This is just a brief note to point out that a few graphs that I have put together showing Ocean Heat Content changes in recent decades had an incorrect scaling for the GISS model data. My error was in assuming that the model output (which were in units W yr/m2) were scaled for the ocean area only, when in fact they were scaled for the entire global surface area (see fig. 2 in Hansen et al, 2005). Therefore, in converting to units of 1022 Joules for the absolute ocean heat content change, I had used a factor of 1.1 (0.7 x 5.1 x 365 x 3600 x 24 x 10-8), instead of the correct value of 1.61 (5.1 x 365 x 3600 x 24 x 10-8). This problem came to light while we were redoing this analysis for the CMIP5 models and from conversations with dana1981 at skepticalscience.com.
These graphs appeared in Dec 2009, May 2010, Jan 2011 and Feb 2012. In each case, I have replaced the graph with a corrected version while leaving a link to the incorrect version. Links to the figures will return the corrected image (and this is noted on the image itself). Where possible I used the data that were current at the time of the original post. Fortunately this only affects the figures used in these blog postings and not in any publications. Apologies for any confusion.
This figure shows the comparison using the most up-to-date observational products (NODC, PMEL):
The basic picture is unchanged – model simulations were able to capture the historical variance in OHC (as best we know it now – there remains significant structural uncertainty in those estimates). There are clear dips related volcanic eruptions (Agung, El Chichon, Pinatubo), and an sharp increase in the 1990s. Note that in GISS-EH (same AGCM but with a different ocean model) OHC increases at a slightly slower rate than seen with GISS-ER above. Looking at the last decade, it is clear that the observed rate of change of upper ocean heat content is a little slower than previously (and below linear extrapolations of the pre-2003 model output), and it remains unclear to what extent that is related to a reduction in net radiative forcing growth (due to the solar cycle, or perhaps larger than expected aerosol forcing growth), or internal variability, model errors, or data processing – arguments have been made for all four, singly and together.
Analyses of the CMIP5 models will provide some insight here since the historical simulations have been extended to 2012 (including the last solar minimum), and have updated aerosol emissions. Watch this space.
References
- J. Hansen, L. Nazarenko, R. Ruedy, M. Sato, J. Willis, A. Del Genio, D. Koch, A. Lacis, K. Lo, S. Menon, T. Novakov, J. Perlwitz, G. Russell, G.A. Schmidt, and N. Tausnev, "Earth's Energy Imbalance: Confirmation and Implications", Science, vol. 308, pp. 1431-1435, 2005. http://dx.doi.org/10.1126/science.1110252
Gavin, the 2009 and 2012 graphs don’t appear to have been corrected yet. On the graph you have shown for this post, why isn’t the model extended like the other graphs show?
[Response: try refreshing your cache? As for the linear extrapolation that is shown in the 2011 figure (and the others), but this one here is focussed on the actual model results. For the post 2003 period, the newer model runs with updated forcings will be more interesting, and we’ll be able to avoid the whole issue. – gavin
Thanks for the update and your help clarifying this issue, Gavin. It would be interesting to see how current model runs fare in comparison to the OHC observational data too (including to greater depths).
That must (have) hurt: https://www.realclimate.org/images/ohc11.jpg
[Response: not really. I learnt a long time ago that a) I’m not infallible, and b) that one should never get personally invested in the results of a model. When things work, one should always remain pleasantly surprised, when they don’t there is possibly a reason that can found – which may be interesting. This is why science is fun. – gavin]
For the non-scientist, Can you explain the blue-dotted line in the graph and the thin black line/ I assume those are overall global temperature. Thanks for the website. I work on climate change policy for a local government and your website allows me to keep up with the science conversation (as much as I understand it).
[Response: Sorry! the blue dotted lines are individual simulations, while the heavy blue line is the ensemble average. The thin black line is the seasonal variability (ie. OHC every 3 months) from the observations, the thick black line is the annual average. – gavin]
Gavin, I can see the updated note under the graphs but the ‘correct’ and ‘incorrect’ graphs look identical.
[Response: Not sure what happened there, but I’ve now fixed it. – gavin]
I thought the purpose of this post was to point out an error/ discuss changes in graphs yet you placed a new graph in the post which show “the basic picture unchanged”. I hope everyone here actually clicks on the link to the Jan2010 OHC graph and then clicks on the ‘incorrect’ graph to see the difference. I wish you had put those two graphs in this post. It would have been more informative. I do appreciate your quick response.
Hi Gavin,
You say that new post 2003 model runs will show heat content based on updated forcings. It seems to me the rate of heat uptake should depend mainly on the ocean surface temperature trend, not the applied forcing, per se.
