Is ocean warming accelerating faster than thought?

by Nic Lewis

*** UPDATE : response to comments by Zeke Hausfather appended

There are a number of statements in Cheng et al. (2019) ‘How fast are the oceans warming’, (‘the paper’) that appear to be mistaken and/or potentially misleading. My analysis of these issues is followed by a reply from the paper’s authors.

Contrary to what the paper indicates:

  • Contemporary estimates of the trend in 0–2000 m depth ocean heat content over 1971–2010 are closely in line with that assessed in the IPCC AR5 report five years ago
  • Contemporary estimates of the trend in 0–2000 m depth ocean heat content over 2005–2017 are significantly (> 95% probability) smaller than the mean CMIP5 model simulation trend.

Ocean warming over 1971–2010 per IPCC AR5 and contemporary estimates

1. The paper states: “The warming is larger over the 1971–2010 period than reported in AR5. The OHC trend for the upper 2000 m in AR5 ranged from 0.20 to 0.32 Wm−2 during this period (4: AR5). The three more contemporary estimates that cover the same time period suggest a warming rate of 0.36 ± 0.05 (6: Ishii ), 0.37 ± 0.04 (10: Domingues), and 0.39 ± 0.09 (2: Cheng) Wm−2.” [Numbered references in this article are to the same numbered references in the paper. The number is followed by the lead author’s name, or AR5, where this aids clarity.]

2. AR5 (4) featured 0–700 m depth ocean heat content (OHC) 1971-2010 linear trend estimates from five studies, ranging from 0.15 to 0.27 Wm−2  of the Earth’s surface. Adding the AR5 700–2000 m OHC 1971-2010 trend estimate of 0.09 Wm−2  brings the range up to 0.24 to 0.36 Wm−2 , not to 0.20 to 0.32 Wm−2 as stated. The warming rates plotted in Supplementary Figure S1 agree to my values, not to those stated in the paper.

3. Importantly, although AR5 featured several OHC trend estimates for 0–700 m depth, its assessment of the Earth’s energy uptake (Section 3.2.3 and Box 3.1) used only the highest one (10: Domingues), adding the Levitus (12) 700–2000 m OHC trend to give a best estimate 0–2000 m warming rate over 1971–2010 of 0.36 Wm−2. That rate is identical to one (6: Ishii) of the three more contemporary estimates given in the paper and extremely close to the other two of them – within the innermost one-third of their uncertainty ranges.

See Figure 1, left hand section, and compare with the ‘Updated OHC estimates compared with AR5’ figure [Fig 2] in the paper. It is therefore misleading to claim that the warming is larger over the 1971–2010 period than reported in AR5.

4. Moreover, over the final decade covered by AR5, 2002–2011, the trend of the 0–2000 m OHC time series that AR5 adopted for its assessment, 0.60 Wm−2, was noticeably higher than those for two of the three more contemporary estimated OHC datasets given in the paper (0.35 (6: Ishii) and 0.52 (2: Cheng) Wm−2) and, unsurprisingly, almost identical to the third (10: Domingues + 12: Levitus).

Figure 1: Updated 0–2000 m OHC linear trend estimates compared with AR5 and the CMIP5 mean. Error bars are 90% confidence intervals; black lines are means. Units relate to the Earth’s entire surface area.

Ocean warming over 2005–2017 per CMIP5 models and contemporary estimates

5. The paper’s ‘Past and future ocean heat content changes’ figure [Fig 1] caption states: “Annual observational OHC changes are consistent with each other and consistent with the ensemble means of the CMIP5 models for historical simulations pre-2005 and projections from 2005–2017, giving confidence in future projections to 2100 (RCP2.6 and RCP8.5).” This does not appear to be true for the linear trends of the annual values for the 2005–2017 projections, at least.

The main text states: “Over this period (2005–2017) for the top 2000 m, the linear warming rate for the ensemble mean of the CMIP5 models is 0.68 ± 0.02 Wm−2, whereas observations give rates of 0.54 ± 0.02 (2), 0.64 ± 0.02 (10), and 0.68 ± 0.60 (11) Wm−2.”

6. Five problems with this claim regarding 2005–2017 warming rates are:

(i)    the CMIP5 RCP2.6 and RCP8.5 projections top 2000 m OHC data archived for the paper shows an ensemble-mean linear warming rate over 2005–2017 of 0.70 ± 0.03 Wm−2, not 0.68 ± 0.02 Wm−2. The same is true when also including data from the third scenario used in the paper (RCP4.5).

(ii)  the underlying time series from which the third observational estimate is derived (Fig. 3.b in 11: Resplandy) spans 1991–2016, and has a lower (and highly uncertain) linear trend from 2005 to 2016 (its final year) than the stated 0.68 Wm−2­ (which is calculated over 1991–2016), so this estimate should be excluded;

(iii)  the statement inexplicably omits the Ishii et al. (6) observational data, which also have a lower estimated trend (0.62 ± 0.07 Wm−2) than per CMIP5 over this period; and

(iv)  the uncertainty range for the Cheng (2) estimate appears to be seriously understated: I calculate that the estimate should be 0.54 ± 0.06 (rounding 0.055 up), not 0.54 ± 0.02.

(v)  adding the uncertainty ranges in quadrature, since CMIP5 and observational errors are independent, the CMIP5 ensemble mean trend is statistically inconsistent with the all three of these observational trend estimates (2: Cheng, 6: Ishii, 10: Domingues);

The right hand section of Figure 1 shows a corrected comparison of  CMIP5 mean and observational 0–2000 m depth ocean warming rates over 2005–2017.

7. Although it is pointed out in the paper’s Supplementary material that volcanic eruptions after 2000 have not been taken into account in CMIP5 models (with a minor effect on projected warming since then), it has been shown that when underestimation of other growth in other drivers of climate change is accounted for there is no overall bias in post-2000 CMIP5 model forcing growth (Outten et al. 2015).

Other issues

8. The straight black line in the ‘Past and future ocean heat content changes’ figure [Fig 1] for the Resplandy et al. (11) OHC estimate gives a misleading impression of close agreement with the three OHC time series based on in situ observations over 1991–2016: its trend uncertainty range is so large (0.08 to 1.28 Wm−2) that the apparent close agreement is most likely due to chance.

9. The Press release for the paper claimed that ‘If no actions are taken (“business as usual”), the upper ocean above 2000 meters will warm by 2020 ZetaJoules by 2081-2100″, which is based on CMIP5 model RCP8.5 scenario simulations. That is misleading. RCP8.5 involves not only no actions (including those already carried out) being taken, but also emissions being unusually high for a business as usual scenario.[ii]

Nicholas Lewis                                                                                          January 2019

[i]  The paper does not directly claim that ocean warming is accelerating faster than thought; that is the headline of The New York Times article about the paper.

[ii]  As the source paper (Riahi, K., et al., 2011: RCP 8.5–A scenario of comparatively high greenhouse gas emissions. DOI 10.1007/s10584-011-0149-y) states: “RCP8.5 combines assumptions about high population and relatively slow income growth with modest rates of technological change and energy intensity improvements, leading in the long term to high energy demand and GHG emissions in absence of climate change policies.”, and that “[RCP] 8.5 corresponds to a high greenhouse gas emissions pathway compared to the scenario literature”. As Riahi et al. (2011) make clear, the assumed energy intensity improvement rates are only about half the historical average while middling world GDP growth is assumed, leading to coal use increasing almost 10 fold by 2100.

