An oceanic indicator of changing cloud cover?

Posted: April 30, 2012 by tallbloke in atmosphere, climate, cosmic rays, general circulation, Measurement, Ocean dynamics

I was very interested to read the latest post over on Bob Tisdale’s blog on the abject failure of the IPCC’s models in relation to sea surface temperatures. His point is well made, so I won’t belabour it here. My purpose with this post is to spark a discussion about that most contentious of issues: the question of whether or not most of the warming of the late C20th was due not to rising co2 levels, but to a reduction in cloud cover, letting more solar radiation warm the upper ocean.

Here is Bob’s Figure 4 from his post:

Bob says of this plot:

“I used the data through the KNMI Climate Explorer so that I could change the base years for anomalies to 1995-2011. This helped to reduce the strong seasonal signal that appears in the data of some ocean basins. The North Pacific (0-65N, 100E-90W) sea surface temperature anomaly data from NOAA [which uses a 1971-2000 baseline climatology], for example, has a very strong seasonal component, as shown in Figure 4. Using the base years of 1995-2011, also illustrated, the seasonal component is drastically reduced.”

The question I want to explore here is why the strong seasonal signal is drastically reduced by shifting the baseline climatology.This is a question Bob doesn’t explore in his post, but I think it is an important question.

It seems to me that the most plausible explanation is that cloud cover was reduced over the earlier climatology period of 1971-2000, compared to the later period of 1995-2011. This would cause the ocean surface to warm more in summer during the earlier period as the brown curve on the plot indicates. That, I believe is  the reason why the earlier baseline climatology shows a stronger seasonal signal. Both the ISCCP tropical low cloud cover data and the Earthshine project of Palle et al show reduced cloud albedo from the start of their series followed by an increase after 1997. This later increase would explain the reduction in seasonal variation using the later climatology baseline. However, this tentative conclusion will be affected by the question of how, and from what areas the baselines are calculated, so I hope Bob drops by to tell us more.

We have seen in earlier posts on this blog that varying sunshine hours are correlated more closely to surface temperature anomalies than co2 is. The latest was Doug Proctor’s excellent study on the issue. My earlier post on Willie Soon’s far east study is worth a look too. Tim C has hinted recently in comments that he has a major post in preparation on the same issue.

I think we are getting close to determining the real cause of the late C20th surface warming, and that this seasonal variation difference between the two climatology periods is further supporting evidence for our finding. The question then becomes: what caused the changes in cloud cover? Strong late C20th Sun reducing GCR’s a la Svensmark’s hypothesis? Latitudinal shifts in the climate circulation zones a la Stephen Wilde’s hypothesis?

A bit of both? Or something else? Please post your thoughts in comments below.

Comments
  1. Harriet Harridan says:

    Hi TB,

    I’ll kick off with a rough and ready plot of cloud cover anomaly against an inverted HadCrut3 plot.

    Looks good to me. http://oi49.tinypic.com/302uzpu.jpg

  2. Joe Lalonde says:

    TB,

    Nobodies still looked at the surface salt changes that have been occurring in the last 4 decades.

    http://www-pord.ucsd.edu/~ltalley/sio219/curryetal_nature2003.pdf

    As for CO2, many scientists use Venus as an example. Problem is the weight of CO2 on Venus is 400 times heavier due to the direct relationship with velocity difference compared to our planet.

    [Reply] Joe, the weight of co2 on Venus is very similar to its weight on Earth. However, because of the height of Venus’ 95% co2 atmosphere, the surface pressure there is around 93 Earth atmospheres. Now it may be that velocity of axial spin has something to do with the much deeper atmosphere Venus has (though I doubt it because Jupiter has an even deeper atmosphere relative to its core diameter and spins about twice as fast as Earth does), but please stop trying to convince us velocity is linked to gravity in any simple way. Look out for an upcoming post which says momentum has something to do with gravity though.