[Response: Not really. The uptake is a function of the ocean model advection and diffusion, which affects the surface temperatures of course but the forcing also does. Thus the SST trend is not predictive of the OHC change on its own. – gavin]
I mean, if you could accurately know the next 20 years of ocean surface temperature trend, it ought to be possible to make a very accurate prediction of OHC trajectory over that period, independent of what forcing would lead to that temperature trend.
[Response: No. small heat capacity + small forcing could give the same temperature trend as large heat capacity and large forcing (and everything inbetween). – gavin]
Which makes me think that you are hinting the updated forcings (rising aerosol influence) will lead to a lower trend in the modeled ocean surface temperature post 2003. Is that right? When will the post 2003 model trend be released?
[Response: The CMIP5 models are already available for download at PCMDI. – gavin]
If i understand it well, the discrepency between models and observations is about 10^23 J ?
[Response: Not really. First these aren’t ‘the models’, it is ‘a model’, second, the discrepancy depends on the specific run even in this small sample, third, linear extrapolation of a model is not really ‘a model’ in the same sense. – gavin]
Thanks for your answer Gavin.
I was merely using the word you were yourself using in your post here :
https://www.realclimate.org/index.php/archives/2012/02/2011-updates-to-model-data-comparisons/
“Figure 2 is the comparison of the ocean heat content (OHC) changes in the models compared to the latest data from NODC.”
It seems that when it agrees it’s “the models”, and when it doesn’t it’s just “a model”.
Anyway, what do other models give for 2010 values then ? Because here the value I read it 10^23 J, and it’s a lot (4 times the CO2 radiative forcing).
[Response: There was a multiple model comparison in Dominigues et al, 2008. But 10^23 J is a total energy, CO2 forcing is a flux. They aren’t comparable. – gavin]
Ok for Domingues, they seem more in line.
What is your 5.1 in “5.1 x 365 x 3600 x 24 x 10-8”.
Is it Stefan constant corrected for emissivity ?
[Response: Surface area of the Earth: 5.1 x 10^14 m^2. – gavin]
Thanks for the update and correction, Gavin; the OHC graphs have been a source of conversation in the past. Could you extend the ensemble mean as you did with the previous versions? Or is there a reason why you haven’t?
[Response: the extension of the ensemble mean can be seen in the 2011 corrected graph. – gavin]
Gavin – Please (you or Jim) update Jim’s diagnosed value of observed upper ocean heat storage in Watts per meter squared that he presented in this communication – http://pielkeclimatesci.files.wordpress.com/2009/09/1116592hansen.pdf.
He wrote
“Our simulated 1993-2003 heat storage rate was 0.6 W/m2 in the upper
750 m of the ocean. The decadal mean planetary energy imbalance, 0.75 W/m2, includes heat storage in the deeper ocean and energy used to melt ice and warm the air and land. 0.85 W/m2 is
the imbalance at the end of the decade.”
Roger
[Response: There is nothing incorrect about that statement – this was the result reported in Hansen et al (2005) using the GISS-ER model in CMIP3 – and so there is nothing to update. We will be reporting on the new CMIP5 simulations soon. – gavin]
Gavin, one of things I tell non-scientists to watch for in distinguishing real science from fake scepticism is that real scientists make mistakes and are happy to correct them; fake sceptics stick to an argument no matter how many holes are knocked in it. What I like about this site is that no matter what you argue, the science is the ultimate decider. Denial blogs are tediously full of people who will argue any point, no matter how absurd, and only attack positions that disagree with their prejudice.
Concerning the Model-Reality-Comparision for the depth up to 2000 m:
Is it right that the “reality”-graph for the deep ocean does not show the anomaly wrt. 1975-1989? Rather it should be shifted upwards by the (unknown) heat uptake between 700 and 2000 m from 1975-1989 to 2005 (according to the GISS-ER-model by about 4*10²² J).
[Response: The estimates of OHC change down to 2000m are actually quite uncertain and the pre-Argo error estimates are quite tricky. So while the model estimate of the fraction of the total OHC that is happening in 0-700m is interesting (about 80%), it is not definitive and varies quite a lot across models. – gavin]
Why just to 700m? The infamous “tritium experiment” clearly showed mixing to >1200m within a few years.
[Response: The deeper you go the less coverage you have and the more difficult it is to calculate changes. 700m reflects the balance between the desire for completeness and the desire to have small error bars – it is not based on a physical theory of ocean mixing or anything. The CFC penetration data also show deep changes of course, though as a percentage of the volume of the deep ocean, it is small. – gavin]
Figure 1 in Cai et al. 2010 seems to suggest that the vertical profile of ocean heat trends in the GISS ER and EH historical model runs is somewhat different than in most other models. Is there any clear reason for this difference in shape?
