Roger Andrews: Global Warming Projections Using Solar Cycles

Posted: May 23, 2011 by tchannon in climate, Ocean dynamics, Solar physics

Guest post by Roger Andrews

Figure 8 shows the net solar and anthropogenic contributions to SAT

In a previous post (https://tallbloke.wordpress.com/2011/02/21/tallbloke-and-tim-channon-a-cycles-analysis-approach-to-predicting-solar-activity/) Tim Channon replicated the Lean TSI reconstruction very closely using a solar cycle-based analysis. Tim’s results are shown below:

In previous attempts to estimate how much global warming was solar and how much anthropogenic I used TSI, i.e. the sum of the amplitudes of all the solar cycles, to define solar forcing, with the implicit assumption being that all the solar cycles impact temperature in direct proportion to their amplitude. But now I’m going to assume that for reasons as yet unexplained some cycles have a larger proportionate impact than others.

So here I will attempt to match the global sea surface (SST) and surface air temperature (SAT) records using a phenomenological model that applies different forcings to each solar cycle. The cycles I used were Tim’s 426.5, 238.9, 112.5, 78.9 and 57.2 year cycles. The shorter 11-year cycles were ignored.

Figure 1 shows the results for ICOADS SST. I can fit SST very closely (R = 0.97 for annual SST data and 0.99 for smoothed) using solar cycles alone.

Figure 1

Figure 2 shows the individual solar cycle contributions. The 238.9 and 112.5 year cycles generate peak-trough temperature changes of 0.6C, the 78.9 year cycle a change of 0.4C and the 57.2 year cycle a change of 0.3C. The 426.5 year cycle has no detectable impact (it hardly varies over this time interval anyway).

Figure 2

Now I’m going to see if I can match the GISSmet surface air temperature series using solar cycles alone. The answer is no, I can’t. I have to include an anthropogenic component. But once I include it I get another good match (R= 0.91 for annual SAT means and 0.98 for smoothed), as shown in Figure 3:

Figure 3

Figure 4 shows the temperature impacts of all SAT forcings. The 238.9 and 78.9 year solar cycles now have the largest impacts and again there is no detectable contribution from the 426.5 year cycle. The anthropogenic forcings are from the GISS tabulation (http://data.giss.nasa.gov/modelforce/RadF.txt), and the factor applied to them equates to a climate sensitivity of 0.8C.

Figure 4

Figure 5 compares anthropogenic warming with the sum of the solar cycle warming. According to these results solar has caused most of the warming since 1880 and it has also contributed about as much warming over the second half of the 20th century as we humans, despite what the IPCC claims:

Figure 5

According to the above model results man-made greenhouse gases warm only the air; the ocean is warmed entirely by the sun. There is no significant anthropogenic warming of the ocean from back radiation or anything else.

Armed with these results we can now project what sea surface and surface air temperatures are going to do for the remainder of the 21st century. To do this for SST we need only extrapolate the solar cycles through 2100 and apply the factors used to obtain the best-fit match to observations. Figure 6 shows the results:

Figure 6

To project surface air temperatures we need an estimate of future anthropogenic as well as solar cycle forcings, and I got these from IPCC SRES A1B, a middle-of-the-road growth scenario that projects a gradual rise to 717ppm atmospheric CO2 by 2100. Applying the best-fit factors to the anthropogenic and solar forcings then gave the results shown in Figure 7:

Figure 7

Figure 8 shows the net solar and anthropogenic contributions to SAT:

Figure 8

According to these results there won’t be any significant global warming between now and 2100 – provided my analysis is realistic. Is it? Well, it’s admittedly speculative. But then, so are the climate models the IPCC uses to make its official global warming predictions. And my models fit observations a lot better too.

Comments
  1. tallbloke says:

    Hi Roger,
    Well done, this is a great post to generate some interesting discussion.

    Leaving aside my own opinions about forcings and the mechanisms they operate by and taking the post on it’s own terms:

    I’m not sure how you’ve combined solar and anthrop for the projected SAT. How is it that a predicted anthrop effect of +0.65C from 2010 to 2100 plus a predicted solar effect of -0.4C over the same period (fig 8 ) ends up as a +0.4C rise when they are combined? (fig 7)

    Shouldn’t it be nearer +0.25C?

  2. Roger Andrews says:

    TB

    Don’t have the numbers in front of me right now, but the SAT increase shown in Figure 7 looks to be around 0.3C, not 0.4C. Will check and get back with more exact data in a few hours’ time.

