereas none was reported previously, is very puzzling. It appears just in time to explain the relatively small heating in the upper 700m. Does this mean the physical response of the ocean to heating has changed, or is it a data issue?

4. There remains a lack of clarity as to whether Argo can track heat if it transfers through the upper 700m into the deeper depths. There is no doubt, however, that if this heat is from the TOA radiative imbalance (i.e. from added CO2 and other greenhouse gases) the heat must travel through this layer.  Gavin Schmidt has stated that “Heat transfer will be mainly continuous, not episodic”.  I challenged him to support this conclusion.

5. Currently, in the Argo network, there are roughly 3000 floats producing 100,000 temperature/salinity profiles per year. The floats go as deep as 2000m. If heat is transported on shorter time periods through the entire upper 700m, than  it could be missed in the sampling. However, if the transport is slower than can be sampled with ~33 profiles per year (~ once every 11 days) than it will be sampled. Analyses such as provided by the ECMWF suggest the transfers are not continuos as Gavin concludes but are in globs; e.g. see

Gavin – I am glad you noticed my post

Torpedoing Of The Use Of The Global Average Surface Temperature Trend As The Diagnostic For Global Warming

We seem to disagree on several points. First, you write

“The second point is related to a posting by Roger Pielke Sr last week, who claimed that the Meehl et al paper ‘torpedoed’ the use of the surface temperature anomaly as a useful metric of global warming. This is odd in a number of respects. First, the surface temperature records are the longest climate records we have from direct measurements and have been independently replicated by multiple independent groups. I’m not aware of anyone who has ever thought that surface temperatures tell us everything there is to know about climate change, but nonetheless in practical terms global warming has for years been defined as the rise in this metric. It is certainly useful to look at the total heat content anomaly (as best as it can be estimated), but the difficulties in assembling such a metric and extending it back in time more than a few decades preclude it from supplanting the surface temperatures in this respect.”

However, we now have a robust way to diagnose upper ocean heat content, we should move to that metric, starting in ~2003, as the primary metric to monitor global warming.

[Response:The idea that there can be only one metric has no basis in anything. Every new stream of information is useful in building up a picture of what is happening – some records have longevity, others have depth, some are regional, some are global. Your desire to dethrone or torpedo the global surface temperature records is merely rhetorical (unless you are seriously suggesting that we stop monitoring the surface temperatures? Surely not!). – gavin]

Also, NASA, GISS and CRU analyze their data differently, but they have a large overlap in their raw data; see my post

Erroneous Information In The Report “Procedural Review of EPA’s Greenhouse Gases Endangerment Finding Data Quality Processes”

[Response:More semantics: – the *analyses* of the raw data are independent, and it is easy to show that you get the same basic trend with completely independent subsets of the data. – gavin]

You write

“the surface temperature records …. have been independently replicated by multiple independent groups”.

This is not correct.

[Response: Yes, it actually is. And the Berkeley effort will show it again. – gavin]

Finally, you write

Obviously heat going below 700m must have passed through the upper ocean. However, the notion that Argo could see this is odd. Argo measures temperature, not flux. The net flux into a layer is calculated by looking at the change in temperature. It cannot tell you how much came in at the top and left at the bottom, only how much remained. – gavin]”.

You are, of course, correct that Argo measures temperatures, but unless you can show the temporal sampling period is too long, or the spatial sampling is too sparse, the downward movement of heat would be seen in positive temperature anomalies as they move towards lower depth. Similarly, if this heat were to re-emerge, we would also see it as the anomalies move upwards.

[Response: This is a continuous process – not lumps of anomalous heat that can be tracked individually. – gavin]

Also, if there is large amounts of heat being stored at depth in the ocean, this means that the global annual average surface temperature trend is not sampling this heat. This surface temperature trend would be underestimating global warming.

[Response:Semantics: You are redefining ‘global warming’ to something different to what anyone else thinks and then claiming that the standard measure of global warming (as understood by everyone else) is not being properly sampled. I can redefine apple to mean an orange, and then claim that people shouldn’t just bite into apples because of the skin. It might make sense logically, but as a method of communicating a fact to an audience, it is woeful. Words do not mean just what *you* say they mean. – gavin]

Gavin –

[Response: Yes, it actually is. And the Berkeley effort will show it again. – gavin]

The Berkeley analysis is a more indepedent assessment. We agree on that, and that it supports the GISS/NCDC/CRU trend findings with respect to the mean. This still does not make the GISS/NCDC/CRU independent.