One thing confuses me about this figure though, which leads me to think I might be misinterpreting it: the comparison with observations seems to suggest there has been significantly more ocean warming below 100m in reality. Yet your own plots suggest pretty good quantitative agreement down to 700m and 2000m, probably even an overestimate compared to observations.
[Response: Not sure what is going there. Possibly it could be an issue with control drift. I did a quick analysis of the 1951-1999 trend in the GISS-ER ensemble mean total OHC and it is 0.15 x 10^22 J/yr. (0.07 to 0.23 x 10^22 J/yr range within the ensemble). It’s possible that Cai et al is only showing a single run? – gavin]
Gavin, in the Feb 2012 post, the graphs shown in the post and linked to as the original are identical. Comparison with other graphs shows that the graph currently displayed in the post is still the original.
[Response: I don’t think you can have updated your cache or something. – gavin]
Gavin, Satellite measurements (eg CERES) of absolute TOA imbalance are poor to useless but month on month and year on year precision is good (ie the deltas).
The jumps in the OHC charts published 1993-2009 (Trenberth 2010) and Lyman composite show OHC gains 2001-03 of about 7E22 Joules (700E20 Joules) which is a rate of about 350E20 Joules/yr.
Trenberth’s 0.9W/sq.m TOA imbalance equates to 145E20 Joules/yr. To get a 350E20 Joules/yr increase in OHC, the TOA imbalance must have leapt from 0.9 to 2.4W/sq.m in the period 2001-03.
Satellite measurement shows no such change in that period.
The period coincides with the XBT to Argo transition, so which is right? – high precision delta measurement by established satellites or XBT-Argo in transition??
I would plug for the precision of the Satellites.
The conclusion is that the step jumps in OHC Charts in this period are an artifact of the transition, and the greater deployment of Argo shows much flatter OHC since ARO 2003. The OHC data before Argo could be considered poor in spatial coverage and unreliable. Drawing a linear trend line through a step jump which is likely wrong is mistaken.
The NODC data on the above chart shows a growth in 0-700m OHC of no more than about 3E22 Joules over a 8-9 year period 2003-2012 which is about 35E20 Joules/yr or equivalent of 0.25W/sq.m. This is a long way short of Trenberth’s 0.9W/sq.m TOA imbalance and Hansen’s 0.6W/sq.m.
[Response: A few things wrong are wrong here. First, Trenberth does not have an independent estimate of the imbalance. His ‘0.9W/m2’ is actually 0.85 W/m2 which comes from the Hansen et al (2005) paper and was that models estimate of the imbalance in 2003. The uncertainty on that was +/-0.15 W/m2 in the ensemble and is bigger once you factor in uncertainties in the forcings themselves and across different models. This is of course the same paper that showed a good match to the 1993-2002 OHC data with an average imbalance of 0.6W/m2 over that period. Second, variations in OHC-700m metric can come from many sources: spatial coverage, ocean internal variability, differences in surface heat flux etc. Assuming that there is a direct one-to-one comparison on annual timescales to TOA imbalance is not valid. Only the long term trends are attributable like that. Finally, by not taking into account even the standard error bars, you are concluding too much. – gavin]
Gavin, I refer you to ‘Tracking the Earth’s Energy – From El Nino to Global Warming’ Trenberth & Fasullo NCAR 15MAY11.
Fig 3 shows the net radiation from TOA from CERES EBAF Ed2.5 from 2000-2010. The value of Rt is the net imbalance at TOA. This is presumably a ‘direct’ measurement which fluctuates at or below the 1W/sq.m for most of the period.
It actually drops well below 1W/sq.m for the 2001-2003 period. In the same 2001-03 period your above NODC chart for the 0-700m is step jumping about 7E22 Joules across this XBT-Argo transition period (a rate of about 2.4W/sq.m). Both cannot be right if 90% of this TOA heat imbalance is being sequestered into global OHC. How do you get a sharp increase in OHC in sync with a decrease in TOA imbalance?
Your comment: “Assuming that there is a direct one-to-one comparison on annual timescales to TOA imbalance is not valid.”
How else is the TOA energy imbalance globally stored in the Earth system if not on the one to one time scale at which it occurs?
