New paper shows transient climate response less than 2°C

See also: Why the new Otto et al climate sensitivity paper is important – it’s a sea change for some IPCC authors

New energy-budget-derived estimates of climate sensitivity and transient response in Nature Geoscience

Guest post by Nic Lewis

Readers may recall that last December I published an informal climate sensitivity study at WUWT, here. The study adopted a heat-balance (energy budget) approach and used recent data, including satellite-observation-derived aerosol forcing estimates. I would like now to draw attention to a new peer-reviewed climate sensitivity study published as a Letter in Nature Geoscience, “Energy budget constraints on climate response”, here. This study uses the same approach as mine, based on changes in global mean temperature, forcing and heat uptake over 100+ year periods, with aerosol forcing adjusted to reflect satellite observations. Headline best estimates of 2.0°C for equilibrium climate sensitivity (ECS) and 1.3°C for the – arguably more policy-relevant – transient climate response (TCR) are obtained, based on changes to the decade 2000–09, which provide the best constrained, and probably most reliable, estimates.

The 5–95% uncertainty ranges are 1.2–3.9°C for ECS and 0.9–2.0°C for TCR. I should declare an interest in this study: you will find my name included in the extensive list of authors: Alexander Otto, Friederike E. L. Otto, Olivier Boucher, John Church, Gabi Hegerl, Piers M. Forster, Nathan P. Gillett, Jonathan Gregory, Gregory C. Johnson, Reto Knutti, Nicholas Lewis, Ulrike Lohmann, Jochem Marotzke, Gunnar Myhre, Drew Shindell, Bjorn Stevens, and Myles R. Allen. I am writing this article in my personal capacity, not as a representative of the author team.

The Nature Geoscience paper, although short, is in my view significant for two particular reasons.

First, using what is probably the most robust method available, it establishes a well-constrained best estimate for TCR that is nearly 30% below the CMIP5 multimodel mean TCR of 1.8°C (per Forster et al. (2013), here). The 95% confidence bound for the Nature Geoscience paper’s 1.3°C TCR best estimate indicates some of the highest-response general circulation models (GCMs) have TCRs that are inconsistent with recent observed changes. Some two-thirds of the CMIP5 models analysed in Forster et. al (2013) have TCRs that lie above the top of the ‘likely’ range for that best estimate, and all the CMIP5 models analysed have an ECS that exceeds the Nature Geoscience paper’s 2.0°C best estimate of ECS. The CMIP5 GCM with the highest TCR, per the Forster et. al (2013) analysis, is the UK Met. Office’s flagship HadGEM2-ES model. It has a TCR of 2.5°C, nearly double the Nature Geoscience paper’s best estimate of 1.3°C and 0.5°C beyond the top of the 5–95% uncertainty range. The paper obtains similar, albeit less well constrained, best estimates using data for earlier periods than 2000–09.

Secondly, the authors include fourteen climate scientists, well known in their fields, who are lead or coordinating lead authors of IPCC AR5 WG1 chapters that are relevant to estimating climate sensitivity. Two of them, professors Myles Allen and Gabi Hegerl, are lead authors for Chapter 10, which deals with estimates of ECS and TCR constrained by observational evidence. The study was principally carried out by a researcher, Alex Otto, who works in Myles Allen’s group.

Very helpfully, Nature’s editors have agreed to make the paper’s main text freely available for a limited period. I would encourage people to read the paper, which is quite short. The details given in the supplementary information (SI) enable the study to be fully understood, and its results replicated. The method used is essentially the same as that employed in my December study, being a more sophisticated version of that used in the Gregory et al. (2002) heat-balance-based climate sensitivity study, here. The approach is to draw sets of samples from the estimated probability distributions applicable to the radiative forcing produced by a doubling of CO2-equivalent greenhouse gas atmospheric concentrations (F) and those applicable to the changes in mean global temperature, radiative forcing and Earth system heat uptake (ΔT, ΔF and ΔQ), taking into account that ΔF is closely correlated with F. Gaussian (normal) error and internal climate variability distributions are assumed. ECS and TCR values are computed from each set of samples using the equations:

(1) ECS = F ΔT / (ΔF − ΔQ) and (2) TCR = F ΔT / ΔF .

With sufficient sets of samples, probability density functions (PDFs) for ECS and TCR can then be obtained from narrow-bin histograms, by counting the number of times the computed ECS and TCR values fall in each bin. Care is needed in dealing with samples where any of the factors in the equations are negative, to ensure that each is correctly included at the low or high end when calculating confidence intervals (CIs). Negative factors occur in a modest, but significant, proportion of samples when estimating ECS using data from the 1970s or the 1980s.

Estimates are made for ECS and TCR using ΔT, ΔF and ΔQ derived from data for the 1970s, 1980s, 1990s, 2000s and 1970–2009, relative to that for 1860–79. The estimates from the 2000s data are probably the most reliable, since that decade had the strongest forcing and, unlike the 1990s, was not affected by any major volcanic eruptions. However, although the method used makes allowance for internal climate system variability, the extent to which confidence should be placed in the results from a single decade depends on how well they are corroborated by results from a longer period. It is therefore reassuring that, although somewhat less well constrained, the best estimates of ECS and TCR using data for 1970–2009 are closely in line with those using data for the 2000s. Note that the validity of the TCR estimate depends on the historical evolution of forcing approximating the 70-year linear ramp that the TCR definition involves. Since from the mid-twentieth century onwards greenhouse gas levels rose much faster than previously, that appears to be a reasonable approximation, particularly for changes to the 2000s.

I have modified the R-code I used for my December study so that it computes and plots PDFs for each of the five periods used in the Nature Geoscience study for estimating ECS and TCR. The resulting ECS and TCR graphs, below, are not as elegant as the confidence region graphs in the Nature Geoscience paper, but are in a more familiar form. For presentation purposes, the PDFs (but not the accompanying box-and-whisker plots) have been truncated at zero and the upper limit of the graph and then normalised to unit total probability. Obviously, these charts do not come from the Nature Geoscience paper and are not to be regarded as associated with it. Any errors in them are entirely my own.

clip_image002

image

The box-and-whisker plots near the bottom of the charts are perhaps more important than the PDF curves. The vertical whisker-end bars and box-ends show (providing they are within the plot boundaries) respectively 5–95% and 17–83% CIs – ‘very likely’ and ‘likely’ uncertainty ranges in IPCC terminology – whilst the vertical bars inside the boxes show the median (50% probability point). For ECS and TCR, whose PDFs are skewed, the median is arguably in general a better central estimate than the mode of the PDF (the location of its peak), which varies according to how skewed and badly-constrained the PDF is. The TCR PDFs (note the halved x-axis scaling), which are unaffected by ΔQ and uncertainty therein, are all better constrained than the ECS PDFs.