  3. Verity Jones says:

    I predict that, if your numbers are anyway accurate and SST does start to behave as projected in Figure 6, that we’ll see some reason being invented for reducing the proportion of SST used in the calculation of gobal average temperature 😉

  4. Roger Andrews says:

    TB

    Just got home.

    To answer you earlier question, the actual SAT numbers were: anthrop warming 0.673C, solar cooling 0.376C, total warming 0.297C.

  5. cementafriend says:

    I would suggest that GISS is not a good basis for surface temperature. I believe that FOI requests have shown that there have been upward adjustments which have no scientific basis. I suggest that you go to KNMI Climate Explorer and retrieve actual SSTs. Also have a look at the following article http://climategate.nl/wp-content/uploads/2011/02/CO2_and_climate_v7.pdf.

  6. tallbloke says:

    Thanks Roger. That’s from when to when? 2010 to 2100?

  7. From 2010 to 2030:

  8. Roger Andrews says:

    From 2011 to 2100

  9. tallbloke says:

    Cool, I see it checks out now.

    So,,, GISS data needs GHG forcing huh. There’s a surprise. 🙂
    Still, it’s a commendably low sensitivity estimate.

    Now, you haven’t used any kind of model for ocean heat retention.
    Wanna talk about that? Hmm? 8)

  10. Roger Andrews says:

    Tallbloke:

    I was afraid you might ask about that.

    The models don’t allow for ocean-air heat transfer, so as far as they are concerned any solar heat added to the ocean stays there. But they also show SST decreasing by almost 1C between now and 2040, and this represents the loss of a large amount of heat from the ocean surface. Where does it go? Well, according to the SAT model little if any of it goes up, therefore most if not all of it must go down, presumably to emerge at some unspecified ocean location at some unknown time in the future.

    Another complication is the direct relationship between solar cycle forcing and SST, which presupposes that the heat from the sun warms the ocean during “up” cycles and then cools it during “down” cycles. But where does the lost heat go? Down again, presumably.

    The models would certainly be improved if they simulated sea-air heat transfer, but I wasn’t able to come up with a good way of doing this. All I could think of was to use the SST-SAT differential as a SAT forcing, but after a little thought it occurred to me that SST-SAT isn’t a forcing, it’s an effect of another forcing (most likely the 110-year solar cycle, as discussed in one of my earlier posts) and I was already using this cycle as a forcing. So I let things be.

    Thoughts?

    On a different topic, can you think of any reason why different solar cycles should generate different forcing impacts? I can fit observations only when I assume they do.

  11. tallbloke says:

    I shall reply to this when I’ve had a walk, drunk some beer while listening to live music, and had a think about it on the way home. 🙂

  12. vukcevic says:

    On a different topic, can you think of any reason why different solar cycles should generate different forcing impacts?

    Hi Roger
    Yes, reverse of Svensmark’s theory, I mentioned idea on WUWTsome time ago.
    Arctic and solar magnetic fields have a negative correlation.
    http://www.vukcevic.talktalk.net/LFC9.htm
    Most of GCR end up in polar areas, while only strongest get through in the equatorial regions. If Svensmark is correct that CR increase cloudiness, than largest effect will be at poles. Since GMF is weakest when solar is strongest, and the Earth’s field is considerably stronger then heliospheric, than increase in cloudiness will be higher when the GMF is weaker, as it is case now, exactly opposite to Svensmark suggests.
    Increased albedo in the arctic autumn, winter and spring, as well as the summer nights has no effect (low or no insolation), but higher cloudiness will act ‘as a warming blanket’ keeping arctic warmer and shortening the sea ice formation season.
    In the summer’s daytime, albedo will reduce Arctic warming, but only in the areas which are not covered by ice or snow.
    Hence: reduced GMF- more cloud- warmer polar circle region (66-90N).
    Speculative, but possible if CR increases the cloud formation to any degree.

