Ferdinand

I ask again (sorry to be nitpicking on that point): Do we agree that GCM do not, cannot by construction and therefore also shouldnt get "implemented" any sun/cloud or whatsoever cycles and correlations. I actually think that you know basically how GCMs work and so I am wondering why you writing again:

"Parametrisation" (in the case of GCMs) is a numeric description of (more or less) known physical processes, often an ensemble description of small scale processes that couldnt be described in a direct approach. For example a typical convective cloud takes place on a spatial scale of typically some km well below the GCM grid scale. However the overall (ensemble) effect of such clouds can be described by upscaling approaches (for the distribution of latent heat/upwelling/subsidence and so on). Anyhow, I just wanted to make sure that we agree (yes/no would be appreciated) on this and that when you write "cycles should be included" you mean in fact "results should be compared against".

Furthermore (again a little repetitive) if you say we have good physical reasons to assume that the sun is undergoing a certain cyclicity (50y,200y whatsoever) from the GCM point of view this a forcing problem. And of course you have to come up with a good estimate for the importance of these forcings.

Some further remarks to your post:

Here I might add a very subjective comment since I am now quite a bit in the business. If I compare state-of-the-art models and in particular their performance concerning clouds with the models when I started I am really euphoric. Simulations seem to me of an incredible quality given the complexity of the task. Anyhow in this case we are discussing if the glass is 1/10th empty or 9/10th full.

There are some imprecisions in what you say. Yes, there are differences in anthropogenic and natural sources just by timing/location and so on. Due to many nonlinear effects (interaction with dust, transport, effects on particle size,distribution, chemical composition etc) the deconvolution between natural and non-natural sources can only be done in models. It's not just residence time, but also the net effect of optical thickness and finally radiative forcing.

RF of aerosols (nitrate & mineral & organic carbon & sulfate & black carbon & biomass burning) is estimated by models and observations. Overall effect by models -0.4 ± 0.3W/m2 and by three obs studies -0.55±0.3W/m2. Sulfate aerosols were decreasing all the time in Europe (from 18 to 4 Tg/y in the last 20 years) and the US (10 to 7), is however increasing in Asia.

Non-linear and indirect aerosol effects (what I said above is only the direct radiative impact) are global in nature as shown by satellite data. These studies demonstrated also a remarkably different droplet size between polluted and non-polluted areas. The global impact of the indirect effect as implemented in GCMs is allways negative and varies between -0.3 to -1.8W/m2 with a mean of -0.8W/m2. In summary there is nothing shocking about the 1W/m2 you were mentioning and it's based on dozends of model and observational studies.

As I said above, I have the impression you didnt understand the different nature of for example the indirect aerosol effect (on droplet size spectra) and a possible relation between TSR and clouds or whatsoever. If you are interested I can try to explain this more. By the by, there are of course many studies looking on the effect of aerosols on cloud droplets and such.

Saludos

Georg

Georg,

As far as I know how GCM's work (will be in Oxford 23rd June to learn more about them...), they should react on changes in forcing, like any cycle in solar forcing. Indeed if they don't resemble the real world results (as is the case for clouds in the tropics and in the Arctic), something is lacking in the equations and/or something is wrong with the parametrisations. That means that the GCM's must be adjusted to that (by including a -non- linear factor, or by changing the parametrisation accordingly), in such a way that the results of the GCM are in line with the observations.

The main problem with several models is that they use quite similar feedbacks for the different forcings. Which is not so straight-forward for solar and volcanic vs. GHGs and tropospheric aerosols. The influence of solar/volcanic changes is largely in the stratosphere, which influences climate (temp, clouds, precipitation) in a different way than GHGs and aerosols. The Hadcm2/3 model e.g. probably underestimates the influence of solar changes on temperature with a factor 2-3 (see part 2c, page 7 or 4085 of Stott ea.. And that is within the constraints of the model (like a fixed minimum influence of aerosols...).

Further, some models spontaneous generate longer time (over 30 year) AMO-like cycles, others don't. In fact, ocean (and coupled ocean-atmospheric) models should generate internal climate cycles like ENSO, even with randomly shifting frequency around a central frequency, if they accurately describe oceanic climate... But I haven't seen many which does that accurately.