The Nature Geoscience ECS estimate based on the most recent data (best estimate 2.0°C, with a 5–95% CI of 1.2–3.9°C) is a little different from that per my very similar December study (best estimate 1.6°C, with a 5–95% CI of 1.0–2.9°C, rounding outwards). The (unstated) TCR estimate implicit in my study, using Equation (2), was 1.3°C, with a 5–95% range of 0.9–2.0°C, precisely in line with the Nature Geoscience paper. In the light of these comparisons, I should perhaps explain the main differences in the data and methodology used in the two studies:

1) The main difference of principle is that the Nature Geoscience study uses GCM-derived estimates of ΔF and F. Multimodel means from CMIP5 runs per Forster et al. (2013) can thus be used as a peer-reviewed source of forcings data. ΔF is accordingly based on simulations reflecting the modelled effects of RCP 4.5 scenario greenhouse gas concentrations, aerosol abundances, etc. My study instead used the RCP 4.5 forcings dataset and the F figure of 3.71°C reflected in that dataset; I adjusted the projected post-2006 solar and volcanic forcings to conform them with estimated actuals. Use of CMIP5-based forcing data results in modestly lower estimates for both ΔF and F (3.44°C for F). Since CO2 is the dominant forcing agent, and its concentration is accurately known, the value of ΔF is closely related to the value of F. The overall effect of the difference in F on the estimates of ECS and TCR is therefore small. As set out in the SI, an adjustment of +0.3 Wm−2 to 2010 forcing was made in the Nature Geoscience study in the light of recent satellite-observation constrained estimates of aerosol forcing. On the face of it, the resulting aerosol forcing is slightly more negative than that used in my December study.

2) The Nature Geoscience study derives ΔQ using the change in estimated 0–2000 m ocean heat content (OHC) – which accounts for most of the Earth system heat uptake – from the start to the end of the relevant decade (or 1970–2009), whereas I computed a linear regression slope estimate using data for all years in the period I took (2002–11). Whilst I used the NODC/NOAA OHC data, which corresponds to Levitus et al. (2012), here, for the entire 0–2000 m ocean layer, the Nature Geoscience study splits that layer between 0–700 m and 700–2000 m. It retains the NODC/NOAA Levitus OHC data for the 700–2000 m layer but uses a different dataset for 0–700 m OHC – an update from Domingues et al. (2008), here.

3) The periods used for the headline results differ slightly. I used changes from 1871–80 to 2002–11, whilst the Nature Geoscience study uses changes from 1860–79 to 2000–09. The effects are very small if the CMIP5 GCM-derived forcing estimates are used, but when employing the RCP 4.5 forcings, switching to using changes from 1860–79 to 2000–09 increases the ECS and TCR estimates by around 0.05°C.

Since the Nature Geoscience study and my December study give identical estimates of TCR, which are unaffected by ΔQ, the difference in their estimates of ECS must come primarily from use of different ΔQ figures. The difference between the ECS uncertainty ranges of the two studies likewise almost entirely reflects the different central estimates for ΔQ they use. The ECS central estimate and 5–95% uncertainty range per my December heat-balance/energy budget study were closely in line with the preferred main results estimate for ECS, allowing for additional forcing etc. uncertainties, per my recent Journal of Climate paper, of 1.6°C with a 5–95% uncertainty range of 1.0–3.0°C. That paper used a more complex method which, although less robust, avoided reliance on external estimates of aerosol forcing.

The take-home message from this study, like several other recent ones, is that the ‘very likely’ 5–95% ranges for ECS and TCR in Chapter 12 of the leaked IPCC AR5 second draft scientific report, of 1.5–6/7°C for ECS and 1–3°C for TCR, and the most likely values of near 3°C for ECS and near 1.8°C for TCR, are out of line with instrumental-period observational evidence.

===============================================================

Here’s a figure of interest from from the SI file – Anthony

Lewis_et_al_SI

Fig. S3| Sensitivity of 95th percentile of TCR to the best estimate and standard error of the change in forcing from the 2000s to the 1860-1879 reference period. The shaded contours show the 95th percentile boundary of the TCR confidence interval, the triangles show cases (black and blue) from the sensitivity Table S2, and a smaller adjustment to aerosol forcing for comparison (red).

Advertisements

  Subscribe  
newest oldest most voted
Notify of

Is that 2C heating…or cooling?

Arno Arrak

Interesting that the positive water vapor feedback that IPCC has used is no longer able to double or triple the Arrhenius warning from pure carbon dioxide alone. Most recently that Arrhenius warming was supposed to be about 1.1 degrees Celsius. They are getting really close to it which tells me that even positive water vapor feedback cannot save their predictions of dangerous greenhouse warming any more. You should know of course that according to Ferenc Miskolczi water vapor feedback is negative and completely cancels out any greenhouse warming from carbon dioxide. Read E&E 21(4):243-262 (2010).

the1pag

“The take-home message from this study, like several other recent ones, is that the ‘very likely’ 5–95% ranges for ECS and TCR in Chapter 12 of the leaked IPCC AR5 second draft scientific report, of 1.5–6/7°C for ECS and 1–3°C for TCR, and the most likely values of near 3°C for ECS and near 1.8°C for TCR, are out of line with instrumental-period observational evidence.”
What else would we expectfrom IPCC except typical exaggeration?

Bill_W

Not impressed. All I see are 14 more authors that have been bought by big something or other and are now one the denier’s list. Also this journal and its editors now need to be shunned. I think we need to redefine what peer-reviewed means. Nature Climate Change and Nature Geoscience are obviously now NOT reputable journals. Unlike NATURE which is 2nd only to the Gospel in its veracity. The libertarian-Kochtopus has obviously wrapped it’s evil tentacles of denialism-conspiracy theory around these journals, probably holding the editors families hostage. /sarc off
REPLY: Good thing you added that /sarc off -Anthony

DirkH

The fun thing is that this will not change the IPCC’s policy prescriptions one iot.