  13. Roger Andrews says:

    Hi Vuk:

    Thanks. Interesting theory. However, I don’t see how it explains why the 238.9 and 112.5 year solar cycles should have about twice the impact on SST, relative to the peak-trough TSI change during the cycle, that the 57.2 year cycle does. Maybe you could elaborate? 🙂

  14. vukcevic says:

    Roger
    I take solar scientists view that the measured TSI is relatively constant, I assume they do professional and reliable work.
    I do not think that long term reconstructions are reliable, since they are based on 10Be and C14.
    I am looking for alternative route, which is the solar magnetic field and possible link to the Earth’s magnetism changes. Overall solar field (22 year cycle) is not particularly strong at 1 AU, but occasional short bursts (solar flares, CNEs) amount up to 1-2% of the E’s field’s strength in the Arctic
    Arctic responds strongly (imprint in long term data is more than obvious), while Antarctica may do, but it is smoothed to such degree that becomes imperceptible in the long term geomagnetic data (have an inkling why the difference, but that is another story).
    Arctic appear to behave as a kind of ‘bar magnet’ every time you hit it with a hammer looses bit of its strength, but (in contrast) tends to recover eventually .
    For the GMF down and up effect to take place a prolong periods are required, certainly lasting longer than one cycle.
    In final analysis GMF – GCR would work in same direction as TSI but on a multidecadal (smoothed) time scale.
    Data does point to some kind of a link to the Arctic and even the global temperatures
    http://www.vukcevic.talktalk.net/LL.htm
    GCR cloud formation creating an ‘Arctic warming effect’ is possible, but even there counts are so low, that I doubt that GCR effect is sufficient to make any significant difference to the cloud coverage.
    Have to wait and see ‘consensus’ on the CLOUD experiment, but if positive, I think the ‘cloud blanket’ has more chance than the ‘cloud albedo’.
    Worth floating the idea and see how it fares.

  15. tallbloke says:

    Vuk,
    did you consider my suggestion to you in email yesterday?

    The solar modulation of GCR’s up to 4GeV is around 25%
    Earth geomag modulation of GCR’s is around 50%

    But the sun’s heliospheric magnetic field varies twice as much as Earth’s geomagnetic field, so if they vary in anticorrelation in the N.H., they may be close to balancing each other in terms of GCR modulation.

    So the Svensmark effect may be more about *where* the cloud seeding takes place, rather than *how much* cloud seeding takes place.

  16. vukcevic says:

    Hi Rog
    I had discussion with L.S on this some time ago, but I am not certain he is correct. His numbers of gmf mod of 50% are, I assume, based on dipole. Dipole in the Antarctica has no perceptible solar modulation (e.g. on the scale to distinguish Dalton min or recent ‘grand max’ from the rest). All multidecadal modulation comes from Arctic, so I am inclined to dismiss his 50-25% numbers.
    Instrument and satellite measurements (since 1970) show that GCR numbers have a ceiling and floor defined by individual cycles (no multidecadal mod), where correlation with ‘misleading 10Be’ ice cores breaks down.

    Solar mod then is about 20% of 0.3-0.6nT or at best around 0.1nT (or possibly less). (Heliospheric field at 100 AU = is about 0.3-0.6nT , where solar GCR mod takes place is estimated to be 0.3-0.6nT
    http://web.mit.edu/space/www/helio.review/axford.suess.html )
    Earth modulation occurs at the entry into magnetosphere where heliospheric and geomagnetic field pressures are balanced at level about 7nT.

    Earth’s field changed from mid-Maunder to 1750 at Arctic surface from 59 to 52 microTesla, i.e. 7 microTesla.
    http://www.vukcevic.talktalk.net/LFC9.htm
    Dipole lost about 1.5 micro Tesla in that time, so you are left with more then 5 microTesla or roughly 10% re Maunder min level.
    Going back to magnetosphere’s magnetopose where modulation takes place and field is 7nT, modulation is 10% or 0.7nT against about 0.1nT at heliopause.
    That is modulation ratio 7:1, not 2:1 as LS would have it.
    My view : Arctic modulation wins ‘hands down’
    You can do your own numbers, or ask L.S. to show his calculations and compare for errors (I am sure there must be some somwhere).

  17. vukcevic says:

    Rog, just an additional note.
    If one wants to consider global GCR modulation then:

    global GCR gmf mod = (Arctic mod + Antarctica mod) / 2
    = (0.7 + 0) / 2 = 0.35
    which would give gmf mod vs. solar mod = 3.5 :1
    which is much closer to L.S.’s numbers 2:1
    However, I am talking Arctic, where temperatures correlate closely to GMF (which is not case in Antarctica)
    http://www.vukcevic.talktalk.net/NFC1.htm
    We’ll see if CLOUD comes up tramps or not, but I doubt there GCR flux is sufficient to do anything even to polar cloudiness, let alone equatorial were albedo would matter.