2. L. Cheng et al., Sci. Adv. 3, e1601545 (2017).
4. M. Rhein et al., in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, T. F. Stocker et al., Eds. (Cambridge Univ. Press, 2013), pp. 215–315.
6. M. Ishii et al., Sci. Online Lett. Atmos. 13, 163 (2017).
10. C. M. Domingues et al., Nature 453, 1090 (2008).
11. L. Resplandy et al., Nature 563, 105 (2018).
12. S. Levitus et al., Geophys. Res. Lett. 39, L10603 (2012).
Additional references
Cheng, L., Abraham, J., Hausfather, Z. and Trenberth, K.E. (2019). How fast are the oceans warming?. Science, 363(6423), pp.128-129.
Outten, S., P. Thorne, I. Bethke, and Ø. Seland (2015), Investigating the recent apparent hiatus in surface temperature increases: 1. Construction of two 30-member Earth System Model ensembles, J. Geophys. Res. Atmos., 120, doi:10.1002/2015JD023859

Addendum: Update 22 January 2019 in reaction to tweets by co-author Zeke Hausfather

The authors seem still not to understand that their Figure 2 AR5 0–2000 m warming rates are mathematically wrong, while those I calculate are correct, not merely a different approach. Perhaps when they see it graphically they will admit their error. The below plot shows how the AR5 Box 3.1 1971-2010 Deep ocean (sub-700 m) OHC values (green line) were made up from Levitus 700–2000 m OHC data (black line) plus warming of the sub-2000 m ocean at a rate of zero until 1991 and then 35 TW from 1992 on (blue line).

Cheng et al. deducted from the AR5 sub-700 m OHC (green line) an assumed sub-2000 m warming at a rate of 35 TW throughout 1971–2010 (orange line) to reach their estimate of AR5 700-2000 m warming (magenta line). As can be seen, their ‘AR5’ 700-2000 m OHC estimate (magenta line) has a substantially lower linear trend (warming rate) than the actual AR5 700-2000 m OHC estimate (black line), which is what my calculation uses.

It follows that Cheng et al.’s Figure 2 values for AR5 0-2000 m warming rates over 1971-2010, which add the trend of the magenta line to the 0-700 m warming rate, are understated by the difference between the trends of the black and the magenta lines.

Honest scientists, unlike activists, are prepared to admit and correct factual mistakes in their papers, whether or not they alter its primary conclusions. I expect that Cheng et al. will accordingly submit a correction to Science to substitute correctly calculated 1971–2010 upper 2000 m AR5 OHC trend values for their erroneous values.

Lijing Cheng has asked me to post also the following reply from the paper’s authors to my critique. I am pleased to do so and I thank him for providing it. I have replaced interspersed text extracted from my article with paragraph number references.  The authors’ responses are shown in blue. I have appended my comments, shown in red, on a number of them. 

Paragraphs 1 and 2

Some questions have been raised concerning the numbers in our article (…) and indeed there is an inconsistency between a value in the supplementary material and the main text.  It relates to the use of linear trends and how to assign a change over various periods.  For longer time intervals, a linear trend is not a good fit to the data and use of that to assign a change can be misleading.  In the IPCC AR5, below 700m depth, it is stated that “the heating below 700 m is 62 TW for 1971-2010”.  They also state “For the ocean from 2000 m to bottom, a uniform rate of energy gain of 35 [6 to 61] TW from warming rates centred on 1992–2005 (Purkey and Johnson, 2010) is applied from 1992 to 2011, with no warming below 2000 m assumed prior to 1992.”  Hence the difference for the 700 to 2000 m layer is 62 -35 = 27 TW.  This is 0.05 W m-2 and is what was used in the main text to produce the numbers quoted.   However, if instead one takes the 2 flat lines below 2000 m and subtracts from the actual values, and then fits a linear trend, the implied change is closer to 45 TW which gives the 0.09 W m-2 plotted in Fig. S1.  If the latter is used instead, then the change from the old AR5 values to the newer OHC values is somewhat reduced (see figure below).  The increase is up to 40% over the prior IPCC estimates, and the average is 24%.  This exercise was prompted by a comment by Nic Lewis who we thank, and it highlights the uncertainty in actual trends and their use to depict changes.  The conclusions in our Perspective remain sound. If the alternative analysis method proposed by Nic Lewis is used, the change is not quite as dramatic as implied in some of the associated press releases.

Based on this:

  • While there is an inconsistency that is not discussed between Fig. 2 and Fig S1, it reflects the uncertainties in previous OHC estimates and the associated methods. In particular, some values before 1980 or so are erratic (high values in the 1960s) and a linear trend is not a good fit to the time series.
  • All of our key points are still valid: (1) the best estimates are collectively higher than the 5 estimates featured in AR5 (0-63% higher). (2). And the best estimates are more consistent with each other (0.36/0/37/0.39 Wm−2 than 0.24~0.36 Wm−2 in AR5). (3). Model ensemble means are higher than 5 estimates featured in AR5 (0.39 Wm−2, 8-63% higher) and consistent with new/updated observations.
  • AR5-Box 3.1 used the strongest estimate without backup literature, we state this in the supplement (also read our replies below). So our study justifies the choice in Box 3.1 as we discussed in supplement. The new estimates could be 0-8% stronger than the selected estimate by AR5-Box 3.1 for 1971-2010. But we didn’t make claims regarding Box 3.1 in Science article, so this is not an issue.
  • This would be an adjusted Fig. 2 (plot below) if we used the different value, the key messages do not change:

Additionally, the Domingues value for 0-700m should have extremely large error bars: all of the values prior to 1970 are much higher than from 1970 to 1980 in AR5, (see AR5 Figure 3.2; given also below) and hence the trends for that estimate are extremely dependent on the period used.  Whether that value was used or not in AR5 (and we state it was), the AR5 message is that they really didn’t know the value at all well, and now we do.

Nic Lewis comments:

a) Their arguments justifying their deduction of 35 TW from the AR5 1971–2010 linear trend below- 700 m ocean heating rate to give their 700–2000 m layer heating rate of 27 TW (0.053 Wm−2) make no sense. AR5 arrived at its sub-700 m deep OHC time series (plotted in Box 3.1 Figure 1) by adding, from 1992 onwards, { (year – 1991) * 1.10 ZJ } to its estimate of 700–2000 m depth OHC [1.10 ZJ / year = 35 TW].

The only correct way to derive the AR5 700–2000 m depth OHC time series is to take its sub-700 m OHC time series and reverse out this addition, which is what I did. Cheng et al.’s method gives the wrong 1971–2010 rate of 700–2000 m depth ocean heating irrespective of whether this is measured by a linear trend or otherwise.

b) Their explanation of the inconsistency between their Fig 2 and their Fig S1 conflicts with the facts. They imply in the supplementary material that for both figures the warming rates are linear trends from an ordinary least squares (OLS) fit. Whether or not an OLS fit is ideal is irrelevant; it is what AR5 did and is what Cheng et al. indicated they did. I have verified that their Fig S1 estimates agree to OLS fits to their data. It is undeniable that the AR5 warming rates plotted in their Fig 2 are erroneous.

c) Numerical simulations using strongly autocorrelated random errors confirm that the 1971–2010 trend uncertainty for the Domingues 0–700 m OHC trend stated in AR5, which is incorporated in the AR5 0–2000 m trend uncertainty plotted in my Fig 1, appears to adequately reflect the large uncertainty that AR5 showed the Domingues estimates as having in pre-Argo years (which dominates the uncertainty shown in Box 3.1 Fig 1).

 Paragraph 3

Please read our supplement, we fully describe the whole story as follows :

“IPCC-AR5 (1) featured five estimates for OHC within 0-700m including Levitus et al. (2) (LEV), Ishii et al. (3) (ISH), Domingues et al. (4) (DOM), Palmer et al. (5) (PAL), Smith and Murphy (6) (SMT), one estimate for 700-2000m: Levitus et al. (2) (LEV) and one estimate below 2000 m: Purkey and Johnson (7) (PG). For the Earth’s energy budget inventory (Box 3.1 in Ref. (1)) and other places, DOM, LEV and PG are used for 0-700m, 700-2000m, and below 2000m respectively. Among the five 0-700m OHC estimates in AR5, the minimum yields an ocean warming of 74 [43 to 105] TW (SMT) within 1971-2010, which is almost half of the maximum, with a rate of OHC change of 137 [120 to 154] TW (DOM). If all of five estimates are treated equally, a huge error bar has to be put in the final OHC estimate, downplaying the reliability of OHC records.