[Response: In the real world, energy is conserved (of course). And so TOA imbalances are equal to the integrated heat content change of the whole system. However, neither measure (OHC from NODC or PMEL) or TOA balance from CERES are perfect. OHC is not the total system, and the estimate of it does not include large parts of the ocean. There are changes on land, in ice, in the Arctic, in the deep ocean, in the water storage etc. that on a year to year basis are significant. Over longer time scales, it is easier to assess these changes and conclude that the estimated OHC number is more representative of the total heat content. Our model suggests that OHC-700m is strongly correlated (not perfectly) to the TOA imbalance and is ~90% of the total heat content change. – gavin]
Ken, Gavin answered you inline above
— considering the uncertainties, you can’t assume (as you continue to do) that the published numbers are accurate and precise, and then based on that assume that the exact wiggles you focus on must mean energy is being stored somewhere.
You’re presuming a physical explanation exists for the wiggles.
The short term wiggles are mostly noise and uncertainty.
Just FWIW, RP Sr’s post today is wrong on both counts. The trend in the historical runs (1951-1999) was 0.15 x 10^22 J/yr, not the trend in the control runs. And point 2 is expanded on in comment #18.
Gavin, your inline comment to my #18 above does not contain any specifics. I am quoting a large discrepancy in the NODC 0-700m OHC increase with regard to the 2001-03 period and the satellite numbers which actually show a decrease.
Ice melt (sea and land), land + atmosphere ALL account for about 10% of the heat energy absorbed from the TOA imbalance. You agree that 90% goes into the oceans.
[Response: Only in general. In any one year it could be quite different. – gavin]
By the way your point: ” And so TOA imbalances are equal to the integrated heat content change of the whole system.” should be “And so *integrated* TOA imbalances are equal to the heat content change of the whole system”.
[Response: Not really: F = d/dt ( int{heat content} ) – but it’s always clearer in maths. – gavin]
The evidence of this large OHC discrepancy in the 2001-2003 period when XBT and other methods were transitioned to Argo points to the ‘step jump’ being an artifact of the transition.
The much flatter OHC increase in the last 7-8 years of Argo also points to better measurement giving lower TOA imbalance (ARO 0.25W/sq.m) which is a big difference from 0.9W/sq.m (0.85 rounded) and even Hansen’s 0.6W/sq.m is a significant difference from the 0.9 number.
[Response: The 0.6 and the 0.9 are from the same paper! They are results from different time frames though (1993-2003 vs. 2005). They are not independent estimates. Neither are they necessarily the best estimate for the last 7 years. That would be affected by the solar minimum, possible changes in aerosol forcings from Asia, and of course as with any short time period, is more influenced by internal variability than is the long term trend. It will be interesting to see what models with updated forcings suggest for this period in CMIP5. – gavin]
If the ‘step jump’ was brought back even to the 0.9W/sq.m number over that 2001-03 period, then the OHC increase would have been 2.9E22 Joules not ARO 7E22 Joules and the 0-700 NODC OHC would drop from ARO 12E22 to 8E22 Joules which is a very significant difference.
[Response: If there was a significant reevaluation of the OHC data than that would be a significant change. Obviously. But since there hasn’t been, speculating on what would happen if there was is a bit pointless. There may well be further reevaluations as problems are found and dealt with, but there is no reason to suppose that those adjustments are going to go the way you’d like. I agree that finding a consistency between the data streams is a useful effort, but it isn’t as simple as you suggest. – gavin]
Hi Gavin,
Thanks for the update. Sad to see that Roger Pielke Sr. seems more interested in scoring points than getting the science right. SkepticalScience has a great post on the OHC issue.
It is unfortunate that Roger Pielke Sr. is only to happy to throw stones from afar from a blog that does not permit comments. Roger is also only too happy to highlight other people’s errors and advising various people to update their sites. It is thus ironic that Roger has still not updated his own error made in this June,14 post on his blog, where it still incorrectly states (to this day) that the oceans accumulated accumulated zero Joules of energy between 2003 and May 2011. He has been aware of that error since at least October 1, 2011 when Dana brought it to his attention.
More of Pielke’s errors are highlighted in the first SkepticalScience post linked to above– but I doubt he will fix them given that he has thus far failed to update his 2011 post.
It is bad enough that Pielke Sr. is misleading reader’s of his blog, but that he seems to have no problem with misleading school students is especially egregious.
Earlier I noted that Pielke Senior was more interested in trying to score (PR) points instead of getting the science right. His latest update on his dictator-style blog (i.e., no scientific critique or challenges permitted) beautifully reinforces that point. So thank you Roger! :)
Gavin, thank you for considering my points.
I would have thought that consistency between data streams was essential to building a case. If you agree that the satellites (CERES and others) were showing decreases in TOA imbalance when OHC was sharply increasing, this must make one or the other data stream incorrect.
Which is it? The OHC chart or the satellites?
