A Harde response

Earlier this year, I wrote a post about a paper by Hermann Harde that argued that most of the rise in atmospheric CO2 was natural. If you want more details of why this suggestion is nonsense, you can read my earlier post. What I was going to mention in this post is that a number of us have just published a response.

The history of this is essentially that Gavin Schmidt, as he suggested in this Realclimate post, set up an Overleaf document and contacted those who had shown interest. It was lead by Peter Köhler, and colleagues, from the Alfred-Wegener-Institut, but also included myself, Eli Rabett, and Richard Zeebe from the University of Hawaii at Manoa. Gavin Cawley also provided some very valuable comments and suggestions.

I don’t need to say too much about the details of our paper. It essentially highlights that the Harde paper confuses the residence time of an individual molecule (years) with the adjustment time for an enhancement of atmospheric CO2 (centuries). It also points out that you can’t model the evolution of atmospheric CO2 with a single equation. You need to consider at least two reservoirs (atmosphere and surface ocean) and this requires at least two equations that are solved simultaneously.

We also point out that it’s important to consider the Revelle factor, which limits how much of our emissions can be taken up by the oceans (we would expect – depending on how much we emit – that 20-30% of our emissions will remain in the atmosphere for thousands of years). Additionally, there were issues with Harde’s application of his model to paleoclimate, and there were a number of papers that he really should have cited (such as citing Essenhigh 2009 paper, while failing to cite Gavin Cawley’s response).

We end our paper by suggesting that Harde’s paper be withdrawn. I’m normally a little uncomfortable with suggesting that a paper that does not involve fraud, or plagiarism, be withdrawn. However, Harde’s paper is so obviously flawed that it is remarkable that it made it through the editorial, and review, process without being rejected. It might be better if it were withdrawn, but at least there is now a formal response that highlights the numerous issues.

This entry was posted in ClimateBall, Gavin Schmidt, Research, Science and tagged , , , , , , , . Bookmark the permalink.

22 Responses to A Harde response

  1. Kevin ONeill says:


  2. Marco says:

    Quite a few papers have been retracted recently for making mistakes. In those cases, however, it supposedly(*) is the author that takes the initiative.

    (*) In reality, in some cases there very likely is pressure from the Editor(s) to retract.

    COPE actually states that “Journal editors should consider retracting a publication if they have clear evidence that the findings are unreliable, either as a result of misconduct (e.g. data fabrication) or honest error (e.g. miscalculation or experimental error)”

    Moreover: “Retraction should usually be reserved for publications that are so seriously flawed (for whatever reason) that their findings or conclusions should not be relied upon.”

    Harde’s paper definitely fits this category…

  3. Marco,
    Thanks, I hadn’t seen the suggestion that retraction should be considered if a paper’s findings are unreliable. It does make sense in cases where it’s obvious that the paper’s results are clearly flawed.

  4. dikranmarsupial says:

    Bravo indeed!

    I think the reason Prof. Harde’s paper made it through review can be found in the author information pack for the journal:

    Please submit, with the manuscript, the names, addresses and e-mail addresses of five potential
    referees. Note that the editor retains the sole right to decide whether or not the suggested reviewers
    are used.

    which seems to me a recipe for pal-review, I suspect I can guess some of those likely to have been on any such list. I suspect that is also how the earlier carbon cycle paper by Humlum et al. (which also was the subject of peer-reviewed comments) that was published in the same journal. IMHO journals should never do this. For any paper within the scope of the journal, there should be an action editor sufficiently familiar with the sub-field to identify suitable reviewers for themselves and if they can’t they shouldn’t handle it. However, they are only human and we all make mistakes…

    The problem with the paper is made very obvious by the fact that Prof. Harde doesn’t provide a plot of the output of his models against the observations, which is a very natural thing to do if you want to show that your model provides a good fit to the data. The closest we get in the paper is figure 3:

    I tried reproducing this myself, but plotting the actual ice core data, rather than giving error bars:

    Not so good. In particular it is clear that in the ice core data, the relationship between temperature and CO2 is linear, and the modern observations don’t fit the pattern of the ice core data (and neither does Prof. Harde’s model). It isn’t at all clear how the error bars for the observations for Harde’s figure 3 were calculated, I suspect they may be “subjective”.

    Note there are substantial problems with my diagram as well (I learned a lot from my discussions with the authors of the paper!), particularly the Vostok temperatures are regional Antarctic temperatures, rather than global temperatures. IIRC the Vostok core includes interglacial periods that were about as warm as it is now (if not warmer?), so if Prof. Harde’s model is correct, it is hard to see why CO2 concentrations were not approaching 400ppm then (I’d need to go and check the details on that, so caveat emptor).

    Anyway, it is a shame that the effort it takes to respond to ill-informed misleading papers, such as this one, is so much higher than it is apparently to produce them, but I’m glad someone took the trouble to do it in this case.