AR5 chose the DOM estimate to assess Earth’s energy budget, rather than any others or an ensemble mean of the five featured estimates by stating:

‘Generally the smaller trends are for estimates that assume zero anomalies in areas of sparse data, as expected for that choice, which will tend to reduce trends and variability. Hence the assessment of the Earth’s energy uptake (Box 3.1) employs a global UOHC estimate (Domingues et al., 2008) chosen because it fills in sparsely sampled areas and estimates uncertainties using a statistical analysis of ocean variability patterns.’

In this way, the ‘conservative error’ of many estimates has been identified in AR5 but not supported by the literature. Since AR5, many studies have been looked into this issue either directly or indirectly (8-13) and several new/revised estimates are available, and are chosen by our study.

For OHC within 0-700m, the new CHG and ISH estimates are consistent with DOM (Figure S1). The three estimates are collectively higher than LEV/ISH/PAL/SMT featured in AR5 (Figure S1). Therefore, the progress after AR5 justifies the choice of DOM in AR5 for OHC 0-700m.”

We note that the AR5 featured five different OHC estimates available at the time in the main body of their assessment and the main figure (Fig. 3.2), shown below. We feel that this justifies comparing newer OHC estimates to all five series, rather than just the Domingues series that the AR5 chose to highlight.
Additionally, when the 700-2000m values from Levitus are used (as discussed above), recent records still show 0% to 8% more warming over the 1971-2010 period than the AR5 Domingues value: 0.36 ± 0.05 (Ishii), 0.37 ± 0.04 (Domingues+Levitus), and 0.39 ± 0.09 (Cheng) compared to the old Domingues value of 0.36 Wm−2.

Incidentally, since we have this figure here: note the big bump in Domingues in the top panel in the 1950s and 60s.  Also note the bump in the 1970s in the 700 to 2000 m layer.

Nic Lewis comments:

None of these points affects what I say in my article. The paper says ” These recent observation-based OHC estimates show highly consistent changes since the late 1950s (see the figure). The warming is larger over the 1971–2010 period than reported in AR5. The OHC trend for the upper 2000 m in AR5 ranged from 0.20 to 0.32 W m−2 during this period (4).” Since the figure referred to shows only 0–2000 m OHC, it is implicit that ” The warming is larger over the 1971–2010 period” in the next sentence refers to warming in the 0–2000 m ocean layer. AR5 only featured 0–700 m OHC dataset other than Domingues when discussing warming of that ocean layer; it did not use any of them to estimate warming over 0–2000 m .

Paragraph 4

The period from 2002-2011 seems somewhat arbitrary, and we chose to focus on the 1971-2010 period as it was the one specifically highlighted in the AR5. We would expect greater agreement between older and newer estimates of OHC changes after around 2005 (when Argo data begins being available), as corrections of XBT measurements and better spatial interpolation approaches – which were the primary changes made to newer OHC datasets – matter much more prior to the early 2000s. And there is a better agreement after 2005, Johnson et al. 2018 BAMS state of climate show this already. We do give updated values for 2005-2017 (Argo period) for comparison with CMIP5.

Further, we could see the time series plot similar to AR5-Fig.3.2 below, the new time series apparently show better consistency than AR5-Fig. 3.2 among estimates.Figure.  Times series of OHC 0-2000m for the four best estimates compared with CMIP5 model ensemble mean and two-sigma model spread.

Nic Lewis comments:

My paragraph 4 is simply an observation; it does not claim to point to any mistake in the paper. Nor does it bear on my point that it is misleading to claim that the warming is larger over the 1971–2010 period than reported in AR5. 

Paragraph 5

First, 2005-2017 is a short period, there are many uncertainties: 1) There are short-term variability in the time series (i.e. Interannual variability such as ENSO) and uncertainty in observations, these can impact the trend calculation in a short period within 2005-2017;  2) CMIP5 models do not contain natural variability in phase with actual natural variability, and 3) do not contain realistic forcings after 2005. We discuss this in some detail in supplement:

“We show in the main text that over the period of 2005-2017, the linear warming rate for the ensemble mean of the CMIP5 models is 0.68±0.02 W m-2, slightly larger than the observations (ranging from 0.54±0.02 to 0.64±0.02). Many studies, including Gleckler et al. (13) and Schmidt et al. (16) have shown that the volcanic eruptions after 2000 have not been taken into account in CMIP5 models. Taking this into account, the Multi-Model-Average of CMIP5 simulations will be more consistent with observations during the recent decade (13).”

Gleckler et al. 2016 explicitly addressed the volcano impacts in ocean heat content comparison between model and observations, after Outton et al. 2015, they suggested a correction for a global volcanic aerosol forcing since 2000 of 0.19±0.09 Wm−2.

Nic Lewis comments:

None of this is relevant to my point that the claim in the caption to Fig 1 of the paper that “Annual observational OHC changes are consistent with each other and consistent with the ensemble means of the CMIP5 models for historical simulations pre-2005 and projections from 2005–2017” is contradicted by the differences in the linear trends of the data involved over 2005–2017, having regard to the trend uncertainty ranges.

 Paragraph 6

(i)   As we point out in the supplementary materials (figure caption) “CMIP5 results (historical runs from 1971 to 2005 and RCP4.5 from 2006 to 2010) are indicated by the green bar”. Using CMIP5 historical + RCP4.5 runs gives us 0.68 ± 0.02 Wm−2. We could have been clearer in the main paper which RCP runs were shown in the trends comparison part of the figure; we did in earlier drafts of the article but it was cut at the suggestion of the editors at Science to shorten/simplify the figure caption.

(ii)  Resplandy explicitly state in their paper that the trustable estimate is the linear trend, rather than annual values, because the O2 and CO2 changes on annual scales are not primarily driven by temperature. Hence we only use their linear trend (the revised version shown in Real Climate).

(iii)  Earlier drafts of the paper did include the Ishii estimate, though it was omitted from the final version due to length constraints as it fell between the 0.54 (Cheng) and 0.64 (Dom/Lev) instrumental estimates noted. We should have made this clearer (e.g. mention that instrumental estimates range from 0.54 to 0.64), although its exclusion here does not impact any of our conclusions. As the 0.62 Ishii estimate is closer to the Dom/Lev than Cheng, its inclusion would make the overall range of instrumental estimates seem closer to CMIP5 over this period.

(iv)  We used the error calculation presented in Foster and Rahmstorf 2011, which takes accounts of the autocorrelations in a time series.



  • OHC_se is standard error using OLS.
  • v=1+2*p1/(1-q)
  • p1=OHC_autocorrelation(2)
  • q= OHC_autocorrelation(3) / OHC_autocorrelation(2)
  • OHC_autocorrelation is the autocorrelation of the time series.

Using this method, we can replicate the error bar provided by AR5, so it should be nearly identical to AR5 method.

We also get 0.06 uncertainty range for Cheng (2) if simply using OLS method. But this does not impact the comparison between new/updated observations and model.

(v)  The four new/updated best estimates are 0.54, 0.62, 0.64, and 0.68 Wm-2. The CMIP5 historical + RCP4.5 model ensemble mean is 0.68 Wm-2. If we focus on instrumental estimates (and exclude Resplandy et al given its large uncertainties), the CMIP5 models are a bit higher than observations during the Argo era, although, as we discuss in the Supplementary Materials and our previous replies, mismatches between projected and observed forcings in the forecast period are expected to give differences over this period.