Yes, the cycles of the forcings should be included. But as already proven in the case of clouds, the results of the parametrisation doesn't fit the observations... And the observed internal cycles (ENSO, AMO,...) should be seen more accurately (in amplitude, less in frequency) in the model results (not "implemented" - that indeed is wrong for internal cycles).

That was done - within the constraints of the model - for the Hadcm3 model, see Stott ea. beforementioned. But I have the impression that they didn't adjust their model accordingly...

I hope so for your model work, as the 2002 results were more 50-50 (as much negative change in heat balance/m2 in the tropics as the global increase by GHGs since the beginning of the industrial revolution!)...

I was talking about only sulfate aerosols, as that are the most important coolers (according to the models!). The global SO2 emissions didn't change much since 1975 (according to the IPCC inventory, must retreeve the URL), with a strong decrease in Europe and North America and a strong increase in SE Asia.

As the main increase in emissions of all types of anthropogenic aerosols is in SE Asia, in the NH (sub)tropics, this should lead to:

Higher albedo (primary and other effects combined) in the tropics: the opposite is observed in total skies (with no observed change in clear skies). More long-lasting clouds in the tropics: the opposite is observed. Less warming in the NH (oceans) vs. the SH: the opposite is observed...

Any reactions?

Ferdinand

Hello Ferdinand

astonishingly we cant agree even on such very fundamental things. Please use your time in Oxford to clarify this with people most probably more educational than I am. A logic that can be introduced into a GCM would be "we have observed/inferred that UV radiation does this or that to cloud droplets. Therefore there should be a modified parametrisation in the cloud scheme"(that's just an example). In the following it might turn out that this process explains that the much higher UV variations of the sun then can have a longterm effect that somehow fits to observation you might consider robust and important (x-year cycle of sun variations). However it does NOT work like this:" Oh there is a cycle of x years found in some observations, so lets introduce a fudge factor which acts on the time scale x on whatever process in the modell. This would be at bets a caricatural imitation of how some sceptics think that GCMs work. That is as clear as I can put it.

The paper states a general underestimation to small forcings. Even with a factor of 3 for solar you cant explain the warming of the last 50 years. Still it stands that an estimated forcing of about 0.1W/m2 must be compared with the 0.5-0.9 by greenhouse gase.s and the contribution of solar varies between 16% to 36% according to one model configuration and depending on the used solar reconstruction. Dont cite selectively.

In my eyes it's rather a strength than a weakness of the models that they feature different behaviour (different periodicities, cloud feedbacks and so on). Stochastic forcing of the ocean by atmospheric variability acts differently and so excites different oscillations. The fact that all these models show comparably similar feedback to GHGs underlines the robustness of climate predictions.

Sulfate aerosols are just one type of aerosols. I gave you the actual estimation for all aerosols. Their interaction is non-linear and dependent on the location and timing of the emission. The global estimates justify perfectly the 1W/m2 you were doubting. Because there are rising emissions in SE Asia there must not necessarily higher albedo values in the tropics due to complex circulation and cloud feedbacks. In particular your inference on NH vs SH SSTs seems a gross oversimplification. The aerosol forcing on the satellite observed period is very small.

Have fun in Oxford 

Georg

Georg, of course, I will ask such questions in Oxford, but (as a retired engineer in process automation), if a model (any model) doesn't fit reality in important items (in this case an exact response of tropic cloud cover to changes in TSI), then simply discard the model, until you know more about the exact mechanism (and the exact results). Or you will find yourself in deep trouble (own experience!) if you try to predict the future by such a model... As in this case the change in insolation (in the tropics) is an order of magnitude larger than the change in forcing by GHGs in the same period, one need to reconsider the basics of the model...