  5. dikranmarsupial says:

    Here is my attempt at implementing Prof. Harde’s model of post-industrial conditions, using an ODE solver to drive the model with observed temperatures:

    This shows why a temperature driven model for atmospheric CO2 with a short “residence” time can’t explain the observed rise in atmospheric CO2. The temperature data has lots of decadal+ scale variability, and a “residence” time of only 4 years isn’t enough to smooth it out, so it will predict that CO2 will rise with similar decadal+ scale variability. The trouble is that it doesn’t, the observed rise in atmospheric CO2 is pretty smooth, and a simple (and conventional) one-box carbon cycle model (based on the one in my paper) with an approx. 50-70 year adjustment time (don’t worry, it also has a residence time of 4-5 years ;o) does a much better job.

  6. dikranmarsupial says:

    Cheers, I’m glad there was a suitable venue to share the diagrams at last! I like your new avatar/icon BTW, if I had an avatar representing my work it would have to be an animated GIF of (apparently meaningless) numbers changing very slowly ;o)

  7. “It also points out that you can’t model the evolution of atmospheric CO2 with a single equation. “

    I can. It’s variously referred to as dispersive diffusion — a classical diffusion equation solved with a MaxEntropy distribution of diffusivity values. This maps to the heuristically determined set of factors known as the Berne formulation.

  8. Geo,
    Well, okay, but that is presumably determining net flux and your diffusion coefficient is presumably implicitly incorporating information about the two reservoirs.

  9. The numerical solution to diffusion in that geometry is a slab model, which is an infinite set of differential equations representing the flow between an infinite number of reservoirs. The analytical solution to that geometry is an erf. As an analogous application, the ocean/atmosphere interface is equivalent to the solid/vapor interface used for diffusional doping of a semiconductor. The industry has this process characterized very well and the erf works effectively.

    But with a range of diffusivities representing the different pathways to CO2 sequestration, the erf can be replaced with another formulation that takes into account the variability. So the Berne formulation is a heuristic for the description of the actual diffusional physics. They use the heuristic because that’s all they need apparently. There’s nothing wrong with it but they will never be able to approximate the diffusional fat-tail at infinite time because they are using a set of exponentials instead of an erf. The slowest exp decay is what they use for the fat tail.

  10. Everett F Sargent says:

    So, who will reply to Harde’s latest missive missile that misses massively? …
    Radiation Transfer Calculations and Assessment of Global Warming by CO2

    “Including solar and cloud effects as well as all relevant feedback processes our simulations give an equilibrium climate sensitivity of Cs = 0.7°C (temperature increase at doubled CO2) and a solar sensitivity of Ss = 0.17°C (at 0.1% increase of the total solar irradiance). Then CO2 contributes 40% and the Sun 60% to global warming over the last century.”

    Rust Never Sleeps

  11. Everett,

    So, who will reply to Harde’s latest missive missile that misses massively? …
    Radiation Transfer Calculations and Assessment of Global Warming by CO2

    I don’t know if anyone is going to bother with that one.

  12. cce says:

    Is there a link to this paper?

  13. cce,
    Good point. I managed to forget to provide a link in the post (fixed now). It’s here, in case you don’t want to read the post again 🙂

  14. angech says:


  15. KHome1990,
    Below is a figure that might help. It shows the cumulative emissions for the 4 different Representative Pathways (RCPs) and shows how the warming depends more on how much we emit, rather than on how fast. It also shows atmospheric concentrations (bubbles) and each dot on each line is a decade. The range is represented by the width band.

  16. Marco says:

    Wrong thread, ATTP!

  17. Everett F Sargent says:

    Figure 2.3 | Global mean surface temperature increase as a function of cumulative total global carbon dioxide (CO2) emissions from various lines of evidence. Multi-model results from a hierarchy of climate carbon-cycle models for each Representative Concentration Pathway (RCP) until 2100 are shown (coloured lines). Model results over the historical period (1860 to 2010) are indicated in black. The coloured plume illustrates the multi-model spread over the four RCP scenarios and fades with the decreasing number of available modelsin RCP8.5. Dots indicate decadal averages, with selected decades labelled. Ellipses show total anthropogenic warming in 2100 versus cumulative CO2 emissions from 1870 to 2100 from a simple climate model (median climate response) under the scenario categories used in WGIII. Temperature values are always given relative to the 1861–1880 period, and emissions are cumulative since 1870. Black filled ellipse shows observed emissions to 2005 and observed temperatures in the decade 2000–2009 with associated uncertainties.(WGI SPM E.8, TS TFE.8, Figure 1, TS.SM.10, 12.5.4, Figure 12.45, WGIII Table SPM.1, Table 6.3)

    (page 63)

  18. Comment on “Scrutinizing the carbon cycle and CO2 residence time in the atmosphere” by H. Harde, needed proof-reading.

  19. Pingback: 2017: A year in review | …and Then There's Physics

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s