Nic Lewis comments:

 (i)    No indication is given in the paper or the supplementary material that “the linear warming rate for the ensemble mean of the CMIP5 models” for the top 2000 m over 2005–2017 referred only to projections based on the RCP4.5 scenario. Although the authors were unlucky to have an editor who was more concerned with presentation than scientific content, they, not the editor, are ultimately responsible for the paper.
The caption to their Fig 1, which states that annual observed OHC changes are consistent with the ensemble means of the CMIP5 models, shows projections based on the RCP2.6 and RCP8.5 scenarios. The relevant 2005–2017  trends for those scenarios are respectively 0.70 and 0.71 Wm-2.

(ii)  This supports my point that the 2005–2017 trend estimatable from the (revised) Resplandy data is highly uncertain. The fact that the estimated 1991–2016 Resplandy trend is somewhat less uncertain (at 0.68 ± 0.60 Wm-2) does not justify treating it as also being the 2005–2017 trend. The fact is that the information available from the Resplandy data is so imprecise that it adds almost nothing to knowledge about ocean warming trends over 2005–2017.

(iii) Noted. IMO this issue illustrates a problem with publishing papers in Science and similar ‘high profile’ journals.

(iv) I also used the error calculation presented in Foster and Rahmstorf 2011. I estimated the relevant autocorrelations over 2005–2017, since that was the period over which the trend was being estimated. They were insignificantly negative for the Cheng data, so no correction to the OLS standard error  estimate of 0.0332 was appropriate. Multiplying this by 1.65 gives a 5–95% uncertainty range of, rounded up, ±0.06. The authors appear to agree with this value. I cannot understand how a correction for autocorrelation could possibly reduce the uncertainty range by a factor of three in these circumstances. The paper’s ±0.02 uncertainty range for the Cheng 0–2000 m 2005–2017 trend seems clearly wrong.

(v)  Using data only from the RCP4.5 scenario simulations, giving an ensemble mean lower than that for RCP2.6 only, for RCP8.5 only, and for all three scenarios combined, appears to be unjustified (even if had been disclosed).

Paragraph 7

Gleckler et al. 2016/Santer et al. 2014 (cited in our supplement) explicitly addressed the volcano impacts, after Outton et al. 2015, see our previous reply.
Schmidt, A., Mills, M. J., Ghan, S., Gregory, J. M., Allan, R. P., Andrews, T., et al. (2018). Volcanic radiative forcing from 1979 to 2015. Journal of Geophysical Research: Atmospheres, 123.

Nic Lewis comments:

This is irrelevant. Outten et al 2015 also included the omission in CMIP5 models of the impacts of recent volcanic eruptions, but found that it was fully offset by the net impact of misestimation of recent changes in other forcings.

Paragraph 8

We agree that the uncertainties in the revised Resplandy estimate are quite large, as we note when including them in the paper over the 2005-2017 Argo period (0.68 ± 0.60 Wm−2). Unfortunately showing the error bars of all the underlying observational series in main text figure as well as those of the climate models would have made it unreadable, and the fact that the Resplandy estimate does not extend back to 1971 means that it is left out of the “Updated OHC estimates compared with AR5” portion of the figure that does show individual series uncertainties. However, Resplandy et al does provide a novel approach to estimating ocean heat content, and we think their median estimate was worth showing alongside the three updated instrumental datasets, even if (unlike the other three datasets) Resplandy’s uncertainties are so large that they limit the claims that can be made regarding agreement with climate models.

Paragraph 9

We agree it is generally better to include the full definition of RCP8.5 to avoid any confusion, but in a press release for the general public we had to simplify. This does not impact the message in the published Science article. We note that there is an ongoing debate within the energy modeling and climate science community regarding RCP8.5 and the extent to which it represents a “business as usual” outcome, and that this is shifting with the availability of the SSP scenarios and the inclusion of a 7 w/m^2 forcing scenario in CMIP6. However, references to RCP8.5 as “business as usual” in the published literature are quite common, and the original paper presenting the RCP8.5 scenario (Riahi et al: explicitly refers to it as “a high-emission business as usual scenario”.

Closing statement by the paper’s authors:

  1. Thanks for the critique, an alternative set of values could be used in Fig.2 in our calculation for 700-2000m OHC in AR5. But the uncertainties are large in those early years.
  2. We believe that all of our conclusions are still valid:
  • After significant progress since AR5, the best OHC estimates show stronger warming than estimates featured in AR5 (0~63%), and they are also more consistent with each other.
  • The models are consistent with the best OHC estimates for the 1971-2010 period. While models are warming slightly faster than most of the observational records during the 2005-2017 period, this is expected because the volcanic aerosol effects are not fully included.

Nic Lewis closing comment:

I thank the authors for their constructive response. I concur that OHC uncertainties are large in the early years of the 1971–2010 period.

None of the authors’ responses refute any of my criticisms concerning factual errors and misleading statements in the paper.

In particular, presenting my method of calculating AR5 0–2000 m warming rates over 1971–2010 as alternative to their method is like claiming that calculating 4 –  2 = 1 is an alternative to calculating  4 – 2 = 2.


92 responses to “Is ocean warming accelerating faster than thought?

  1. Well done, Nic Lewis.
    An additional point concerning OHC in the ARGO era. I recently guest posted an article at WUWT, “ARGO fit for purpose?” Concerning OHC, the design intent accuracy (given randon ~3 degree lat/lon spacing and the minute water temperature diffences caused by large heat storage differences) was ~10W/square meter. At any of the annual heat deltas Nic references, it takes substantialy more than a decade after full 2005 deployment for ARGO to provide any rigor (the precise design intent term).
    And, it will take well more than two decades to reach any rigorous conclusions about the ‘true’ rate of OHC change, since the 2005 baseline from earlier OHC estimation methods is so uncertain.

    Therefor ALL the OHC papers cited, not just Cheng, woefully understate the uncertainty in their estimates.

    • This is not a comment that makes any sense to me. Positions of the floats that have delivered data within the last 30 days :

      Each of the instruments has an accuracy of 0.002 K and 2 mbar. So apart from throwing around arbitrary numbers (10 W/m2?) there is not much to claims of great uncertainty in the data. Conflating this with the problem of statistical significance of short term data on a phenomenon with long term variability confuses the issue.

      It is precise enough despite churlish skeptic memes. But there is something else happening.

      “We find a marked 0.83 ± 0.41 Wm−2 reduction in global mean reflected shortwave (SW) top-of-atmosphere (TOA) flux during the three years following the hiatus that results in an increase in net energy into the climate system. A partial radiative perturbation analysis reveals that decreases in low cloud cover are the primary driver of the decrease in SW TOA flux. The regional distribution of the SW TOA flux changes associated with the decreases in low cloud cover closely matches that of sea-surface temperature warming, which shows a pattern typical of the positive phase of the Pacific Decadal Oscillation.”

      At what temporal scales does SST in the Pacific vary?

      • It is usually recommended to read the referenced post before commenting if you do not want to look foolish. You are correct about the temp data accuracy, +/-0.002C. The SBE38 temperature sensor itself is NIST calibrated to +/-0.001C, with a sensor drift per 6 months (1/8 of an ARGO float life) of +/-0.001C.
        You are incorrect about ARGO OHC resolution.The mission intent design is explicitly a resolution of about 10W/square meter (0-2000 meter depth) over a ‘pixel’ (finest resolution) about 1000km on a side, ‘rigorous’ depending on which of the three main oceanography missions in a time frame of several months to much longer. That is referenced directly from the internationally agreed ARGO mission documents signed off in 2000.

      • Reading WUWT posts is never recommended. The 3° array – about 3 times greater than a 1000 km grid – resolves surface layer heat variance at seasonal scales.