Christoph Dollis

Well that’s really interesting.
I just put this video up on Richard Dawkins Foundation for Reason and Science in the vain hope I can get Dawkins to come to reason on this issue. While that seems unlikely (and who knows if he’ll even see it, although he did retweet something I said the other day, which resulted in me getting in an an argument with Lucy Lawless a.k.a. Xena: Warrior Princess — which, granted, is not something I ever expected to happen in my life!
Anyway, this is what I posted:

Why AGW is not a likely risk — and what is.
Richard Dawkins, a fantastic and rightly-esteemed scientist in many respects, is fundamentally wrong about global warming. This is an excellent summary of why and also how this error is dangerous to the people of the world in terms of setting policy.
The entire video should be watched for the evidence to be developed, but the first 9 minutes and last 1 must be, at a minimum.

P.S. I find the word “hoax” in the title that the YouTuber added regrettable. I suspect the belief started with a sincere but flawed hypothesis. And the thing is, it isn’t *entirely* flawed. But it’s a minor and self-limiting part of a much more complex equation. The big danger facing us is not global warming, as you will see and hopefully understand.

Where this video (which is excellent and well worth watching and sharing!) is relevant is the 2 deg figure correlates very closely with the figure Geography Professor Bob Carter (over 100 papers published, plus some op-eds in major newspapers) refers to in his talk.

This Nature letter suffers from the same problem as so many other studies : It is not possible to derive ECS or TCS from observations of temperature if you don’t know what the other drivers of climate are. ie, the letter assumes the temperature is driven by CO2. If in fact something else is driving temperature, then ECs and TCS are simply unknown.

Christoph Dollis

P.S. I didn’t embed that video correctly. Anyone know where instructions to do just that can be found?
REPLY: just post the link, no tags, WordPress does it automatically. I’ll fix it for you – Anthony

Christoph Dollis

Thanks, Anthony.

Bill_W

[double post, we got the first one – Anthony]

DirkH

Christoph Dollis says:
May 19, 2013 at 2:13 pm
“Richard Dawkins, a fantastic and rightly-esteemed scientist in many respects, is fundamentally wrong about global warming. This is an excellent summary of why and also how this error is dangerous to the people of the world in terms of setting policy.”
Well, for some reason Darwin always needs a bulldog, Thomas Huxley, or a Rottweiler, Dawkins. Good luck convincing him of anything he isn’t already convinced off.

Gary Pearse

The sensitivities are lower for cooler periods and higher for warmer periods. This suggests to me that there is still some natural variability included in the estimates.

Christoph Dollis

Well, for some reason Darwin always needs a bulldog, Thomas Huxley, or a Rottweiler, Dawkins. Good luck convincing him of anything he isn’t already convinced off.

Oh, I forgot to add that on Dawkins Foundation’s FB page, you’d be surprised how many of his commenters grow at his pro “climate change” unthinking posts. A lot of them really do get it.
They love him on meme theory and selfish gene theory, and for standing up for human rights (as do I), but climate change and another topic I won’t get into at the moment lose him.

Pamela Gray

And clouds were modeled how? Why don’t we run these models with various cloud settings? Equatorial wind and SST is teleconnected with greater/lesser reflective clouds. Try it. Run the models with equatorial clouds and without clouds. You can even have your CO2 fudge factor. Send in the clouds.

Christoph Dollis

*groan

Rud Istvan

Nic Lewis–terrific. Special thanks for the clarification of differences from your previous paper.
Dare we hope from your remarks that AR5 SOD will see further substantial revision, now that the leaked version has been so roundly criticized by so many?
That would seem a litmus test for the current IPCC process, which the Climategate emails show was rigged for AR4. But much has changed since. Climategate, Climate Audit, Climate Etc., WUWT, the pause exceeding the length NASA said falsified the GCM models,…
ECS and TCR are the purest and simplest form of this litmus test for IPCC scientific integrity. Thanks to the work of you, you colleagues, and many others, we now have a clear indicator.

David, UK

As has been said MANY times, in so many words: if the effects of a little CO2 really were amplified multi-fold by water vapour in a feedback loop, we wouldn’t have lasted a blink of an eye. And yet here we are – life forms – 4 billion years on. Somehow that’s too hard to understand for some. Go figure.

Great work Nic.
I’ve been itching to talk about this for sometime especially give the crap that SkS said about your earlier work
Anthony. that is the big bottom line here. You had cook and company trashing Nic and it appears that 14 IPCC authors think differenly than the Cook and company

@Mosher I agree. Cook and Co. are advocates, so like Romm, they tend to do those sorts of things. Now, it appears Cook and Nuccitelli have reached the level of paid advocates.

Eliza

Unfortunately I have to disagree with the author because I don’t think C02 at this stage of evidence has ANY effect on atmospheric temperatures: as Prof of Physics Happer said recently with 3000ppm C02 we had an ice age 1000’s of years ago…. So by logic high C02 seems to be related if at all with very LOW mean global temperatures. Also submarines regularly have 3000 ppm atmosphere and everybody seems to be happy down there, so as far as us humans as concerned it ain’t a problem and plants love it. LOL

Manfred

Nic Lewis,
you are really leaving a footprint in climatescience !
What is your opinion about the massive increase in 0-2000m ocean heat content during the switch to the ARGO system ? Your link above does not show yearly data for that layer (only 5 years averages), but there is a graphic here (posted recently by Willis Eschenbach):
http://wattsupwiththat.files.wordpress.com/2013/05/changes-in-ocean-heat-content-noaa-layers.jpg?w=640
I would say that jump in 0-2000m data (and 0-700m as well) around 2003 is completely implausible, the increase in 3 years is about as much as in the other 37 years of the last 40 years combined. Such an increase would have required an epic decrease in cloud coverage.
Assuming the increase is mostly an error, what effect would this have on the sensitivity fall in the 2000-2009 decade ?

Great article. However, I could not find the link to get the “free” copy from Nature.

Rosco

I have conducted a simple experiment that proves the fundamental assumption that you can simply sum 2 equal radiative fluxes and use that sum to calculate a temperature using the SB equation is NOT VALID – not meaning to appear rude by using capitals.
I have proved that 2 equal sources of energy capable of heating something to 30 degrees C each – about 478.9 W/sq.metre each – are only capable of heating that same object by an extra few degrees C when both are operating at the same time.
The assumption that you can add these fluxes together to arrive at 957.8 W/sq.metre and calculate a temperature of ~360 K using the SB equation from this sum DOES NOT ACTUALLY HAPPEN IN REAL LIFE.
Anthony can contact me and I’ll submit the experiment in Microsoft word document form and anyone is welcome to try it for themselves.
I am confident I am right – not precise in actual measurements – but right in claiming 2 30 degree heat sources simply cannot combine to produce 87 degrees C as a simple sum of the radiative fluxes implies and as is used all the time in climate science !