  3. Joe Lalonde says:

    TB,

    Then why is the CO2 on Venus NOT a liquid under that amount of pressure? We can contain CO2 on our planet in cylinders and it becomes liquid.

    [Reply] Because Venus surface is hot.

  4. mt says:

    The brown line is 1995-2012 anomaly data relative to the 1971-2000 climatology. Which means that the climatology has less variation, since the anomaly data shows higher variability. The 1995-2011 climatology is built using the graphed data, so it should have the summer peaks, which is why they drop out of the red anomaly data.

    So there should be a weaker seasonal signal in the 1971-1995 to reduce the strong signal in 1995-2011, not a stronger one.

  5. Joe Lalonde says:

    TB,

    Thank you on that one! Never thought to add the heat difference.
    Velocity is unlike inertia. In a rotating orb, it is different speeds from core to equator. And from equator to poles. The strongest energy is always at 90 degrees which is the same with centrifugal force. Inertia effects the speeds of all the velocities but not any of the individual velocities. Gases and liquids are effected differently due to the different densities.

  6. tallbloke says:

    MT: Thanks. I’m still unsure on this, hence the question mark in the thread title and the ‘tentative’ conclusion.

    When you say:
    “The brown line is 1995-2012 anomaly data relative to the 1971-2000 climatology. Which means that the climatology has less variation”

    Can you explain why? Sorry if I’m failing to see exactly what you are saying.

    Thanks

  7. mt says:

    If you’re asking for a physical explaination for a change in the data, I have no idea. There may have been less seasonal variability, or the peaks may have been shifted. The cloud vs HadCRUT graph above shows higher than average cloud cover from ~83-93, which if your “lower cloud cover = higher seasonal variability” is true, is consistent with reduced seasonal signal in the longer climatology.

    If you’re asking why I think the 1971-2000 climatology has less seasonal variability just from the graph, look at the red line vs the brown line. The peaks in the brown are generally around June/July. Since most years from 1995-2012 have those peaks, that feature will show up in the shorter climatology. Now look at the red line where the peaks are off a bit, 2002 is early, 2006 is late, and those peaks remain in the anolamy data from the shorter climatology, since those are different.

  8. mt says:

    BTW, “climatology has less variation” is sloppy on my part. A better way to say that would be that the longer climatology is missing or has a reduced seasonal component seen in the 1995-2011 data.

  9. Doug Proctor says:

    Rog says: The question I want to explore here is why the strong seasonal signal is drastically reduced by shifting the baseline climatology.

    Roger, it’s not just a baseline shift, it is a data-value shift. The lows and highs change values.
    But when you change reference period, all you do is change the number from which your datapoints are measured.

    Something odd: it’s like the rolling period for averaging has changed, or that the actual reference period is a rolling forward average. Weird.

  10. tallbloke says:

    OK, well it looks like an opportunity to learn something to me. I hope Bob spots the pingback on his blog and drops by to spread some knowledge here. :)

  11. Roger Andrews says:

    “It seems to me that the most plausible explanation is that cloud cover was reduced over the earlier climatology period of 1971-2000, compared to the later period of 1995-2011. This would cause the ocean surface to warm more in summer during the earlier period..”

    In this case we would expect seasonal SST ranges in 1971-2000 to be higher than in 1995-2011. But the NOAA-NCDC ERSST v3b2 record for the North Pacific shows a change in the opposite sense (data from KNMI):

    1971-2000: Mean seasonal range 4.49C, standard deviation 0.14C
    1995-2011: Mean seasonal range 4.63C, standard deviation 0.15C

  12. Roger Andrews says:

    “The question I want to explore here is why the strong seasonal signal is drastically reduced by shifting the baseline climatology.”

    The reason is that the seasonal detrending algorithm for the 1995-2011 period removes almost all of the seasonal variations from the raw record between 1995 and 2011 while the algorithm for the 1971-2000 period removes only about 95% of them. And the 5% that’s left gives the +/- 0.2C residual seasonal variations we see in the upper plot in Bob Tisdale’s Figure 4.