        “Spectral analysis of altimetric data shows that, on a global basis, half of the variance in sea level is at wavelengths shorter than 1000 km (Wunsch and Stammer, 1995). If the climate signal of interest includes all wavelengths longer than 1000 km, then a float array with 3° spacing would resolve these signals with a signal-to-noise ratio of nearly 3:1. The unresolved variability — fronts, mesoscale eddies, etc. — has short time-scales, typically 10-20 days, compared with the seasonal and longer climate signals. Therefore, temporal averaging can further increase the signal-to-noise ratio. As a function of latitude, the halfpower point in the altimetric spectrum varies from 1300-km wavelength in the tropics to 700 km at 50°
        N (Stammer, 1997). This shortening of the spatial scales with increasing latitude is the reason why Argo requires a higher density of floats at high latitude. A 3 ° array has twice the density of instruments at 60 ° latitude as at the equator.”

      • Then you also missed my companion post at WUWT on Jason3 sat alt. NASAs newest SLR bird repeatable ‘same pixel ocean spot’ precision is speced at 3.3cm, even though the data is reported to 0.1mm. The spec is available on line at NASA. This limited resolution is inherent in 1. Earths lumpy reference ellipsoid, 2. varying surface waves (Jason2 processing assumed all were 2 meters, Jason3 attempts a real time resolution down to 0.4meters using peak trough relectance difference), and 3. varying atmospheric humidity which affects signal pulse duration.
        So sat alt based steric rise estimes are highly uncertain.

      • The scale of the processes captured by Argo is hundreds of kilometers and not meters. This has no relevance.

      • “It is usually recommended to read the referenced post before commenting if you do not want to look foolish.”

        The published literature on a purposefully designed and constructed 21st century monitoring system might be a better source than WUWT.

    • Жан Марк Ван Белл

      i think it is time, that we create some serious statistics.

      A temperature on both sides of the ocean, is one thing, but we should have some measuring points as well like they did in the end of 19th, beginning 20th Century: with weather ballons they found out the different behaviour of the Troposhere, Mesosphere… On you find a complex article but in fact the thickness of some spheres just varies and is much less above than else (and the speed of Earth is only 40.000 km per 24 hours at the equator.

      So, apart from all these very technical details, we may not become a that is Arab, not for an ‘unbelieving’ person, but for somebody whos heart is ‘covered’ (koffer/couvrir in lots of langauges as well and it resides from Hebrew and later Arab) from its brain that gets lots in the figures and no longer sees the ‘relativity’ in this all. We are talking about 0,0005% and forget the main factor oxygen and nytrogen in the natural cycle, and nature gets to the same balance of 21%/78%/1% mostly no reacting gases up to 100 km high, and only then the balance is different. Though the temperature cools down first with every kilometer higher, then goes up again, much higher cools down down again, to rise again when we probaly can say the atmosphere ends ‘somewhere’.

      And we may not forget to choose points at both sides of the equator… and give them the same weight in our statistical calculations (that is what this article and critic was about in essence).

      But thanks for the extra information to all here.

      Jean Marc VAN BELLE +je

  2. ristvan ==> It is possibly the most significant problem in Science today — the over-confidence in numerical values produced from huge data sets, while ignoring the very basic problems of the original accuracy and precision of measurements. For long term trends, the fact that older data in the data set is inadequately accurate (has huge uncertainty bars) is simply ignored (sometimes only downplayed) and the “mean” used as if it were both accurate and precise.

    It is similar to “The Epidemiologists’ Fallacy” in which medical/health studies claim to have “corrected for” large numbers of confounders in the data.

    • Kip

      I would add this to your excellent comment: often there is only one original measurement anyway, so we can’t average out say seven different readings and in the process discarding the highest and lowest.

      We also need to know the circumstandes under which they were taken and the accuracy of the instrument, let alone the mind set of the observer. This all means we need to be cautious in according magical status to data, as whilst We can often get a ‘generality’ getting a specific figure accurate to tenths of a degree is problematic.

      Whilst modern automatic readings may be more accurate than readings made by people, they still have their problems and in any case, like satellite readings, provide us data covering only the blink of an eye, from which it is dangerous to assert any definitive trends

      I think Judith needs to introduce the ‘extrapolation’ monster to her herds of climate beasts, it is a very large one.


    • @Kip Hansen Amplifying the over-confidence plague is the rhetorical power of computer generated graphics which visually imply an exaggerated degree of precision in the measurement. Compare with any manually plotted graph paper charts to see what I mean.

  3. Thanks Nic. Moreover a development of “qualtity management” in science is remarkable: in earlier days a peer reviewed response in the publishing journal was necessary to get a response from the authors. These days are gone IMO. Your critique relating Resplandy was followed by a quick response by the author(s) and also in this case the exchange was rather quick and well on target. This makes hope: (I)The discussion transferes to the public and doesen’t remain in the comfort zone of pay walled journals and (II) some PR-addicted editor takes more care :-)

    • Frank, This raises the same questions as the Replandy episode. Is the peer review process this broken? It seems yes. Is the quality of recent science papers this low? It seems yes. We should probably be grateful for the IPCC which seems to do a reasonably good job of presenting all the evidence, even though one never knows how many of the papers relied on are wrong or biased.

      I can’t help but return to the cultural issues that need to be addressed. At the very least, the huge cost of wasted human capital and talent is immense. I could say a lot more but will stop there.

      • From a layman’s perspective I would like to know if Dr. Curry could address this. What are the standards necessary to complete a peer review of a paper?

      • James, there are no clearly defined standards. Basically if the paper is well written and doesn’t contradict too strongly the consensus in the field, getting it published is time consuming but usually not difficult. I’ve gone through this process perhaps 40 times myself. In virtually all cases, only superficial change were suggested by reviewers, who in many cases I suspect didn’t read the paper in detail.

        This publication process has become something scientists have gotten very good at because their jobs and career advancement depend on it. There are various games that can be played to get more visibility for your work. Press releases have become very common and are often misleading or incomplete. Press stories based on these releases can be very wrong and exaggerated. Don’t believe anything you read in the popular media about science. It’s mostly rubbish.

        Peer review is an invention of the last 60 years and is mostly a rubber stamp. The one exception is if your paper challenges a widely held tenet of the leaders of the field or if the paper presents too many negative results that might cast doubt on the need for continued funding of the field. Generally the scientific literature has a strong positive bias. It gives the impression that the science is much more advanced and accurate than it really is.

        The reasons for this go very deep in scientific culture. The old days of individual researchers paid by universities to do teaching and research with little or no monitoring are over. You get paid the big bucks by bringing in lots of soft money to pay administrative “overhead” which has gone up scandalously, hire graduate students and post docs (who are paid poorly), and get lots of media attention. Further many scientific issues can only be tackled by large teams of people using massive computing resources to run very complex models.

        In more pessimistic moments I think the only way to get unbiased science these days is to have independent outsiders (like Nic Lewis who has done spectacularly good work) or retired scientists do their own quality checking on a paid basis.

      • dpy6629 ,it is easy to see where the problems lie when reading the statement relating to “simplified for the press release to the public”. I strongly suspect that 97% of the public are not interested in a boring claim regarding a few zettajoules made by people that have apparently been educated way beyond their intelligence.

        I had to laugh at the comparison chart for CMIP5 ,if i hadn’t looked at my screen from an angle i would have missed the error bars,tells me all i need to know.

        Anyone that thinks we know ocean heat content to single figure zettajoules in the top 2 m of global ocean ,never mind 2000 m is beyond delusional. If you know anyone that does believe this could you let them know i have a bridge for sale.

  4. michaelrath250

    Judith, could you comment on the following article? I found it very informative and concise.

    • Geoff Weatherford

      Thanks for that link. Extremely well done. And yes, I’d like to see Judith, Nic’s, and others feedback on what the person had to say.

    • The author of that article has no idea about atmospheric radiative physics.

      There is an equation which is derived from first principles, the equation of radiative transfer. It calculates absorption and emission through the atmosphere.