The issue of why you persist in retaining the surface temperature trend as the primary metric of global warming is a puzzle to me. We, of course, need surface temperatures for a wide variety of other reasons. However, if significant heat is being transported to deeper depths, I assume you would agree that the surface temperature trend would underestimate global warming and influence the calculaiton of “climate sensitivity”. But let us know if you disagree and why.

[Response:“Climate sensitivity” is classically defined as the change in surface temperature as a response to radiative forcing. It is an equilibrium concept that is almost completely divorced from the flux of heat into the deep ocean. One could define a new concept – “total heat content sensitivity” (in J per W/m2 – odd unit) which would be related to the standard sensitivity, but also to the ocean mixing processes. Given that concept one could attempt to estimate it from observations and diagnose it in models and one could try and make a case that this was somehow more relevant in terms of impacts or vulnerability. All of these things could be done. But, as far as I am aware, none of them have. Thus, the standard climate sensitivity remains the focus of attention. I would suggest that if you want to change that, you should embark on the steps I gave above rather than simply co-opting language and changing standard definitions. So, to directly answer your question, since surface temperature changes define global warming, they cannot underestimate it. If you really mean to say that surface temperature increases don’t tell you much about deep ocean heat content changes, then this is of course true. But in that case I’m not sure what point you are trying to make. – gavin]

Gavin – You write

surface temperature changes define global warming”.

Here is where we have a fundamental disagreement. Global warming is defined by the accumulation of heat in the units of Joules. Surface temperature changes by itself is not heat.

[Response:I am well aware that temperature is a different quantity than heat, and have no objection to people tracking the accumulation of heat, but ‘global warming’ is simply not defined in this way. This is not a ‘fundamental disagreement’, this is simply you redefining the term ‘global warming’. For me (and almost anyone else you care to ask) global warming refers to the increase in global surface temperature anomaly. Indeed, ‘warmth’ is not a pure function of Joules – ice and water at 0 deg C have the same ‘warmth’, but very different heat contents. The very natural definition of warming is in terms of temperature; when people say that something has warmed, it means that the temperature has risen. You would be much more effective at communicating your scientific points if you used words in ways other people were already used to. – gavin]

Gavin – This is one reason why we have a different view of this issue. You write

“Indeed, ‘warmth’ is not a pure function of Joules – ice and water at 0 deg C have the same ‘warmth’, but very different heat contents”.

They do not have the same “warmth”, just the same temperature. There is more “warmth” with the liquid water. This is not semantics, but basic physics. If we want to properly monitor global warming, it must be in units of heat.

You are correct that when ” people say that something has warmed, it means that the temperature has risen”. However, when a scientist say that something has warmed, it means that the Joules have increased.

[Response:We’ll just have to agree to disagree then, because I agree with ‘people’ in this context. Warming means an increase in temperature for almost anybody you ask, and redefining it to mean something different just leads to confusion. And indeed, since I can find no antecedent for ‘climate sensitivity’ referring to anything else other than the global surface temperature change, I’m pretty confident that most scientists will agree. But regardless of what the common usage is, if you want to be understood clearly, I strongly suggest you define your terms every time you make a statement if you do not want to be misunderstood. For future reference, any time I use the phrase global warming, it is to be understood to refer to the increase in global mean surface temperatures. – gavin 

Gavin – If the temporal and spatial resolution are sufficient, all that is needed is to monitor the Joules of heat in the upper 700m to follow their upward and downward motion (which is given by the temperatures).

In the plot that you show of heating down to 2km [note: the plot is given in a comment below], do you have evidence that this heat was first in the upper 700m? All you need to do is show this being transported through that layer, and I will be convinced of the robustness of the heating . deeper than 700m.

There is no need to explicitly calculate the fluxes; their net effect (the flux divergence of heat) is accomplished by integrating the mass-weighted temperature anomalies over time. This is presumably what Jim did in the comment he provided.

[Response: You’ve lost me. How did heat content increase below 2000m without it passing through the top 700m? (I think we can safely assume that we don’t have a coincident increase in undersea geothermal heat flux!). – gavin]

Gavin – You wrote

“You’ve lost me. How did heat content increase below 2000m without it passing through the top 700m? (I think we can safely assume that we don’t have a coincident increase in undersea geothermal heat flux!). – gavin]”

This is precisely my point. We should have seen this heat be transfered through the upper 700 meters. From my understanding, the temporal and spatial resolution in the upper 700m is good enough to see this (but we certainly need colleagues at JPL and elsewhere to tell us if it does not have this accuracy).