Christoph Dollis

Unfortunately I have to disagree with the author because I don’t think C02 at this stage of evidence has ANY effect on atmospheric temperatures: as Prof of Physics Happer said recently with 3000ppm C02 we had an ice age 1000′s of years ago…. So by logic high C02 seems to be related if at all with very LOW mean global temperatures.

No, Eliza.
While he doesn’t mention this exact point, Professor Carter in the video I posted above, pointed out that there is a logarithmic curve with the greenhouse effect of increasing CO2 in the atmosphere being less and less per additional unit of increase of CO2. Please see the video starting at the 20:21 mark.
So 3000 ppm is not 10 times more effective in direct greenhouse effects than 300 ppm. CO2 can be both a greenhouse gas and overwhelmed by other temperature forcing factors.

RomanM

Well done, Nic!

Eliza

Cris Dollisto
OK, OK I totally agree with Carters thinking by the way, but empirically I even disagree with him re the earth’s atmosphere which is basically 99% water vapour. I ask you am I right or wrong?. Yes iexcessive Co2 may “attempt” to conserve heat but it does not by data and past evidence, Sorry I have become a blatant denier that C02 has ANY effect of global temperatures. BTW I totally agree with Scafetta re Solar influence and that at this moment the only changing variable is in fact SOLAR (Solar 24 is a minimum SSN on average since 1914?)

William Astley

In reply to
Mike Jonas says:
May 19, 2013 at 2:14 pm
This Nature letter suffers from the same problem as so many other studies : It is not possible to derive ECS or TCS from observations of temperature if you don’t know what the other drivers of climate are. ie, the letter assumes the temperature is driven by CO2. If in fact something else is driving temperature, then ECs and TCS are simply unknown.
I somewhat agree with you comment. This is however a very important paper. Thank-you Dr. Lewis, for the link to the paper and your enlightening comments concerning the paper.
Best wishes,
William
There is additional data and analysis that needs to be considered. There appears to be a connected set of anomalous observations that require and have a physical explanation.
The IPCC has assumed that 100% of the 20th century warming was caused by atmospheric CO2 increase. It appears that assumption is not correct.
If a significant portion of the 20th century temperature rise was due to solar magnetic cycle modulation of planetary cloud cover, the 20th portion of warming due to CO2 could be significantly less.
It is interesting that the majority of the 20th century warming was in the Arctic and high Northern latitudes which is not in agreement with the general circulation models (GCM). The GCM predicted that the majority of the warming due to atmospheric CO2 increases would be in the tropics where there is the most amount of long wave radiation emitted to space and the largest amount of water to amplify the CO2 forcing.
The regions that warmed in the 20th century, Northern Hemisphere, particularly high latitude regions in the Northern hemisphere is the same regions that the paleo climatic records shows warms and cools cyclically. The paleo climatic specialist call this cyclic Northern Hemispheric warming a Dansgaard-Oeschger cycle. There are cosmogenic isotopes changes that correlate with the D-O cycles which indicate that solar magnetic cycle changes are somehow causing the D-O cycle. The late Gerald Bond was able to track 23 D-O cycles through the current interglacial and into the last glacial cycle. The D-O cycles have a periodicity of 1450 years and also occur at 950 years and 1950 years which is 1450 years plus or minus 500 years. This matches an observed cycle of solar variability.
The solar magnetic cycle activity in the last 70 years was the highest in 8000 years and the longest period of high activity in 11,000 years.
http://cc.oulu.fi/~usoskin/personal/nature02995.pdf
Unusual activity of the Sun during recent decades compared to the previous 11,000 years by S. K. Solanki, I. G. Usoskin, B. Kromer, M. Schussler & J. Beer
As the solar magnetic cycle is currently slowing down and it appears the sun is going to enter a Maunder like minimum, the solar contribution to the 20th century warming will become known.
This is a graph, that is also located at the above site, that compares solar cycle 24 to the weakest solar magnetic cycles in the last 150 years.
http://www.solen.info/solar/images/comparison_similar_cycles.png
http://arxiv.org/abs/1009.0784v1
There is an anomalous delay of 10 to 12 years from the time the solar cycle changes to the first observed cooling when there is a step change in the solar magnetic cycle.
http://arxiv.org/abs/1112.3256
Solar activity and Svalbard temperatures
The long temperature series at Svalbard (Longyearbyen) show large variations, and a positive trend since its start in 1912. During this period solar activity has increased, as indicated by shorter solar cycles.
The temperature at Svalbard is negatively correlated with the length of the solar cycle. The strongest negative correlation is found with lags 10 to 12 years. These models show that 60 per cent of the annual and winter temperature variations are explained by solar activity. For the spring, summer and fall temperatures autocorrelations in the residuals exists, and additional variables may contribute to the variations. These models can be applied as forecasting models.
William: Latitude and longitude of Svalbard (Longyearbyen)
78.2167° N, 15.6333° E Svalbard Longyearbyen, Coordinates
We predict an annual mean temperature decrease for Svalbard of 3.5 ±2C from solar cycle 23 to solar cycle 24 (2009 to 2020) and a decrease in the winter temperature of ≈6 C.
Greenland ice temperature, last 11,000 years determined from ice core analysis, Richard Alley’s paper.
http://www.climate4you.com/images/GISP2%20TemperatureSince10700%20BP%20with%20CO2%20from%20EPICA%20DomeC.gif
http://www.essc.psu.edu/essc_web/seminars/spring2006/Mar1/Bond%20et%20al%202001.pdf
Persistent Solar Influence on North Atlantic Climate During the Holocene (William: Holocene is the name for this interglacial period)
http://www.agu.org/pubs/crossref/2003/2003GL017115.shtml
Timing of abrupt climate change: A precise clock by Stefan Rahmstorf
https://ams.confex.com/ams/pdfpapers/74103.pdf
The Sun-Climate Connection by John A. Eddy, National Solar Observatory
Solar Influence on North Atlantic Climate during the Holocene
The specific regions of the planet that warm and cool the most during the D-O cycle is the Northern hemisphere and particularly high latitudes in the northern hemisphere.
http://rivernet.ncsu.edu/courselocker/PaleoClimate/Bond%20et%20al%201999%20%20N.%20Atlantic%201-2.PDF
http://www.agu.org/pubs/crossref/2003/2003GL017115.shtml
Timing of abrupt climate change: A precise clock by Stefan Rahmstorf
Many paleoclimatic data reveal a approx. 1,500 year cyclicity of unknown origin. A crucial question is how stable and regular this cycle is. An analysis of the GISP2 ice core record from Greenland reveals that abrupt climate events appear to be paced by a 1,470-year cycle with a period that is probably stable to within a few percent; with 95% confidence the period is maintained to better than 12% over at least 23 cycles. This highly precise clock points to an origin outside the Earth system; oscillatory modes within the Earth system can be expected to be far more irregular in period.
Lindzen and Choi found by analyzing top of the atmosphere radiation Vs ocean surface temperature changes that the planet resists temperature changes (negative feedback) by an increase or decrease in planetary cloud cover in the tropics thereby reflecting more or less radiation off into space. Based on Lindzen and Choi’s results the planet resists greenhouse gas forcing changes rather than amplifies greenhouse gas forcing and the fact that there is no observed tropical tropospheric warming (the extreme greenhouse forcing theory requires that greenhouse gas forcing – any greenhouse gas – will cause there to be an increase in water vapor in the tropical troposphere at around 8 km above the surface of the planet. This increase in water vapor will amplify the CO2 forcing. There is no observed tropical tropospheric warming in the last 15 years.
http://www.johnstonanalytics.com/yahoo_site_admin/assets/docs/LindzenChoi2011.235213033.pdf
http://icecap.us/images/uploads/DOUGLASPAPER.pdf
Based on past on analysis of the paleo temperature record and past solar magnetic cycle changes and GCR changes it is estimated that 0.47C of the 20th century temperature rise is due to solar magnetic cycle changes.
http://www.eike-klima-energie.eu/uploads/media/Shaviv.pdf
“We examine the results linking cosmic ray flux (CRF) variations to global climate change. …then proceed to study various periods over which there are estimates for radiative forcing, temperature change and CRF variations relative to today. These include the Phanerozoic as a whole, the Cretaceous, the Eocene, the Last Glacial Maximum, the 20th century, as well as the 11 year cycle… Subject to the above caveats and those described in the text, the CRF/climate link therefore implies that the increased solar luminosity and reduced CRF over the previous century should have contributed a warming of 0.47 +/-0.19C, while the rest should be mainly attributed to anthropogenic causes. Without any effect of cosmic rays, the increase in solar luminosity would correspond to an increased temperature 0.16C +/-C.”
There has been 16 years in which atmospheric CO2 has risen and there is no increase in planetary temperature.
http://www.drroyspencer.com/2013/04/global-warming-slowdown-the-view-from-space/
As it appears the solar magnetic cycle is now anomalously slowing down, we will likely have an opportunity to determine by observation what portion of the 20th century warming was cause by solar magnetic cycle changes.
It appears that there is now the first observational evidence of cooling in high latitude regions of the Northern hemisphere.