  13. tallbloke says:

    “the algorithm for the 1971-2000 period removes only about 95% of them. And the 5% that’s left gives the +/- 0.2C residual”

    Thanks Roger. So we could expect to see ~4C of seasonal variation if we saw the raw data?? ;)
    Where can we see the details of the algorithms?

    Cheers

  14. Roger Andrews says:

    TB:

    Your wish is my command :-)

    Re seasonal detrending algorithms, here’s the one applied to the NCDC raw data

    As I understand it these algorithms are constructed using harmonic functions that give the best fit to two average annual cycles, but because there is no such thing as an average annual cycle they never match any cycle exactly. So we almost always see residual seasonal effects in the adjusted data, and they sometimes get pretty large – the Arctic sea ice extent record is an example.

    Another problem with removing seasonal effects from SST records is that ENSO impacts should be backed out first, but I’m not sure they are.

    In fact the only sure-fire way of removing seasonal effects from a monthly record is to take a 12-month running average. but you lose detail when you do that.

  15. tallbloke says:

    Wow! there it is, the 4C seasonal adjustment. :)
    Note well everyone, this is for 0-90 N this trick doesn’t work across the equator.

  16. Roger Andrews says:

    Well, actually it does, but with reduced amplitude.

    And while on the subject of seasonal variations by hemisphere I thought the following graph might be of interest:

    Why are seasonal SST variations so much larger in the NH?

  17. tallbloke says:

    Lots of cold water running off cold landsurfaces into the near surface waters in winter? And the opposite in summer?

  18. The answer to the why question as to the change in seasonal peaks changes with choice of baseline is going to be a little involved; the outer planets cause a pulse of heating as the synod conjunction causes a larger than normal flux of tropical air to be perturbed toward both poles as the outer planets are passed. That was very obvious this March with Mars and early April with Saturn on the 15th.

    The reason the base line shifts the spikiness was due to the slow progression of Neptune and Uranus moving from separated spring conjunctions in the 80′s had a synod conjunction between them in April of 1993 (if memory serves, am away from home do not have reference books with me) and we were (the Earth) passing them in June/July in the 90′s by 1998 they were starting to separate and as a result prolonged the effect to produce the super Elnino and were both synod conjunction with the earth and Jupiter got into the mix by 2005 to create the huge hurricane production numbers, just as the declinational angle of the moon was max for the 18.6 mn cycle at culmination.

    So for the period from 1990 through 2008 was sprinkled with spikes in the summer time the hetrodyning of the patterns in the seasonal effects is the result, and by choosing different base periods it is constructive or destructive of the modulation patterns by the outer planets effects.

  19. Green Sand says:

    BOM keep a daily watch on “Cloudiness” (OLR) at the Pacific Equatorial Dateline, interesting.

    http://www.bom.gov.au/climate/enso/

  20. tchannon says:

    The linear trend for an integer number of cycles is zero.

    If someone can send me the data I might be able to say how it has changed.

  21. Joe Lalonde says:

    TB,

    Interesting reading…this is why I study the different velocities, they all change as the planet slows.
    The algorithms can NEVER be exactly the same for the same reason. The planet is ALWAYS changing from year to year.
    Cloud cover can NEVER go from pole to pole and they NEVER cross the equator. Just watch any time frame of cloud cover or even any cyclones, etc.This is a direct relationship to the speed of the equatorial velocity and the pressure difference.

    This is why I know our measuring of atmospheric pressure by using water as a variable is incorrect. The atmosphere is layered very much like the ocean pressure. Water vapor has a different density and it is the layered atmospheric pressure that squeezes it to a different layer of pressure.

  22. Brian H says:

    Is not the SH-NH difference in variability not mainly due to the higher specific heat of the SH water coverage vs NH land/rock etc.? The same heat flux produces different temperature effects.