I never said that GHGs have no influence, only that in most(/all) models solar influences are underestimated, and that the influence of aerosols is largely overestimated. That means that, to fit the past temperatures, one need to reduce the influence of GHGs too. Not in basic forcing, but the difference is in the feedbacks, and mainly in the cloud feedback... If the influence of TSI changes doubles by cloud cover changes (that estimate is based on what is seen in cloud cover changes over the past two sun cycles), that means that the influence of solar in the first halve of the previous century is more near 100% (instead of 50%) and 30-70% in the second halve, which means that the real increase in temperature from a CO2 doubling is more to the low side (around 1.5 K) of the IPCC range...

Or they have all the same (logical) errors...

The increase in ocean heat content (not only SST) in all oceans is (about twice) higher in the NH than in the SH, if corrected for surface. There is little ocean, air and (especially) aerosols exchange between NH and SH, anyway not enough to explain the large difference. 90% of all human induced aerosols are emitted in the NH. Thus from the 1 W/m2 you refer too, 0.9 W/m2 is the average cooling by aerosols in the NH, 0.1 W/m2 in the SH. But the NH is warming faster?

Thanks, will travel in England-Wales-Ireland-Wales-England from 3rd June until 21st June and then in Oxford for a few days...

Moreover, from the study of two ice cores on Svalbard:

Have a look at the Jan Mayen temperarure record, which indicates that the 1930-1960 period was near as warm as today...

Ferdinand

I am very disappointed by the restrictive policy of RC for any comment that sheds even the slightest doubts on the HS graph.  In this case, the discussion was about the positive feedback of CO2 levels caused by climate changes.

To start with, the original comment:

Further I wonder how Sheffer ea. could derive anything from the miniscule changes in CO2 levels in one century of the LIA. The variation in temperature over the past millennium was between 0.2 K (MBH98/99) and 1.0 K (Huang, bore holes), which means a CO2 variation of 1.6-8 ppmv. The effect of this on temperature is unmeasurable... (BTW, the variation in CO2 between MWP and LIA was near 9 ppmv in the Taylor Dome ice core).

There was a response to this from Mike Mann:

The numbers you cite for "the variation in temperature over the past millennium" are nonsensical (the range of total variation only differs by about 0.4C for the full range of reconstructions, such as shown in this previous post). For actual quantitative comparisons between observed pre-industrial CO2 variations and what is expected given estimated temperature variations and their uncertainties, see the Gerber et al (2003) paper cited above. The comparisons are consistent with 2xCO2 sensitivities within the conventionally cited range of 1.5 to 4.5, but not much higher. --mike

I have had a look at the Gerber ea. paper, and that triggered my comments following here:

Mike,
I suppose that there is some misunderstanding here, the graph of the different reconstructions over the past millennium at Wikipedia gives a good summary of the variations in reconstructed temperature between the MWP and the LIA. This is 0.2 K for MBH99 and 0.8 K for Esper, Moberg and Huang (-1 K vs. year ~2000), with other reconstructions somewhere in between. I interpretated the Huang amplitude too large, as the year 2000 temperature is somewhat 0.2 K above the Wikipedia zero line.
The Gerber ea. model test for CO2 changes caused by temperature changes is very interesting. But I have a few remarks:
- I may have missed it, but I didn't see any reference to oceanic photosynthesis (which is a huge player in the carbon balance).
- Since the publication, more reconstructions were published which show a broader MWP-LIA variability than previous ones.
- The model underestimates the 20th century warming. According to Knutti ea., this may be due to an overestimation of the influence of aerosols combined with the (relative) low standard sensitivity (2.5 K for 2xCO2) of the model. An alternative explanation is that the standard model underestimates solar sensitivity (or forcing, or both).
- All runs with low solar forcing and different climate sensitivities are at the very low side of (or underestimate) measured CO2 differences. Only the standard and high climate sensitivity combined with the upper range solar forcing (2.6x minimum) are quite on average CO2 variability. This too points to either a high end solar forcing or a high sensitivity for solar (or both).
- I know that Raypierre will disagree, but a high climate sensitivity for solar doesn't imply a similar sensitivity for GHGs. See the comment by Esper and many others in Quaternary Science Reviews (quoted in RC)...
- The model results imply that for the given CO2 variability, the most likely pre-industrial temperature variability is at the high end of the reconstructions.
- As an aside, CO2 variations in the past millennium are much larger in stomata data than in ice cores (in part as ice cores smooth out fast changes), but that is another discussion...

http://www.realclimate.org/wp-comments-popup.php?p=311&c=1#comment-14014

just wanted to ask you what the "nonsense" was.

my question about the Eemian was considered a "good question"

http://www.realclimate.org/wp-comments-popup.php?p=311&c=1#comment-13975

Hello Hans,

But that response was from David Archer... and didn't remotely touch the HS...