Mike jarosz

This site and Global Warming Policy Foundation are my two favorite sites in large part to a free copy of Steve Goreham’s Mad,Mad,Mad World of Climatism that got me interested in climate change.The science and the economics are driving to a conclusion that will save the world from those socialist CAGW bastards. Keep up the good work and thanks.

Steve

You must consider : this ‘article’ and recent climate studies that could well further discredit the IPCC ‘agenda’ and the current US administration ( think EPA ) will have no bearing on the outlook of the general public wrt ‘global warming’ and NOTHING will change the minds of the AGW cult. The liberal media will keep these findings out of print and off the airwaves, also, those who have bought into the church of fraud will NEVER change their minds – they will hold the political position they have bought into no matter what current science or solar physics ‘finds’. Such will take their position to the grave with them.
Come to Seattle and just try and get ANY of the liberal media to report these findings, and then try and convince the state governor, mayor of Seattle, city council, county council or university types to even listen to such findings. The democrats laugh here, knowing they have sold and delivered the goods on climate and the ship of fools will sail on.
So, this article really accomplishes nothing. Interesting and clarifying science it may be, but let me hear you argue how this will ‘change’ anything. Do you see the current president rushing to hold a press conference to announce a complete make over of the EPA where he states that the above study has forced him to abandon his administration’s previous position on ‘global warming’ ?
Do you ?

pochas

Its less worse than we thought. Maybe a lot less worse.

Nic Lewis

Steven Mosher, Roman M
Thanks!
Gary Pearse
“The sensitivities are lower for cooler periods and higher for warmer periods. This suggests to me that there is still some natural variability included in the estimates.”
Yes, but the uncertainty ranges allow for natural internal variability. Results from the 1970s are most affected, because the changes in forcing and temperature (from 1860-79) involved were smallest for that (cold) decade. And the 1990s were probably more affected than other decades by imbalanced ENSO and volcanic activity. The 2000s had little volcanic activity and a reasonable ENSO balance, as well as the highest forcing and temperature changes. And results from the 2000s and from the long 1970-2009 period, which should be much less affected by natural internal variability, are quite closely in line.

Birdieshooter

The reasons to be a skeptic appear to be building up faster and faster with each passing week. It is hard to not pay attention to these studies and papers in recent weeks.

Nic Lewis

Manfred
“Assuming the increase [in ocean heat content around 2003] is mostly an error, what effect would this have on the sensitivity fall in the 2000-2009 decade ?”
I don’t have a simple answer to that, but I am planning a post about ocean heat content estimates. It is a tricky area.