Below 700 meters, however, the reported warming is based on sparser data so the uncertainty in the plot you have shown for those depths is higher. What is the estimated uncertainty? Before accepting it as sufficiently accurate, we also need to look at the heat transfers through the upper 700 meters.

[Response:I think this is becoming a little repetitive. You have not given anyone a reason to think we can measure absolute downward heat flux at 700m (or elsewhere), nor have you given a reference for anyone performing such a calculation. In my opinion, such an estimate would be extremely difficult and the uncertainties would be so large as to make the attempt much less of a constraint than is required. But regardless of that problem, your request simply doesn’t make any sense for a problem dominated by (effective) diffusion. Take an anomalous temperature profile that is linear in z: T'(z) = T0 * (1 – z/700) above 700m and zero below. The downward diffusive flux is constant (K*T0/700) (assuming constant K) and equal to the flux in at the surface. How would you be able to ‘see’ the heat being transferred? There is no shortcut here!

As for the error bars on the 0-2000m numbers, these are provided in the data file from NODC, and I’ve added them to the plot I gave before. – gavin]

Gavin – Your plot did not appear.

[Response:Fixed. thanks. – gavin]

On your question regarding heat transfer downward, if the models show warming at depth, why don’t you show plots as to the magnitude of its fluxes over time and space through the upper 700m of the ocean. Then one could look at the observations to see if this flux is there with the same pattern and magnitude as the real world data.

I assume our disagreement is in the form of these fluxes. If they are diffuse and distributed across the upper oceans, I agree they would be hard to see in the Argo data. However, if this transfer occurs in globs associated with mesoscale and larger ocean circulation features (as suggested in the ECMWF data), we should clearly see this movement of heat.

[Response:Model estimates of total heat flux through 700m are possible of course, but non-trivial to compute (including effects of the resolved circulation, isopycnal diffusion, and vertical mixing) – though this might well be worth doing for the CMIP5 models. However, I don’t have the answers handy. But I have no confidence that the observations will be sufficient to distinguish the anomalous heat flux from the climatological mean with sufficient precision to be helpful. If you think it is, please point to a study that has attempted this. – gavin]

I am still waiting to hear back, but this text from the Argo website implies that monitoring the vertical, as well as the horizontal fluxes of ocean heat, is major focus of the Argo upgrade. On their website they write

“Lack of sustained observations of the atmosphere, oceans and land have hindered the development and validation of climate models. An example comes from a recent analysis which concluded that the currents transporting heat northwards in the Atlantic and influencing western European climate had weakened by 30% in the past decade. This result had to be based on just five research measurements spread over 40 years. Was this change part of a trend that might lead to a major change in the Atlantic circulation, or due to natural variability that will reverse in the future, or is it an artifact of the limited observations?

In 1999, to combat this lack of data, an innovative step was taken by scientists to greatly improve the collection of observations inside the ocean through increased sampling of old and new quantities and increased coverage in terms of time and area.

That step was Argo.”

[Response:This is not related to our discussion. Instead it refers to the geostrophic calculations of Bryden et al (2005) and the subsequent realisation that the deep ocean circulation changes were being aliased (Cunningham et al, 2007). All of these calculations are of the horizontal flow (via the geostrophic relationships), and with vertical fluxes being implied as a residual. – gavin]

From they also write on this website

“It will provide a quantitative description of the changing state of the upper ocean and the patterns of ocean climate variability from months to decades, including heat and freshwater storage and transport.”

“Currently, there are roughly 3000 floats producing 100,000 temperature/salinity profiles per year. The floats go as deep as 2000m”

If heat is transported on shorter time periods through the entire upper 700m, than I agree it could be missed in the sampling. However, if the transport is slower than can be sampled with ~33 profiles per year (~ once every 11 days) than it will be sampled.

[Response:Heat transfer will be mainly continuous, not episodic. – gavin]

Response: Heat transfer will be mainly continuous, not episodic. – gavin]

Gavin – How do you know this?

On your earlier response, the key text is

In 1999, to combat this lack of data, an innovative step was taken by scientists to greatly improve the collection of observations inside the ocean through increased sampling of old and new quantities and increased coverage in terms of time and area.”

The question is whether the improved network can track heat vertical transfers.

Bart Verheggen – Good summary, except if there is no heat accumulation in the upper 700m as it diffuses slowly downward, we should still see a slight elevation in the temperature anomalies IF the Argo data precision is good enough. I do not know the precision of the temperature data measurements, and hope someone else can answer that.

[Response: Huh? If 0-700m temperature anomalies continue to increase, so will heat content. What are trying to say here? – gavin]

 In one post Bryan S. effectively summarizes the issue when he writes
Bryan S says:

Folks, several of you are taking Roger Pielke’s comments way out of context! I hope this in not intentional.