      It is unknown to the author of that article, along with most of the entertainingly named “climate skeptic” websites. (Ok, I know half of you are just running parody sites).

      It’s like debunking gravity without reference to the formula for gravitational attraction, instead pointing out that the explanation on an episode of Blue Peter wasn’t quite right.

      See Understanding Atmospheric Radiation and the “Greenhouse” Effect – Part Six – The Equations.

      Apply that equation, add more CO2 (or methane) and the outgoing long wave radiation drops, causing warming of the climate system.

      Water vapor has an effect on the radiative balance higher up in the atmosphere, but little effect close to the earth’s surface. Again apply the equation of radiative transfer. See Visualizing Atmospheric Radiation.

      Lots of stuff that matches your prior beliefs sounds very attractive. But open a physics textbook and you find equations. Where does the author debunk the equations of radiative transfer?

      It’s clear he has no clue that these equations even exist. Just like the army of confused people commenting on the lack of effect of CO2 on the radiative balance in various Science of Doom articles.

  5. Why do these papers continue to cite RP ? for changes in warming when there is no correlation of emissions to atmospheric CO2 nor temperature? (

    • I don’t know what “RP” stands for, but if you don’t think that emitting CO2 into the atmosphere raises the level of CO2 in the atmosphere, you’re mistaken.

      It is true that many other factors also affect atmospheric CO2 levels.

      It is true that negative feedbacks are removing CO2 from the atmosphere about half as fast as mankind is adding it, with the result that CO2 levels are only increasing by 2 to 2.5 ppmv per year, rather than 5 ppmv per year.

      It is true that the rate at which those feedbacks remove CO2 from the atmosphere depends mostly on the current atmospheric CO2 level, rather than on the current rate of emission.

      But those facts do not mean that adding CO2 to the atmosphere doesn’t increase the level of CO2 in the atmosphere.

  6. Pingback: Ocean Warming | Transterrestrial Musings

  7. My thanks to Nic Lewis and to the authors ( Lijing Cheng, John Abraham, Zeke Hausfather and Kevin Trenberth) for this contribution to the discussion.

  8. “Although the authors were unlucky to have an editor who was more concerned with presentation than scientific content, they, not the editor, are ultimately responsible for the paper.”

    This is the crux of the matter in so-called climate “science.” In the same way that “narrative journalism” is currently being taught as best practice in journalism schools (i.e. lying for a good cause), ‘narrative science’ is becoming accepted as better than mere ‘science’ (which is disinterested, boring and dubious because it does not serve a moral cause).

    Climate scientists may not be actively engaged in this perversion of their discipline (that would be risky, and therefore require some degree of courage – which is not a quality to be found in any academy yet invented). Instead most of their sins are those of omission (they allow the unscrupulous to misrepresent their work and say nothing).

  9. Pingback: Another Headline-Grabbing Ocean Warming Study Is Full Of ‘Factual Errors And Misleading Statements,’ Scientist Says - Conservative Daily News

  10. Nic, you have already become the bête noir for prominent climate alarmists. You have my sincerest admiration, but I don’t know if I should congratulate you. You better watch out.

  11. Pingback: Another Headline-Grabbing Ocean Warming Study Is Full Of ‘Factual Errors And Misleading Statements,’ Scientist Says – TheConservativeWorld

  12. sheldonjwalker

    ❶①❶① . . . The recent Slowdown – on trial . . .

    Alarmists have started a legal battle, in an effort to convict the recent Slowdown of a serious crime. The crime in question is, “impersonating a real Slowdown”. This heinous crime carries a maximum sentence of 20 years of watching Al Gore “documentaries”.

    The trial is about to begin. We have managed to get our “climate reporter”, Sheldon Walker, on to the jury hearing the case against the recent Slowdown. We asked Sheldon if he thought that it was “fair”, for him to be on the jury? Sheldon replied, “Is it “fair”, that Alarmists won’t admit that there was a small, temporary Slowdown, that doesn’t have any significant long-term implications for global warming”?

    Sheldon is prepared to go to extreme lengths to help his friend. He has taught himself to text message with his toes, using a cellphone that is hidden in his shoe. Sheldon will be sending us text message “reports” from inside the room where the jury members are deliberating. These text message reports will be limited to 160 character per text message (Sheldon refuses to use Twitter), so Sheldon will use abbreviations where necessary.

  13. Has no one considered that the Ishi dataset only goes to 1500 m depth, and that it thereby may miss up to 10% of the warming in the 0-2000 m layer?

    Also, I don’t think that the period with good Argo data starts in 2005. The Argo array wasn’t fully deployed until 2007, and there were also widespread pressure sensor errors in the early years. There is significant disagreement between various OHC datasets in the years 2004-2006.
    I understand that it may be tempting for motivated minds to start a trend with the bump that some datasets have around 2005, but my advice is to use only the period with really good Argo data 2007- now (2017 or 2018).

    I also wonder why the authors blend the Levitus dataset with Domingues? It must be better to use pure datasets 0-2000 m. From 1955 and on, we have Levitus, IAP/Cheng, Ishi (if one compensates for the missing 1500-2000 m), and a fourth alternative could be EN4.

    • Fair point re the Ishii dataset, although the likely bias is quite small.
      I largely agree with you re Argo, although I think Argo 0-700 m coverage wasn’t bad in 2005 and 2006 and the overall effect on 2005-17 trends of Argo inadequacies in 2005 and 2006 is likely to be small.
      I think the authors blend the Levitus dataset with Domingues 0-700 m because that is what AR5 did.

      • I don’t think that the depth bias in Ishii is small. According to data from Argo Marine atlas, the increase of heat content in the 1500-2000 m layer, between 2007 and 2017, is 1.13*10^22 J. That is around 10% of the Ishii OHC change.

        I have included a corrected Ishii dataset in the following chart, which compares various OHC estimations with model data from Cheng et al 2019, during the period with really good Argo data, 2007-2018.

        The observational datasets is close to the middle of the model ensemble.
        Still, I believe that the selection of models is biased towards more warming. Its not normal that the rcp2.6 average has larger OHC increase than the rcp4.5 average, not even by chance..

  14. Nic, thanks for this post. One thing that fascinated me about Judith’s preceding post was Wunsch’s estimate that 20% of the heat the ocean is taking up comes from undersea volcanoes. If that’s true it would mean that the models are even further off in their estimate of heat uptake since they only capture the heat transferred from the atmosphere or the sun.

    • That number has been around for a long time:

    • dpy6629, I think that the text JCH posted is a fair summary. It is only if geothermal heating changes from its long term value that there would be implications for ocean heat uptake and comparisons with models.

      • AGG ocean warming is nominally the result of a reduction in the temperature gradient across the ocean/atmosphere boundary reducing heat loss. The source of the heat is irrelevant – but geothermal heat has implications for the time to renew ocean/atmosphere thermal equilibrium. The latter has been predicated on very slow mixing to depth – hence heat in the pipeline.

        But it is just all too simple physical reasoning. Satellites all show warming in SW and cooling in IR.

  15. Climate change from human activities mainly results from the energy imbalance in Earth’s climate system caused by rising concentrations of heat-trapping gases.

    ” About 93% of the energy imbalance accumulates in the ocean as increased ocean heat content (OHC).”

    Why? On several levels. Nothing about heat accumulation on land which has lots of shallow subsurface water which can retain heat [some of the missing heat?]

    Say the CO2 went up 4ppm. and stayed at this new level for 1 year, 100 years, 10000 years.
    Would the ocean energy imbalance still be 93%?
    Fair question. There is a lot more water than air to transfer the energy into. Would the rate be 93% at a lower PPM change.
    Or would it be 93% if the PPM was going up at 0.04 PPM for a hundred years?