This study assumes the forcings that caused the temperature change were greenhouse gas emissions, aerosols and volcanoes. Volcanoes have no effect on the long term trends. This completely ignores the overwhelming evidence that solar magnetic effects have caused at least half of the 20th century warming. The omission of any solar magnetic effects has been described as “omitted variable fraud” by Alec Rawls:
http://wattsupwiththat.com/2012/02/22/omitted-variable-fraud-vast-evidence-for-solar-climate-driver-rates-one-oblique-sentence-in-ar5/
The helio-magnetic field strength has increased by a factor of 9 from 1895 to 1991.
http://www.friendsofscience.org/assets/documents/FOS%20Essay/Rao_CR_HMF.jpg
A paper by Nir Shaviv 2005 titled “On Climate Response To Changes In The Cosmic Ray Flux And Radiative Budget” was rated “Endorsement Level: 2 by the Cook et al survey and is included in the alleged 97% of papers supporting the AGW alarmism. The abstract says “the [Cosmic Ray Flux] CRF/climate link therefore implies that the increased solar luminosity and reduced CRF over the previous century should have contributed a warming of 0.47 ± 0.19K,…” This is 60% of the 0.8 C 20th century warming. [I think CRF and solar luminosity are only part of the total solar effects.]
Total solar effects are about 7 times greater than what can be explained by changes in TSI.
http://www.friendsofscience.org/index.php?id=425
The method of comparing greenhouse gas forcing to temperature or heat content to determine climate sensitivity should only be applied to periods when there was little solar forcing, such as from 1960 to 1975, after 2002, or before 1895. The maximum warming response of a century of increasing solar forcing is delayed about 11 years after the maximum solar magnetic forcing due to the large heat capacity of the oceans. The maximum helio-magnetic field strength in 1991 causes a maximum temperature response in 2002. There has been global cooling since 2002 despite increasing CO2 emissions.

nvw

Nice job Nic with the new paper.
The BBC has an article on the paper too (http://www.bbc.co.uk/news/science-environment-22567023). Your lead author (Otto) is quoted as being largely defensive of the status quo and says that although the short-term rates of warming will be lower, over the long-term the models still produce a range of 0.9-5.0C. That is, given enough time warming rates will be back to what the IPCC earlier predicted. Questions: Why are not the long-term warming rates similarly reduced?

Steve Reynolds

Nic,
Thanks for the great work and clear explanation.
I look forward to seeing an update with data to the present.

Rosco

I am amazed.
More than an hour ago I claimed that it is invalid to simply add up radiative fluxes and calculate a blackbody temperature using the SB equation.
I claimed a simple experiment proved this and anyone can do it.
If valid it proves the greenhouse effect is insignificant and all the thought bubbles are irrelevant.
Yet everyone still continues on hypothesising without even thinking about actually trying an actual simple experiment claimed to disprove one of the fundamental propositions of the greenhouse effect.
Amazing !

Steve from Rockwood

four more authors and they could have held a soccer match.

Josh

Not that the two models are congruent, but consider for a moment, your own body. I weigh 140 lbs, or 63.5029 kg. When I have a headache or a fever, I take a couple ibuprofen. To be precise, I take 400 mg of ibuprofen. In other words, I have a body mass of 635029 g, and I need 0.4 g of ibuprofen to bring down a headache or a fever. Wolfram Alpha tells me that puts the amount of ibuprofen in my body at the time that it’s working is 0.00006299%, or 62.9 ppm.
I was in the hospital to get my spleen out, once. I was 120 pounds at the time, or 54.4311 kg. While I was there, they gave me a morphine drip that would administer 10 mg doses of morphine to me. That puts me at 544311 g and the morphine at 0.01 g, 1.837 ppm.
Again, I don’t intend to conflate the human body into a system as complex and multi-faceted as the climate of the world, but I think it’s important to have some perspective with regard to just how little of a substance is really necessary to make a large impact.

William Astley says:
May 19, 2013 at 4:10 pm
The solar magnetic cycle activity in the last 70 years was the highest in 8000 years and the longest period of high activity in 11,000 years.
You didn’t learn a thing. So, here goes again:
http://www.leif.org/EOS/muscheler05nat_nature04045.pdf :
“our extended analysis of the radiocarbon record reveals several periods during past centuries in which the strength of the magnetic field in the solar wind was similar to, or even higher than, that of today. …
Solanki et al. combine radiocarbon (14C) data, visually observed sunspot numbers and models to extend the historical sunspot record over the Holocene. They exclude the most recent 100 years of the 14C record, which are influenced by 14C-depleted fossil-fuel emissions and atomic-bomb tests conducted since AD 1950. …
irrespective of the data set applied, the recent solar activity is not exceptionally high (Fig. 2). The 14C results are broadly consistent with earlier reconstructions based on 10Be data from the South Pole, which show that production rates around AD 1780 and in the twelfth century were comparable to those observed today. …
our reconstruction indicates that solar activity around AD 1150 and 1600 and in the late eighteenth century was probably comparable to the recent satellite-based observations. In any case, as noted by Solanki et al., solar activity reconstructions tell us that only a minor fraction of the recent global warming can be explained by the variable Sun.”
http://www.leif.org/EOS/2009GL038004-Berggren.pdf :
“A comparison with sunspot and neutron records confirms that ice core 10Be reflects solar Schwabe cycle variations, and continued 10Be variability suggests cyclic solar activity throughout the Maunder and Spoerer grand solar activity minima. Recent 10Be values are low; however, they do not indicate unusually high recent solar activity compared to the last 600 years. …
Periodicity in 10Be during the Maunder minimum reconfirms that the solar dynamo retains cyclic behavior even during grand solar minima. We observe that although recent 10Be flux in NGRIP is low, there is no indication of unusually high recent solar activity in relation to other parts of the investigated period.”
As I said the whole question has recently been re-examined by a panel of experts at a workshop dedicated to this problem: Leif Svalgaard, Mike Lockwood, Jürg Beer, Andre Balogh, Paul Charbonneau, Ed Cliver, Nancy Crooker, Marc DeRosa, Ken McCracken, Matt Owens, Pete Riley, George Siscoe, Sami Solanki, Friedhelm Steinhilber, Ilya Usoskin, and Yi-Ming Wang. The conclusion is that recent solar activity was not exceptionally high. Around 1780, activity seems to have been even higher than today.
Since the Sun has not behaved in a way compatible with your other references, they are now moot and irrelevant. I think I have pointed all this out several times, but you have a hard time coming to grips with reality.
One more: http://www.leif.org/EOS/muscheler07qsr.pdf :
“The solar modulation maximum around 1780 AD indicated by the 14C and 10Be data was on the level of the second part of the 20th century or even higher. …
“The cosmogenic radio-nuclide records indicate that the current solar activity is relatively high compared to the period [just] before 1950 AD. However, as the mean value during the last 55 yr was reached or exceeded several times during the past 1000 yr the current level of solar activity can be regarded as relatively common”
So, it is time to bury the wrong notion of recent exceptionally high solar activity. This is, of course, difficult to do because once people have locked on to ‘findings’ that confirm their agenda and beliefs, they get stuck on the wrong science and can’t give it up. You are a good example of someone afflicted with that syndrome.