[Response: Unclear. Other possibilities are that they are both right but the magnitude and variability of the missing terms is underestimated. – gavin]
“Response: Not really: F = d/dt ( int{heat content} ) – but it’s always clearer in maths. – gavin]”
What you have expressed is that the Forcing (TOA imbalance) = differential of heat content change wrt time. Conversely the integral of TOA imbalance wrt time = change in heat content – which is what I said above.
The 0.9W/M2 I refer to is that consistently maintained by Dr Trenberth. The 0.6W/M2 (2005-2011 period) I refer to is that reduced value argued by Jim Hansen & von Schuckmann in “Earth’s Energy Imbalance and Implications” – December 2011 final publication.
[Response: Trenberth’s number is only derived from the earlier Hansen paper. It is not an independent estimate. – gavin]
Trenberth does not believe Hansen’s argument that Asian aerosols are mainly causing this TOA imbalance reduction ‘for a minute’ – see skeptical science discussion on this point.
Trenberth argues that the missing heat is still missing – probably in the deep oceans and that Hansen is wrong because the Asian aerosols are already taken account of and the TOA imbalance is still 0.9.
[Response: Do you have an actual quote for this? I doubt very much that Trenberth has said that aerosol forcing is completely known – and since it isn’t there is an unarguable uncertainty in what the imbalance should be. – gavin]
Hansen accepts that it never arrived in the first place in the 2005-11 period. I would have thought a major difference in opinion between two on the world’s leading climate scientists.
[Response: If you read both Hansen’s paper, he gives his reasoning quite thoroughly. But since Trenberth is only quoting Hansen’s earlier paper for his estimate, I don’t see how this can be counted as a major difference of opinion. Both of them acknowledge the uncertainty in the data (both the forcings and the OHC numbers) and have proposed slightly different resolutions. I’m sure that both will be happy to see it resolved by better data regardless of what that resolution is. – gavin]
Gavin, thanks for the detailed reply.
In respect of Kevin Trenberth’s opinion on Asian aerosols and the imbalance – the following quote from SKS http://www.skepticalscience.com/news.php?p=2&t=213&&n=865 is relevant:
Quote:
Kevin Trenberth at 02:29 AM on 25 July, 2011
As has been noted, I have been traveling, and I have quickly gone through the 67 comments. A few responses follow.
There seems to be some confusion over Fig. 3. This represents the total net radiation from CERES EBAF and so it does include effects of clouds. It is not just the clear sky component (that is a lot more uncertain). Ironically the working definition of “clear sky” used in the community excludes clouds but includes aerosol. Since aerosols affect clouds (the indirect effect), I find this rather unsatisfactory. Still it does not affect things here.
There is discussion in the comments of the supposed finding that increasing aerosol (pollution) from China may be the explanation for the stasis in surface temperatures and I do not believe this for a moment. Similarly, Jim Hansen has discussed the role of aerosol as a source of discrepancy. However, the radiation measurements at the top of the atmosphere from satellites (CERES) include all of the aerosol effects, and so they are not extra. They may well be an important ingredient regionally, and I have no doubt they are, but globally they are not the explanation.” endquote
Dr Trenberth is clearly saying that the imbalance (presumably ARO 0.9W/sq.m) “include all of the aerosol effects, and so they are not extra”.
[Response: Thanks for the link. Note that his ‘not for an instant’ comment is related to the surface air temperatures, not OHC changes or the imbalance. Obviously any radiation measurements include all effects (clouds, aerosols, etc.) and so there is nothing ‘extra’ to be added to those measurements – but the fact is that they are not good enough to determine the imbalance (the offset in the raw CERES SW and LW is around 5 W/m2). The 0.9W/m2 comes from the 2003 average in the Hansen et al (2005) paper – nowhere else. And if the aerosol forcing in those runs was underestimated, and the solar forcing was overestimated (both plausible), or these factors were different from expected after 2003, the imbalance that you would have got would be less. The fact of an imbalance is therefore much more robust than the exact magnitude. – gavin]
Kevin Trenberth at 02:29 AM on 25 July, 2011
As has been noted, I have been traveling, and I have quickly gone through the 67 comments. A few responses follow.
There seems to be some confusion over Fig. 3. This represents the total net radiation from CERES EBAF and so it does include effects of clouds. It is not just the clear sky component (that is a lot more uncertain). Ironically the working definition of “clear sky” used in the community excludes clouds but includes aerosol. Since aerosols affect clouds (the indirect effect), I find this rather unsatisfactory. Still it does not affect things here.