    Or is it that the expected Resplandy Trenbath, Gavin heat for a rise of 40% in CO2 which is missing is dictated at a rate which means 93% has to have gone into the sea if the SST and land temps have not risen enough.
    Even if they can only find 55% at the moment.
    Resplandy was claiming a 40% increase over what has been detected to date.

  16. ” The ocean record of this imbalance is much less affected by internal variability and is thus better suited for detecting and attributing human influences (1) than more commonly used surface temperature records.”

    A weird comment given the long lasting nature of ocean currents which can give short 2 year changes like El Nino and prolonged ones like JCH’s favourites. Air with its ephemeral time span repeats on a much shorter daily scale and despite the large daily noise has a much softer longterm imbalance than the ocean .
    Plus it is very hard to measure temperature changes of 0.001 C over a year accurately. OHC, Pielke favourite, extremely untrustworthy because of the measurement difficulties.

    • You really pay no attention at all. I embrace all three phases of ENSO as all three are warming. “The next big La Niña!” I roll on the floor. It’s daft.

      Why would I be upset by the specter of another record warmest La Niña in the instrument record? Do ’em back-to-back! Lol. Bring it.

      The coolest record warmest year in the warmest decade of the instrument record was 2014: ENSO neutral: hotter than all prior El Niño record years.

      Why no land heat content in an OHC paper? When I was a kid we buried water pipes at 6 foot. Only the village idot would bury them at 5 1/2 feet as the pipes would freeze and break. Now the village idot can do whatever he wants.

      • “Why no land heat content in an OHC paper? When I was a kid we buried water pipes at 6 foot. Only the village idiot would bury them at 5 1/2 feet as the pipes would freeze and break.”
        JCH, did you live in Northern Alaska as a child?
        2 meters is quite a depth to put pipes and does not apply to much of the world, does it?
        And in your example only in winter.
        Thanks for pointing out that heat and cold can be present at depth on land and is grossly under represented.
        Most pipes where I live run along just subsurface and are easy to maintain.

      • The North Dakota border.

      • ok

  17. Nic, Could it be that Cheng et al in their figure 2 are only looking at 0-2000m and thus the below 2000m heat is omitted from all calculated rates of warming? Your AR5 heat content rate of change however includes this deep ocean heat so its perhaps not comparable.

    • dpy6629,
      Both Cheng and I have made an adjustment to deduct warming below 2000 m from the AR5 sub-700 m depth warming estimate. But my adjustment is correct and theirs is wrong. See the Addendum that I have recently added at the bottom of my article, just below the references.

      • Nic

        Below 2000 metres? That must be the highly speculative purkey and Johnson paper.

        I was at a climate conference a couple of years ago when Thomas stocker stated categorically said that we did not have the technology to measure the heat below 2000 metres.

        This sort of extremely hypothetical study is the sort of thing that gets climate science a bad name.


      • To which the greatest oceanographer of all time Carl got almost the same answer.

      • Also, Thomas Stocker cites Purkey and Johnson in his oceanography book:

      • Saint Stephen with a rose, in and out of the garden he goes. Country garden in the wind and the rain, wherever he goes the people all complain.

        ~garcia, hunter & lesh

      • TonB
        did Purkey and Johnson measure temp and thus heat below 2000 m or between 700 and 2000 or model it all down to precise estimates w error bars explained.

        Is Argo II or plus measuring in place down to depth now and what extent of coverage?

        This seems like 10,000 angels on teh head of a pin.

      • Jch

        Highly ambivalent he is too. So ‘suggest’ is now a scientific term, is it?
        Stocker made his remark in answer to a question at a climate conference around 2013. Richard Betts was also there as were many of the great and the good.(and me)

        As Scott suggests, angels dancing on the heads of pins. This sort of speculation, infilling, interpolation and wishful thinking does no credit to the climate science community

        We simply do not know


      • JCh

        Let’s repeat that. “We do not have the technology to measure the deep oceans’.

        . That is to say. we do not know.

        How on earth do you manage to get to, if they are wrong it can’t be by much? Is guessing or speculating or suggesting the way we carry out post modern climate science?

        We just do not know.


      • Tony

        “Is guessing or speculating or suggesting the way we carry out post modern climate science?”


    • Thanks Nic, that clarifies it. Your graph makes their error clear.

  18. Given that at all altitudes temperature is colder than standard sea level, can a CO2 molecule at 40km and 250K (arbitrary) absorb and reradiate standard sea level upwelling ground radiation without that temperature being reduced by surrounding air temperature?

  19. Pingback: Is ocean warming accelerating faster than thought? In a word, no. | Watts Up With That?

  20. Pingback: Is ocean warming accelerating faster than thought? In a word, no. |

  21. I am grateful to Judith and Nic for their steadfast application of science and logic to the climate change debate.

  22. The error identified by Nic and discussed in his latest update seems like a simple mathematical mistake that even a non-scientist like me with a bit of maths can just about understand. Can it really be that such an error would appear in a reputable scientific paper ?

    • Yes. All the time. See Nic’s critique of the Replandy paper for another very recent example. And in economics, there is a fairly recent example where two Harvard professors made a basic Excel speadsheet error caught by a grad student at a different university, which completely negated their result.

    • If seven all-star referees can’t see the bleedin’ obvious in an NFL championship game, then one might suppose that anything is possible…

  23. OHC estimates are all over the place.


    Some of it is down to ‘climatologies’ – used to infill sparse data.


    Some of it is due to the period of averaging over interannual to decadal variability – the latter being the background signal of perpetual Earth system change.

    Rather than quibbling about minor differences in trend calculations as Nic Lewis does – it may be better to discount pre 1990’s data and question the XBT/Argo splice. This is a far cry – I note – from rejection of greenhouse gas energy implications or denial of ocean warming.

    Some 3 in 10 decades have SST and total energy changes in the system of opposite signs – – as a result of energy transfers between oceans and atmosphere. Giving relatively large changes in atmospheric heat content and proportionately little change in ocean heat. Change in OHC is more a function of net downwelling radiant energy flux. This in satellite data is all SW warming and IR cooling – clouds and aerosols. It is still all about how much is natural and how much anthropogenic?

  24. Question to Dr. Curry

    The Clausius-Clapeyron equation describes the phase transition of matter.
    dP/dT = L/(T dV)
    dP/dT = derivative of co-existence curve, L = specific latent heat, T = temperature, dV = specific volume

    Expressed in terms of saturation vapor pressure:
    dP/dT = L P/(R T^2)
    P = saturation vapor pressure, R = gas constant

    Solving this equation for P of water gives temperature-dependent values lower than atmospheric pressure. Of course at atmospheric pressure, T = 100 C or the boiling point of water. The question is why does water evaporate at atmospheric pressure below 100 C? According to Clausius-Clapeyron, it should not. But it does!

    • Dr.S: Because the temperature of a volume of water is described by the average speed of the molecules inside. There are a few which ave a much lower speed ( colder) and a few with much higher speed (hotter). The “hottest” can outgo the volume of water ( they evapurate) leaving the colder ones behind. The volume of water will cool down because there is a new lower average speed of all the molecules inside.

      • If there’s wind, it cools the water surface and evaporation should decrease by your description. But the opposite happen, evaporation increases.

      • Dr.S: If windspeed is enhanced the water saturated layer ( hindering further evaporation) above the watersurface will be blown away and the evaporation is enhanced. These issues are very basic physics better to google before asking others here IMO.

      • Even without wind the saturated air will be removed by convection because it is warmer due to hotter water vapor molecules. Show me your references to read. TY

      • “Show me your references to read.” I would suggest a very basic textbook or wiki :-)

      • A simple experiment you can do at home.

        In enclosed bathroom, observe how long for a wet hand towel to dry by evaporation. Wet the towel again. This time use a hair dryer but turn off the heater so the air blown is at ambient temperature. Notice the towel will dry quicker. The air blown has the same humidity as air inside bathroom. This simple experiment proves the result is not due to removal of saturated air.