DocMartyn

Well done Nic. Keep chipping away.

Nic Lewis

nvw
Thanks!
You refer to the BBC article and ask, in the light of the quote by the lead author (Otto) “Why are not the long-term warming rates similarly reduced?”
You will have to address that question to Alex Otto. Maybe he was getting confused. Maybe the BBC misquoted him.
I stand by what I say in my concluding paragraph. If instead of a 5-95% uncertainty range one takes the 17-83% (IPCC ‘likely’) range of 2-4.5 deg. C for equilibrium (long-term) climate sensitivity, per both AR4 and the leaked AR5 second draft, then the correct comparison is with 17-83% uncertainty ranges from the Nature Geosciene study. On my own calculations, those are 1.5-2.8 C using data for 2000-09 (arguably the most reliable, and best constrained, estimate), or 1.3-3.0 K using data for 1970-2009 (less well constrained due to lower average forcing, but a longer period). To my mind, either of those ranges represents a sizeable downwards movement from 2.0-4.5 K.

Justthinkin

nvw asks “Questions: Why are not the long-term warming rates similarly reduced?”
Simple. By the time these long-term “whatevers” are going to start appearing,you,I, and our grandkids will be nothing but dust. I predict that by 2150, the buck will not be worth a duck. Prove me wrong.

Ken Gregory says:
May 19, 2013 at 4:45 pm
The helio-magnetic field strength has increased by a factor of 9 from 1895 to 1991.
No, it has not: see figure 10 of http://www.leif.org/research/2009JA015069.pdf

Jimbo

I have been at the Guardian and I see that their article does not appear in their Climate Change page. I must have missed something but I did click refresh. Does anyone know what I might be doing wrong?
http://www.guardian.co.uk/science/scienceofclimatechange
http://www.guardian.co.uk/environment/2013/may/19/climate-change-meltdown-unlikely-research

William Astley

lsvalgaard says:
May 19, 2013 at 5:22 pm
William Astley says:
May 19, 2013 at 4:10 pm
The solar magnetic cycle activity in the last 70 years was the highest in 8000 years and the longest period of high activity in 11,000 years.
You didn’t learn a thing. So, here goes again:
You confusing what you are repeating with what happened physically in the past and what is happening currently.
The 2007 paper you quote is not correct. (The 2007 paper asserts that the 1999, 2001, and 2004 papers are not correct. The analysis use Be10 analysis in Antarctic to challenge the Greenland Ice sheet analysis and ignores the fact the ocean sediment analysis and dendrochronological analysis supports the 1999, 2001, and 2004 papers. It is sad it is possible to find reviewers who will support warmist propaganda.)
Quite obviously, it was easy in 2007 to get a paper published that asserts solar activity in the 20th century was not the highest in 8000 years. (There is a cottage industry to try adjust observations and theory to push the extreme AGW theory.) In 2013, following 16 years of no planetary temperature rise there is an anomaly that requires explanation. There have been a series of papers published each with a different hypothesis to try to explain why there is a lack of warming for 16 years.
I can explain why the paper you quote is incorrect, however I see no point. I cannot change your mind. Others in the forum will not understand the detail issues.
I do not need however to defend my assertion using theoretical arguments.
I am asserting and quite certain, based on my understanding of the mechanisms that the same regions that warmed during the 20th century will now cool due to the abrupt slow down of the solar magnetic cycle.
The cooling of the same regions will unequivocally prove the assertion that the majority of the 20th century warming has caused by solar magnetic cycle changes rather than the increase in atmospheric CO2 is correct.
You also seem obvious to the other observations and logical arguments that support the assertion that I supporting. (i.e. The warming and cooling occurs cyclically in the paleoclimatic record. The same regions that warmed in during a D-O cycle are the same regions that warmed during the 20th century. If the 20th century warming has primarily caused by the increase in CO2 the majority of the warming would have occurred in the tropics not in high latitude regions of the Northern Hemisphere and so on.)
As the solar magnetic cycle is currently slowing down and it appears the sun is going to enter a Maunder like minimum, the solar contribution to the 20th century warming will become known.
This is a graph, that is also located at the above site, that compares solar cycle 24 to the weakest solar magnetic cycles in the last 150 years.
http://www.solen.info/solar/images/comparison_similar_cycles.png
http://arxiv.org/abs/1009.0784v1
There is an anomalous delay of 10 to 12 years from the time the solar cycle changes to the first observed cooling when there is a step change in the solar magnetic cycle.
http://arxiv.org/abs/1112.3256
Solar activity and Svalbard temperatures
The long temperature series at Svalbard (Longyearbyen) show large variations, and a positive trend since its start in 1912. During this period solar activity has increased, as indicated by shorter solar cycles.
The temperature at Svalbard is negatively correlated with the length of the solar cycle. The strongest negative correlation is found with lags 10 to 12 years. These models show that 60 per cent of the annual and winter temperature variations are explained by solar activity. For the spring, summer and fall temperatures autocorrelations in the residuals exists, and additional variables may contribute to the variations. These models can be applied as forecasting models.
William: Latitude and longitude of Svalbard (Longyearbyen)
78.2167° N, 15.6333° E Svalbard Longyearbyen, Coordinates
We predict an annual mean temperature decrease for Svalbard of 3.5 ±2C from solar cycle 23 to solar cycle 24 (2009 to 2020) and a decrease in the winter temperature of ≈6 C.
Greenland ice temperature, last 11,000 years determined from ice core analysis, Richard Alley’s paper.
http://www.climate4you.com/images/GISP2%20TemperatureSince10700%20BP%20with%20CO2%20from%20EPICA%20DomeC.gif
http://www.essc.psu.edu/essc_web/seminars/spring2006/Mar1/Bond%20et%20al%202001.pdf
Persistent Solar Influence on North Atlantic Climate During the Holocene (William: Holocene is the name for this interglacial period)
http://www.agu.org/pubs/crossref/2003/2003GL017115.shtml
Timing of abrupt climate change: A precise clock by Stefan Rahmstorf
https://ams.confex.com/ams/pdfpapers/74103.pdf
The Sun-Climate Connection by John A. Eddy, National Solar Observatory
Solar Influence on North Atlantic Climate during the Holocene
The specific regions of the planet that warm and cool the most during the D-O cycle is the Northern hemisphere and particularly high latitudes in the northern hemisphere.
http://rivernet.ncsu.edu/courselocker/PaleoClimate/Bond%20et%20al%201999%20%20N.%20Atlantic%201-2.PDF
http://www.agu.org/pubs/crossref/2003/2003GL017115.shtml
Timing of abrupt climate change: A precise clock by Stefan Rahmstorf
Many paleoclimatic data reveal a approx. 1,500 year cyclicity of unknown origin. A crucial question is how stable and regular this cycle is. An analysis of the GISP2 ice core record from Greenland reveals that abrupt climate events appear to be paced by a 1,470-year cycle with a period that is probably stable to within a few percent; with 95% confidence the period is maintained to better than 12% over at least 23 cycles. This highly precise clock points to an origin outside the Earth system; oscillatory modes within the Earth system can be expected to be far more irregular in period.
Lindzen and Choi found by analyzing top of the atmosphere radiation Vs ocean surface temperature changes that the planet resists temperature changes (negative feedback) by an increase or decrease in planetary cloud cover in the tropics thereby reflecting more or less radiation off into space. Based on Lindzen and Choi’s results the planet resists greenhouse gas forcing changes rather than amplifies greenhouse gas forcing and the fact that there is no observed tropical tropospheric warming (the extreme greenhouse forcing theory requires that greenhouse gas forcing – any greenhouse gas – will cause there to be an increase in water vapor in the tropical troposphere at around 8 km above the surface of the planet. This increase in water vapor will amplify the CO2 forcing. There is no observed tropical tropospheric warming in the last 16 years.
http://www.johnstonanalytics.com/yahoo_site_admin/assets/docs/LindzenChoi2011.235213033.pdf
http://icecap.us/images/uploads/DOUGLASPAPER.pdf
Based on past on analysis of the paleo temperature record and past solar magnetic cycle changes and GCR changes it is estimated that 0.47C of the 20th century temperature rise is due to solar magnetic cycle changes.
http://www.eike-klima-energie.eu/uploads/media/Shaviv.pdf
“We examine the results linking cosmic ray flux (CRF) variations to global climate change. …then proceed to study various periods over which there are estimates for radiative forcing, temperature change and CRF variations relative to today. These include the Phanerozoic as a whole, the Cretaceous, the Eocene, the Last Glacial Maximum, the 20th century, as well as the 11 year cycle… Subject to the above caveats and those described in the text, the CRF/climate link therefore implies that the increased solar luminosity and reduced CRF over the previous century should have contributed a warming of 0.47 +/-0.19C, while the rest should be mainly attributed to anthropogenic causes. Without any effect of cosmic rays, the increase in solar luminosity would correspond to an increased temperature 0.16C +/-C.”
There has been 16 years in which atmospheric CO2 has risen and there is no increase in planetary temperature.
http://www.drroyspencer.com/2013/04/global-warming-slowdown-the-view-from-space/
As it appears the solar magnetic cycle is now anomalously slowing down, we will likely have an opportunity to determine by observation what portion of the 20th century warming was cause by solar magnetic cycle changes.