  18. tallbloke says:

    Thanks Vuk, food for thought. Anyway, we’ve wandered off topic a bit here, so let’s concentrate on giving Roger more feedback on his post, and I’ll put something together for a new post later.

  19. Roger Andrews says:

    Tallbloke:

    “I shall reply to this when I’ve had a walk, drunk some beer while listening to live music, and had a think about it on the way home.”

    Are you still down at the pub?

  20. tallbloke says:

    Roger, apologies, it’s been another busy week.
    You said:
    But they also show SST decreasing by almost 1C between now and 2040, and this represents the loss of a large amount of heat from the ocean surface. Where does it go? Well, according to the SAT model little if any of it goes up, therefore most if not all of it must go down, presumably to emerge at some unspecified ocean location at some unknown time in the future.

    Another complication is the direct relationship between solar cycle forcing and SST, which presupposes that the heat from the sun warms the ocean during “up” cycles and then cools it during “down” cycles. But where does the lost heat go? Down again, presumably.

    I think the answer is that this ‘missing’ heat doesn’t arrive in the ocean in the first place. If solar activity drops as the curve predicts, then in accordance with my empirical finding that the ocean loses heat below an average SSN<40 rather then gaining heat, then what heat does arrive in the ocean is lost again (along with more heat stored while the solar activity was high) warming the atmosphere from below while the atmosphere loses heat to space.

    can you think of any reason why different solar cycles should generate different forcing impacts? I can fit observations only when I assume they do.

    Well, although TSI doesn’t change much overall, the sub-components of the overall flux, (U.V., near I.R. etc) vary a lot more. Maybe these bigger variations relate to the different cycle lengths. SInce we have a low understanding of how the solar signal is amplified at the Earth’s surface, it could be something like big variations in U.V. affecting ozone levels, plankton concentrations etc.

    Certainly an interesting aspect to keep an eye on as improved and more detailed solar data builds over the years.

  21. Roger Andrews says:

    Thanks TB:

    Any way of converting your SSN>40 to a TSI threshold? Or alternatively converting projected TSI to SSN?

  22. tallbloke says:

    I used http://woodfortrees.org to overlay SSN and TSI. I can’t remember the TSI value offhand. But in any case, it’s not as simple as that. What I found was that the values obtained by a cumulative total of the SSN departing from that equilibrium value generated a curve which relates to changes in SST. This in turn relates to ocean heat content.
    https://tallbloke.wordpress.com/2010/01/05/my-simple-solar-planetary-energy-model/

    It’s an incomplete model, in that you really need to integrate the changing outgoing longwave radiation from the top of the atmosphere as well. We only have good OHC and OLR data for the last 6 years though.
    https://tallbloke.wordpress.com/2010/12/20/working-out-where-the-energy-goes-part-2-peter-berenyi/

    To get a good match to SST, we would also need to take account of the big oceanic cycles, PDO and AMO. I just noted the fact that the curve I got matched SST reasonably well if you took off the peak of the 1940 ‘blip’ during the positive PDO phase back then, and dropped it into the ‘hole’ in SST around 1910 caused by the negative phase of the PDO at that time.
    https://tallbloke.wordpress.com/2010/07/21/nailing-the-solar-activity-global-temperature-divergence-lie/

  23. […] 3. Global Warming Projections using Solar Cycles […]

  24. Stephen Wilde says:

    “On a different topic, can you think of any reason why different solar cycles should generate different forcing impacts? I can fit observations only when I assume they do.”

    I’m of the view that the ocean cycles are involved because they affect the zonality/meridionality of the jets and the latitudinal positions of the permanent climate zones from below.

    The ocean cycles drift in and out of phase with the solar cycles and so the combined solar/oceanic effects would produce different forcing impacts in different solar cycles.

    The forcing effect being the amount of solar energy getting into the oceans which would be related to total global cloudiness at the time.

    As regards ‘missing heat’ I think it just gets reflected back out to space when cloudiness/albedo is higher and so is lost to the system forever.

  25. Roger Andrews says:

    Stephen

    Thanks. Don’t know how we’d go about proving this, but it sounds plausible.

  26. P.G. Sharrow says:

    I agree, the “missing heat” is lost to space. Not hidden, just book keeping error.
    Time to include the most important part of the hydrological thermostat of earth, CLOUDS, into the climate models. pg

  27. adolfogiurfa says:

    @P.G. Sharrow : That thermostat….don´t forget that without charge there are no clouds, no possible tons of water “levitating” against sacred gravity laws.