        Frankclimate said this is very basic physics. Okay show me links to textbooks or lecture notes that explain this phenomenon. Einstein said we discover new things by asking silly questions that only children ask. We stop learning when we stop asking questions.

        I agree with Popesclimatetheory but why does wind make a lot of difference?

      • Dr. Strangelove, the evaporation rate increases when the humidity is lower and the temperature is higher in the air immediately adjacent to the moist towel.

        Q: So, what happens to the immediately adjacent air when moisture evaporates from the towel?

        A: It gets moister, and cooler.

        That reduces the rate of evaporation from the towel, until that cooler, moister air is moved away, and replaced by drier, warmer air.

        Your fan or hairdryer causes that cool, moist air to be moved away and replaced by drier, warmer air more quickly.

      • Initially the hair dryer removes moist air above wet towel. After a while the dryer mixes the air in the room to uniform humidity and temperature. There is more to this than just removal of moist air

    • Solving this equation for P of water gives temperature-dependent values lower than atmospheric pressure. Of course at atmospheric pressure, T = 100 C or the boiling point of water. The question is why does water evaporate at atmospheric pressure below 100 C? According to Clausius-Clapeyron, it should not. But it does!

      There is a balance of water vapor in the atmosphere. The vapor pressure is a function of the water temperature. When humidity is 100% no more water evaporates, with or without wind. When humidity is less than 100%, wind makes a lot of difference. If Clausius-Clapeyron says water should not evaporate below 100 C then discount everything Clausius-Clapeyron says.

    • Since nobody can show me any textbook that explains my hair dryer experiment, I invented my Vortex theory from fluid mechanics and kinetic molecular theory.

      Conservation of energy in fluids is given by Bernoulli’s equation:
      v^2 /2 + g z + P/d = k
      where: v = wind velocity, g = gravitational acceleration, z = elevation, P = static pressure, d = mass density of air, k = constant
      Let: z = 0, multiply the equation by d to obtain the pressure form of Bernoulli’s equation:
      d v^2 /2 + P = k = Pa
      The first term is dynamic pressure. Pa is standard atmospheric pressure (depends on elevation)
      P = Pa – d v^2 /2

      Atmospheric pressure is related to temperature at constant volume by Gay-Lussac’s law:
      Pa/Ts = k = P/Td
      Where: Ts = static air temperature (v = 0), Td = dynamic air temperature (v > 0)
      Substitute P from Bernoulli’s equation:
      Pa/Ts = (Pa – d v^2 /2)/Ta
      Td = Ts/Pa (Pa – d v^2 /2) Eq. 1
      Equation 1 shows that dynamic air temperature (Td) is inversely proportional to wind velocity (v)

      Temperature translates to mean molecular velocity by kinetic molecular theory:
      v^2 = 8/pi k Td/m Eq. 2
      where: v = mean molecular velocity, k = Boltzmann constant, m = mass of water molecule
      Equation 2 shows that mean molecular velocity (v) is proportional to dynamic air temperature (Td). Gas molecules with velocities greater than mean velocity can escape atmospheric pressure and evaporate.

      Kinetic temperature can be derived from ideal gas law:
      Ps V = n R T
      T = Ps/(n/V R)
      Where: T = kinetic temperature, Ps = saturated vapor pressure, n/V = molar density, R = ideal gas constant
      Molar density can be expressed in terms of humidity:
      n/V = Ha/M
      where: Ha = absolute humidity, M = molar mass of water
      Substitute n/V to the gas equation:
      T = Ps/(Ha/M R)
      T = Ps M/(Ha R) Eq. 3

      Gas molecules in water vapor have different velocities given by Maxwell- Boltzmann distribution:
      P = (m/(2 pi k T)^0.5 e^(-m v^2 /2 k T)
      Where: P = probability of gas molecules with velocity v
      Integrating the equation from mean molecular velocity (v) to infinity gives the evaporation ratio (E). This is the ratio of mass of water vapor and mass of water exposed to air before evaporation.
      E = ∫ (m/(2 pi k T)^0.5 e^(-m v^2 /2 k T) ] v to ∞ Eq. 4

      Equation 4 is my Evaporation equation. This integral equation can be expressed as a summation of an infinite series that converge to a finite value equal to the evaporation ratio:
      E = ∑ E1 + E2 + E3 +… + En where n = ∞
      The evaporation ratio can be determined by solving the four equations simultaneously (Equations 1 to 4)

      These four equations describe evaporation for laminar air flow. In Part 2, I will derive the equations for turbulent air flow (where the name Vortex theory comes from). Yup this is very basic physics, except nobody can point to any textbook that discusses it.

    • Part 2 of my theory
      Turbulent flow occurs when the wind velocity is at angle with a water surface. The collision of wind and water breaks the laminar air flow. In turbulent flow, vortices are formed over the water surface. A vortex is like a tiny tornado where air move in circles at high speed. The center of a vortex is low pressure. Turbulent flow is more complicated than laminar flow because there are many vortices at varying speeds.

      To quantify turbulence, I invented the Strangelove number (s). This is similar to the Reynolds number in fluid mechanics, which differentiates laminar and turbulent flows. Reynolds number applies to fluid flow in pipes whereas Strangelove number applies to air flow over water or wet surfaces.
      s = 1 + tan B + 4 a/y
      where: B = angle between wind velocity and water surface 1 is turbulent. Laminar flow is when wind velocity is parallel to the water surface and no waves. When there are different waves in turbulent flow, s is the average of Strangelove numbers.

      Strangelove equations:
      Td = Ts/Pa (Pa – d v^2 /2)
      v^2 = 8/pi k Td/m
      T = Ps/(k Ha/m + Pa/Ts)
      E = ∫ (m/(2 pi k T)^0.5 e^(-m v^2 /2 k T) ] v to ∞
      e = s (C + D v) (Hs – H) A

      (Note: I modified Eq. 3)
      Since turbulent flow creates vortices and my theory has five equations, I will name it the Vortex V theory (pun intended). I invented this theory to answer how a hair dryer works. We discover new things by asking silly questions that only children ask.

  25. “The paper does not directly claim that ocean warming is accelerating faster than thought; that is the headline of The New York Times article about the paper.”

    Did any of the paper’s authors respond to NYT to correct that mistake? No, I didn’t think so either. It’s always seems to be like that with alarmists. They keep their scientific reputation by allowing someone else to make false claims. But when are those false claims ever corrected by alarmists authors?

    Someone with time & patience should write a paper on this.

    • The establishment would gain much credibility as real scientists if they would publicly refute some of the more outrageous apocalyptic stories. Don’t expect that to happen any time soon since they have a symbiotic relationship with the media and at the core they are all activists. When the fate of the world hangs in the balance, it’s magical how the mind can rationalize not making a fuss.

      This whole multidecadal group think on steroids will give social psychology text books fodder for case studies for generations to come. What fun it would be to read them all. I’ll alert my grandkids to watch for them in 30 or 40 years.

  26. Nic, The scientific community is blessed to have you doing proper peer review. Please keep up the good work!

    People are being psychologically damaged by the omnipresent fear based propaganda surrounding climate science. Voices of reason are needed to bring sanity and reason back to science.

    Many thanks and best regards!

  27. Treemometers were abandoned from the 60s on because of the decline. Are Landometers transitioning to oceans meters because the former only show the max temp not rising fast enough?

  28. Pingback: Nachrichten­medien verliehen fehler­hafter Klima­studie große Aufmerk­samkeit – EIKE – Europäisches Institut für Klima & Energie

  29. Pingback: Weekly Climate and Energy News Roundup #345 |

  30. Pingback: Another Headline-Grabbing Ocean Warming Study Is Full Of ‘Factual Errors And Misleading Statements,’ Scientist Says | Prime Patriot

  31. Reblogged this on Climate Collections.