There is discussion in the comments of the supposed finding that increasing aerosol (pollution) from China may be the explanation for the stasis in surface temperatures and I do not believe this for a moment. Similarly, Jim Hansen has discussed the role of aerosol as a source of discrepancy. However, the radiation measurements at the top of the atmosphere from satellites (CERES) include all of the aerosol effects, and so they are not extra. They may well be an important ingredient regionally, and I have no doubt they are, but globally they are not the explanation.
How did the imbalance occur (comment 2) can be seen from Fig 3 broken into ASR and OLR (not shown here). ASR increased, suggesting fewer clouds as occurs in La Nina over the tropical Pacific, but OLR decreased. The latter seems to be mainly a temperature signal: colder conditions mean less radiation to space. This is often a complex relationship because the biggest variations occur in the Tropics and there is typically a large offset in OLR and ASR signals in association with variations in convection that largely relate to albedo effects being offset by the radiation to space from tops of clouds. Thus fewer clouds means more ASR and more OLR (since the radiation to space comes from warmer lower levels). But that works only in the tropics. At higher latitudes OLR is dominated by surface temperature effects.
Comment 17 asks about “back radiation” which is really “downwelling radiation” that is the downward component emitted from the atmosphere in all directions. Clouds, water vapor, and all the greenhouse gases play a key role and the emissions correspond to the temperature of the air. For clouds, the key temperatures are the cloud top temperature for emissions to space and the cloud bottom temperature for emissions back toward the surface. An important point is that to understand the energy flows (which include radiation), the full three (or really four) dimensional structure of the atmosphere is needed, and the simple Figure 1 does not show the vertical structure of temperature.
In response to 26: yes melting permafrost can take up some energy but the amount turns out to be very tiny.
The last topic I’ll touch on is the ocean heat content (OHC). A couple of references were made to the von Schuckmann and Traon paper, which was nice to see, but has some flaws. For instance the data down to 2000 m in the ocean have increased since 2002 and since the beginning of that analysis, yet their error bars are constant. New analyses will be of considerable interest and are underway.
I discussed this in this article here in Nature:
Trenberth, K. E., 2010: The ocean is warming, isn’t it? Nature, 465, 304. http://www.cgd.ucar.edu/cas/Trenberth/trenberth.papers/NatureNV10.pdf
ENSO involves a redistribution of OHC and losses to the atmosphere in the latter part of El Nino, and gains during La Nina, so this is internal to the climate system, not external (comment 31). The southern ocean is clearly playing a role (comments 48, 49)in taking up heat and mixing it deep, even though the magnitude of the observed warming is small. But the data are fragmentary and unsatisfactory in many respects. Nonetheless, the southern oceans, while playing some role, are not the main place where the heat goes in our model. We have a paper submitted that describes and documents that in more detail so it is premature to go into detail here.
A nice paper is in press in GRL by Palmer et al (UKMO) using two Hadley climate models that details the relationship between SST and OHC to different levels in their model. Going all the way to the bottom accounts for all of the OHC but the upper OHC in the top 300 m and the SST (which relates to that) are not always good indicators of total OHC. So they also find that energy can go missing into the deeper ocean, and moreover the main phenomenon in their model associated with this is La Nina. [This latter point is not in their article].
The bottom line is that the ocean plays a major role in climate and especially in interannual and decadal variability, and a lot more will be written on this topic. Maintaining an adequate observing system is extremely important.
Kevin Trenberth
Gavin, I noticed in the Hansen 2005, fig 2, that your paper G A Schmidt 2005 in preparation, is referenced for the model. Do you have a link to this paper? TIA
[Response: Schmidt et al, 2006, J. Clim. – gavin]
Gavin, your comment at #25 thus:
“The 0.9W/m2 comes from the 2003 average in the Hansen et al (2005) paper – nowhere else.”
Please refer to Figs 2 and 4 of Kevin Trenberth’s paper 3AUG09 “An Imperative for Climate Change Planning: Tracking Earth’s Global Energy” here:
http://www.cgd.ucar.edu/cas/Trenberth/trenberth.papers/EnergyDiagnostics09final2.pdf
Trenberth seems to have the number of 0.9W/sq.m with error bars fom Refs 6&7.
[Response: Ref 6 is the Trenberth BAMS paper, where the statement is: “The TOA energy imbalance can probably be most accurately determined from climate models and is estimated to be 0.85 ± 0.15 W m−2 by Hansen et al. (2005) and is supported by estimated recent changes in ocean heat content (Willis et al. 2004; Hansen et al. 2005).”. Which is, as we have discussed the 2003 value in the GISS-ER ensemble. Frankly, I’m a little surprised that I haven’t seen an assessment of the current imbalance across the all the CMIP3 models…. did I miss it? – gavin]
Note from Fig 4 he reconciles the AR4 Radiative Forcings of +1.6W/sq.m to +0.9W/sq.m with a Radiative cooling term of -2.8W/sq.m and a WV + IA feedback of +2.1W/sq.m . Was this all taken from the Hansen 2005 paper?