The physics here is a simple matter in concept.

For illustration purposes:

If the TOA radiative imbalance is 1 W/m^2, and the downward flux of heat below 700 meters is only 0.2 W/m^2, then there must be positive flux of heat into the 0-700 integral of 0.8 W/m^2. If the TOA radiative imbalance is real, then the upper ocean cannot have a “flat” heat content unless the downward flux of heat below 700 meters=1 W/m^2 (ignoring of course the smaller reservoirs of heat and interannual weather noise for illustration purposes).

No empirical data or model output that I am aware of suggests that any downward flux of heat below 700 meters is close to the the modeled TOA radiative imbalance averaged over a decadal period. Therefore, what Roger is saying is exactly correct.

The Palmer et al., 2011 paper cited above states that the behavior of several examined models leads to the conclusion that the range of uncertainty in the absolute radiative imbalance estimates might be reduced by as much as 30% by integrating the full volume of ocean below 700 meters, but this does not change what I have just stated above. This still implies that the bulk of the radiative imbalance must be seen in the upper ocean over decadal time periods.

Maybe further research will show these assumptions must change radically. While the Palmer results suggest the need for further research, as it stands now, the simple working model I have outlined above seems like a good working assumption.


First, I stated that the Argo data density was fine enough to see the movement of the heat downward, but am now unclear on this, and look forward to an Argo specialist to give us an overview of capability in this regards.

On the other issue you raised, you wrote

“it is good to see that he recognizes that heat can be sequestered in the deeper ocean, and that if it is then the surface temperature record is probably underestimating the amount of warming. And yes, I realize that him saying that is inconsistent with his research that claims the surface temperature record has a warm bias, especially at nighttime.”

I have always recognized that heat can be transfered to greater depth. In my paper

Pielke Sr., R.A., 2003: Heat storage within the Earth system. Bull. Amer. Meteor. Soc., 84, 331-335.

“An assessment of the heat storage within the earth.s climate system offers a unique perspective on global change. If the heat actually remains within the earth system in the deeper ocean, for example, while the heat content of the remainder of the heat reservoirs in the earth system remains unchanged, sudden transfers of the heat between components of the system (from the ocean into the atmosphere) could produce rapid, unanticipated changes in global weather.

Similarly, relatively large warming and cooling radiative forcings (e.g., well-mixed greenhouse gases and the indirect effect of aerosols) could be in near balance at present, suggesting that sudden climate changes could occur if one of these forcings becomes dominant. On the other hand, a loss of space to a large portion of the increased radiative fluxes, as the atmosphere adjusts, such as through a change in cloud cover (e.g., Lindzen et al. 2001), would suggest that the climate system is relatively more resilient to continued anthropogenic heating effects than conventionally assumed.”

My view now is that if the heat moves to deeper depths, it would become relatively dispersed, and I do not see how it could be quickly transfered back to the surface.

On the warm nighttime bias we found, this is still a robust finding between the MSU LT data and the surface temperature data. It is an independent issue from the ocean heating.

The sequestration of the heat at depth, however, if this is accurate, does show a serious inadequacy in using the global annual average surface temperature trend as the metric to diagnose global warming.

[Response:Isn’t this where we started? – gavin]

Gavin – Regarding

Isn’t this where we started? – gavin

I have found the set of comments quite informative, so I do not feel we are back where we started. Bryan S has summarized the issue very well.

Thank you for the opportunity to interact on your weblog. Roger

I recommend reading all of the other exchanges also, such as

flxible – If one is using the annual global average surface temperature trend to diagnose global warming (and cooling), but some of the heat is being sequestered in the deep ocean, the trend would be underestimated when using the surface temperature trend to monitor global warming.

This issue provides yet another reason we should move to the ocean heat content changes as the primary metric to diagnose global warming as as a check against the time integrated TOA radiative imbalance as measured from satellite.


Would using ocean heat content changes be a satisfactory primary metric for diagnosing heat storage changes within the Earth system?”

The answer is yes. I have heard Jim Hansen say the same thing at one of NRC (2005) committee meetings. This is a suggestion for the present (2003 and later when the Argo network obtain their planned coverage and into the future.

P.S. I disagree that the land temperatures are well sampled, but that is an issue for a different thread.

Your related questions are clearly presented and answers needed.

Comments Off on My Recent Discussion with Gavin Schmidt On Real Climate

Filed under Climate Change Metrics, Q & A on Climate Science

Comments are closed.