Cheers,

Ferdinand

Hello Ferdinand

I dont know why your comments hasnt been posted. The form is fine and, at least if you havent send these or similar comments already several times before, I would have led them through. I only have some comments on what I disagree (some of your points I would call wrong, but from there to nonsensical, hmm).

Only recently global ocean carbon models have been equipped with a dynamic description of phytoplankton. Classically they use a formulation based on the so called Redfield Ratio which regulates the consumption of phosphate , nitrate and CO2. The formed organic material is then exported to lower layer (the famous "export rate") and remineralized. To make it short, photosynthesis is included in an implicit way (computing net effects) in ocean models such as the Bern model.

Exactly, and so you can judge yourself from the results (the simulated CO2 variations) which of the variations are the most probable from such a carbon model's point of view.

A LIA to MWP amplitude of much more than 1C can effectively be excluded since 1C would correspond already to an amplitude of > 12ppm compared to the about 10ppm amplitude observed in icecores. The study offers CO2 constrains to climate reconstructions.

The model (as used here) does mainly test how much the atmospheric carbon is changing if you impose a certain temperature (or climate) change. So your question above is not really relevant.

Here is the summary from Gerber et al:

Such rapid and strong changes (>20ppm in 50years) cant be explained by today's carbon models. It's still very difficult to bring the stomata data in agreement with our understanding of the carbon cycle.

Georg

Hi Georg,

This was the first time I commented on Gerber, didn't know of the paper. I have some suspicion that my comment was seen indirectly as a critique on the Mann's HS, as the variability of the MBH99 reconstruction seems to be too low in the carbon model... See also the comment of M. Mann on a previous comment, where he called the variation of 0.8 K in the "spagetti graph", "nonsensical", as it is only 0.4 K (in average, but that is not the point of discussion!), according to him. But as one has a look at the graph in detail, the variation in different reconstructions is indeed 0.2-0.8 K...

Agreed, that only excludes the 5xSolar, but the study also should exclude the lower border CO2 results. For some reason they include a four sigma variation, which allows the inclusion of low-end results! That is the main problem with this study. They allow for inclusion of the MBH99 HS, but the result of all runs of the model with low-medium-high climate sensitivity and low solar forcing give a CO2 variation which is 2.5-4.1-6.8 ppmv. That is (too) low, but I don't know what the (more normal) 2 sigma variation of the measurements is (in addition, the Taylor dome has a MWP-LIA change of 9 ppmv, this should reduce sigma further...). Either the model has a too low climate sensitivity (but even the highest sensitivity gives low results...) or the model underestimates solar. And as you can see, 2.6xSolar gives results (5.8-8.0-10.6 ppmv) which are right on target for the standard to high climate sensitivity of the model... With high solar, the same model probably will give better results for the 20th century too.

Don't agree here. If the standard model doesn't fit the past century, something is underestimated, see previous comment...

The rapid changes are only in part from the biogenic response, most is from rapid changes in SST. There is some indication for that in certain periods of the MWP-LIA transition, where European stomata response is very fast to North Atlantic SST changes (and in current global CO2 - and plant growth - variations during ENSO events). And ice cores have a long (~60 years for Law Dome, which has the highest accumulation) smoothing time...

Ferdinand

Hi Ferdinand

quoting becomes more and more complicate. Hope this is still readable.

The YD was associated with about 10-15ppm in the Monnin et al paper (Science,2001) on Dome  C. If stomata data are right it means that an event of similar importance has been taken place completely undetected in  the last 1000 years over a period of less than 60 years? Is this what you propose?

Regards Georg