[Response: The total forcings comes from AR4, but has a big uncertainty (+/- 1 W/m2 perhaps) because of the aerosol term. The LW term comes from 0.75ºC increase in T and stefan-boltzmann at the emitting temperature (255K). The feedback term doesn’t normally get thought of like this, but it could be written like that (i.e. lambda*T-planck) but is just a residual – and so collects all of the uncertainties. Which is why the recent period is not much of a constraint on sensitivity.- gavin]
Another thing on errors and corrections. In 2010, the AMSU-A near-surface temperature data was looking like a real outlier, making 2010 a much warmer year than any other on the record (GISTEMP has it as warmest but not significantly; HadCRUT as a little below the warmest).
I had another look at the site recently and it now reports that channel 4 (near surface) went bad in 2008. OK good, an error was corrected. When I dug a bit deeper, it seems that channel 4 was known to be bad (if my HTML anchor doesn’t work, scroll down to the heading “Yet Another Update”) since March 2008 – i.e., this wasn’t an error only picked up 2 years later by checking the data.
Everyone can make mistakes, as we see in Gavin’s article. But what I wonder why this other blunder wasn’t all over the blogosphere. Or are some people allowed to make mistakes without causing offence to the sort of people (Roy Spencer was one of the authors of the AMSU-A temperature site) who demand absolute perfection in other circumstances?
Gavin, I have a number of questions from your responses to my #25 and #28.
Dr Trenberth in ‘Tracking the Earth’s Energy – From El Nino to Global Warming’ Trenberth & Fasullo NCAR 15MAY11 states that; “The TOA imbalance can probably be most accurately determined from climate models and Fasullo and Trenberth(2008a)deduced the imbalance to be 0.9W/sq.m where the error bars are +/-0.5W/sq.m.”
[Response: … which is referenced to Hansen et al (2005), with a slightly expanded error bar. – gavin]
Dr Trenberth in his part 3 discussion of modelling suggests that ‘the implication is that the missing heat is being deposited mainly in the region below 700m depth’.
I disagree with your interpretation of Trenberth’s comment on the SKS website quoted in #25 applies only to surface temperatures. quote:
“Similarly, Jim Hansen has discussed the role of aerosol as a source of discrepancy. However, the radiation measurements at the top of the atmosphere from satellites (CERES) include all of the aerosol effects, and so they are not extra. They may well be an important ingredient regionally, and I have no doubt they are, but globally they are not the explanation” endquote
Dr Trenberth is clearly referring to Hansen’s point that the ‘imbalance’ has reduced due to Chinese aerosols (presumably reflecting heat out to space): Dr Trenberth suggests that this might happen regionally but not globally presumably because the CERES measurements are not showing a reduction in the 2005-2011 period. Have a look at Rt for the period in Fig 3 of same paper.
Hansen plugs for a reduced imbalance from lower OHC measurement and Trenberth is for unreduced imbalance with the discrepancy being missing ocean heat hidden below 700m. That is a significant difference.
I have a problem with the whole issue of how the CERES raw readings are corrected. I have seen +6.4W/sq.m as the raw imbalance and you have quoted +5W/sq.m which is corrected down to ARO 0.9W/sq.m. This correction is huge, and it is done to match the models ie. say +5/sq.m has a -4.1W/sq.m correction applied down to 0.9W/sq.m?
[Response: The fact is that the satellite measurements don’t have a good enough absolute calibration to measure the imbalance directly (the ‘raw’ 5 W/m2 is clearly not real). So in building a budget where you are using multiple sources of data (satellite, reanalyses, OHC etc.), the model based imbalance was used ‘faute de mieux’. It doesn’t have much impact on the rest of the budget, but it needs to be borne in mind that the current ‘best estimate’ of the budget is not set in stone and may vary in the light of improved data. – gavin]
You say “The fact of an imbalance is therefore much more robust than the exact magnitude”.
[Response:Yes. This is evidenced by the undisputed increase in OHC. – gavin]
With such a large correction (80-90% of the raw number) – how sure are we that the imbalance is even positive from CERES? If the imbalance is closer to 0.25W/sq.m in the post Argo period then the correction would be 95% of the raw number.