  28. J Martin says:

    As Roger’s model very effectively shows that there can be no back radiation to the sea, then there can also be no back radiation to the land. If that’s the case then how does co2 act as an anthropogenic component.

    Perhaps we need to find something else, most likely candidates being GCRs and a transfer of heat from SSTs to the atmosphere and hence to the land, or I don’t know what.

    Perhaps there is a threshold above which the the sea gives up it’s acquired heat (> Tallblokes 40 SSN ?), and does so over a longer period of time.

    It’d be nice to get rid of that co2 component.

  29. Roger Andrews says:

    J Martin:

    “As Roger’s model very effectively shows…” I think the models are a little too crude to draw any firm conclusions from, but thanks for the vote of confidence anyway.

    But if we accept the model results we also have to accept that CO2 does in fact heat the air – which is OK because physics says it should, all other things equal. But according the models CO2 doesn’t heat the oceans (i.e no back radiation), and this brings up a point I hadn’t thought about before.

    If CO2 doesn’t heat the oceans then the air temperature response to increasing CO2 isn’t delayed by ocean thermal inertia. We don’t have to wait hundreds of years for the CO2-induced heat in the oceans to seep back into the air because it was never there to begin with. The temperature response to increasing CO2 is felt only in the atmosphere, and it’s effectively instantaneous (again with all other things equal).

    Which means that there’s no such thing as “equilibrium” climate sensitivity – only an “instantaneous” climate sensitivity. And the SAT model gets a best-fit at an instantaneous sensitivity of 0.8C. This is a fraction of the 2-4.5C range of equilibrium sensitivities and only about half of the “transient” sensitivities estimated from climate models. (The transient sensitivity is actually an equilibrium sensitivity estimated over twenty rather than hundreds of years. It still contains a significant ocean thermal inertia CO2 feedback component.)

  30. tallbloke says:

    Hi Roger. 0.8C/2xCo2 sounds more reasonable. 450Myr BP the co2 level was 20 times higher than now, and the air temp was about 22C. (i.e. ~8C higher than now)

  31. Roger Andrews says:

    TB:

    Those numbers actually give a CS of 1.85. But according to

    which I assume is the reference you used, the uncertainties are a little on the large side.

  32. tallbloke says:

    Roger, yes, I was recalling that graph from memory (always risky for me. I should have said 550Myr ago co2 was 20 times higher. You’re right about the theoretical CS. Uncertainty noted,

    We saw 0.8C over the last century. I think half of that at least was nothing to do with co2. The alleged co2 effect is logarithmic so we’ve already seen most of what a doubling from 280ppm might give…

  33. Roger Andrews says:

    TB:

    You win the cement bicycle. According to my model only about 40% of the SAT warming in the 20th century came from CO2 (https://tallbloke.files.wordpress.com/2011/05/ra-fig5.png?w=600&h=434 )

    And I’ve just come across yet another example of apparent data fudging. The GISS TOA radiative forcing estimates I used to factor temperatures in the model have just been revised (by Hansen et al., 2011). No longer is there a 1.92 w/m2 increase in net forcing between 1800 and 2003. It’s now down to 1.50 w/m2, with all of the 0.42 w/m2 decrease occurring between 1990 and 2003 (previously we had a 0.52 w/m2 increase over this period, now we have only 0.10 w/m2).

    What’s the impact of the “revisions”? Well, they explain why the “surface” temperature record shows no warming over the last ten years or so. It’s because there’s been no increase in net TOA forcing over this period. Since 2000 the positive forcings contributed by greenhouse gases, black carbon etc. have been totally offset by negative forcings contributed by sulfate aerosols. But just you wait until those aerosols go away ….

    Verily this doth smell of fish.

  34. tallbloke says:

    Roger, the real reason there’s no increase in net forcing is because the oceans have been losing net energy to space since 2003. The ‘missing heat’ is on it’s way to intergalactic space.

    I have no faith in Hansen’s ongoing tinkerings with the historical data. In fact little faith in the historical data even before the tinkerings. That’s why I’m working on rough estimates and large scale wiggle matches to try to understand the major drivers.

    Well done documenting Hansens’s manoevres though.

  35. Roger Andrews says:

    A scientist woke up one morning
    Distressed by the absence of warming
    Said he “sooner or later
    We must tweak the data
    Or our theories will all die aborning”

    🙂