Joel Shore

Congratulations, Nic, on your work on this paper. If the central estimates for TCR and ECS from these papers turn out to be correct, it looks like we have a better chance of being able to avoid the 2 deg C of warming that most scientists agree to be roughly the major danger threshhold, assuming that we do work pretty diligently to gradually wean ourselves off of fossil fuels over the next several decades (or implement sequestration, etc.)
That is certainly a more optimistic assessment than one would get if the middle- or high-end IPCC numbers were to turn out to be correct…in which case we are pretty screwed…and would need to be much more aggressive in stopping, or even reversing, the rise in CO2.

Richard M

I’m not completely sure I understand what the 2C relates to. It sounds like that is the warming for a doubling of CO2 without any other considerations. However, as long as we burn fossil fuels and as long as 3rd world countries are burning charcoal, etc. there will be aerosols released as well. and, if I’ve understood this estimate correctly, these aerosols will continue to cancel out some of the warming. Hence, any real warming will be less that 2C. Have I got that right?
And, to top it off, if any of the warming is absorbed into the deep oceans, then that warming will also disappear. The bottom line is we still have no idea how much warming will actually occur (if any).

joeldshore

Richard M says:

However, as long as we burn fossil fuels and as long as 3rd world countries are burning charcoal, etc. there will be aerosols released as well. and, if I’ve understood this estimate correctly, these aerosols will continue to cancel out some of the warming. Hence, any real warming will be less that 2C.

There are also other greenhouse gases, like CH4, which contribute warming…So, I think, roughly speaking, the effect of the aerosols and of the non-CO2 greenhouse gases may about cancel. (Aerosols and CH4 also have a shorter perturbation time in the atmosphere…whereas a perturbation in CO2 levels lasts a long time.)

And, to top it off, if any of the warming is absorbed into the deep oceans, then that warming will also disappear.

Warming doesn’t “disappear” by going into the deep oceans. It might get delayed…i.e., it will take longer to warm if more goes into the deep oceans than we expect and will warm more rapidly if less goes into the deep oceans than we expect. But, the ECS will not depend on this rate of transfer to the deep oceans. [What might be true is that certain diagnostic methods for trying to estimate the ECS might be off if the transfers to the deep ocean are either less or more than expected…but that’s another kettle of fish.]

Josh says:
May 19, 2013 at 5:22 pm
Not that the two models are congruent, but consider for a moment, your own body. I weigh 140 lbs, or 63.5029 kg. When I have a headache or a fever, I take a couple ibuprofen. To be precise, I take 400 mg of ibuprofen. In other words, I have a body mass of 635029 g, and I need 0.4 g of ibuprofen to bring down a headache or a fever. Wolfram Alpha tells me that puts the amount of ibuprofen in my body at the time that it’s working is 0.00006299%, or 62.9 ppm.
I was in the hospital to get my spleen out, once. I was 120 pounds at the time, or 54.4311 kg. While I was there, they gave me a morphine drip that would administer 10 mg doses of morphine to me. That puts me at 544311 g and the morphine at 0.01 g, 1.837 ppm.
Again, I don’t intend to conflate the human body into a system as complex and multi-faceted as the climate of the world, but I think it’s important to have some perspective with regard to just how little of a substance is really necessary to make a large impact.

But CO2 isn’t “working” chemically the way drugs are. it isn’t even a catalyst. It’s just a physical blocker–and one whose effect decreases logarithmically.