[Response: You are not sure of anything just based on CERES. – gavin]
To suggest that CERES is evidence of an imbalance of any baseline magnitude at all is a circular argument. This is what Dr Trenberth said in the 2009 paper:
“The Clouds and the Earth’s Radiant Energy System (CERES)
measurements from March 2000 to 2005 were used at top of
atmosphere(TOA) but adjusted to an estimated imbalance
from the enhanced greenhouse effect of 0.9 +/- 0.5 W m-2
(with 90% confidence limits) [7*].”
CERES is probably good for monitoring changes in the TOA radiation not any absolute value.
[Response: Agreed. I have never suggested that CERES is a source of certainty in the TOA imbalance. – gavin]
I notice that SORCE has a -4.5W/sq.m TSI discrepancy (it measures 1361.5W/sq.m when other satellites measure 1366) since deployment in 2005, and last time I looked at the website there was no explanation for this difference. It would make a -1.1W/sq.m reduction in incoming solar radiation if it were right, which is a story similar to CERES – an arbitrary absolute correction is being applied which could be right or wrong, so only difference measurements are useful.
[Response: There was discussion of the TSI issue at AGU last year. – gavin]
Thanks Gavin for your responses. Your referenced December 2011 AGU meeting had some very interesting information on the TSI measurement which I had not seen viz:
“Combined with a full implementation of an annually varying temperature correction, their latest processed data product has reduced the discrepancy with the TIM instrument from over 5 W/m2 to less than 0.5 W/m2 – a huge improvement. The new PREMOS instrument onboard Picard, a french satellite, was also tested before launch last year, and they improved their calibration as well – and the data that they reported was also very close to the SORCE/TIM data: around 1361 W/m2 at solar minimum.”
Does this mean that the SORCE number of 1361 W/sq.m at solar minimum should be adopted rather than the mean value 1366 W/sq.m so far widely used?
[Response: Yes. This will take a little time to work its way through to published results (since they will be relying on CMIP5 runs done last year and this), but I would anticipate that almost all post-CMIP5 model simulations will use 1361 instead of 1365. – gavin]
If so that would give a new mean value of 1361.5 W/sq.m (about 1 W/sq.m max to min) and an Incoming Solar Radiation number of 340.4 W/sq.m.
Dr Trenberth’s 15MAY11 paper quoted at #30 shows the Fig 1 “Global Energy Flows” diagram with an incoming Solar Radiation of 341.3 W/sq.m with Refected Solar Radiation at 101.9 and LW outgoing at 238.5. The sum of 341.3 – (101.9 + 238.5) = 0.9 W/sq.m – our oft quoted imbalance number.
If the Incoming Solar number was reduced to 340.4 from 341.3 (coincidentally a drop of 0.9 W/sq.m) then the imbalance would be zero if the Reflected term and the outgoing LW values remained the same.
[Response:No, because you are not following the reasoning used. Instead, it is more likely that the albedo estimate will change. This is the residual number, not the imbalance.]
If you assumed that the reflected term stayed at the same fraction of the incoming (29.85%) then it would reduce to 101.6 W/sq.m and the sum would be :
340.4 – (101.6 + 238.5) = 0.3 W/sq.m
This is remarkably close to my number of 0.25 W/sq.m at #17 from the last 7-8 years of OHC data.
Any thoughts on this?
Gavin,
I have already suggested an alternative explanation by reducing the reflected (albedo) term using the same fraction as the non-SORCE estimates viz:
“If you assumed that the reflected term stayed at the same fraction of the incoming (29.85%) then it would reduce to 101.6 W/sq.m and the sum would be :
340.4 – (101.6 + 238.5) = 0.3 W/sq.m”
Again trying to construct a positive imbalance of 0.9 or 0.6 or 0.25 from differences in numbers of magnitude 340 needs an accuracy of better than 1 in 340 which is quite unrealistic in the context of the wide error bars of some of the forcing components – particularly aerosol effects and feedbacks.
[Response: Uncertainties in aerosols/feedbacks only affect estimates from climate models. Measurements of TOA radiation would simply be the ‘answer’ (if they were good enough). – gavin]
Very accurate ocean heat content has to be the key to quantifying the imbalance.
[Response: Probably. Not sure I’ve indicated anything different… – gavin]
Gavi,
I note that you have not commented on my calculation which keeps the reflected term at the same fraction (29.85%) of the incoming solar radiation.
It seems logical to suggest that if the current Energy flow chart is budgeted around an imbalance number of +0.9W/sq.m then the ratios of incoming to reflected to achieve that budget would hold same if the absolute value of the incoming was refined to a more accutate figure (TSI of 1361.5) – ISR of 340.4W/sq.m.
Maybe it is time to ask Kevin Trenberth what he will make of the chart with a TSI of 1361.5 instead of 1366.