Bill Illis: Clouds account for most of the variability in net radiation at TOA

Posted: September 10, 2011 by Rog Tallbloke in atmosphere, climate, Energy, Ocean dynamics

Over on the Spencer Good, Bad and Ugly response to Dessler 2011 thread on WUWT, Bill Illis quietly drops this little bombshell:

Bill Illis says:

While we are having no luck finding a good correlation between clouds and temperatures in a feedback sense (the scatters are providing r^2 of 0.02) which indicates there is probably NO cloud feedback either way (and the IPCC calculates that positive cloud feedback might be half of the total feedbacks so that is very clearly in question now) …

There is a very interesting relationship between the Net Cloud Radiation levels and the Total Global Net Radiation as measured by the CERES satellite (which I don’t think anyone has looked yet being busy trying to find the temperature feedbacks).

I’m getting Cloud variability being a very large part of the variability in the total Global Net Radiation Budget – anywhere from 65% to 100% (with R^2 between 0.29 and 0.77).

First the (not really convincing but better) scatter using the CERES data (that Steve McIntyre and Roy Spencer made available).

And then the (much, much better) relationship over time.

And then the versions of the data that Dessler provided (where adjustments where made according to the ERA reanalysis dataset which some think is actually a little more accurate). 100% of Net Radiation governed by Clouds with R^2 at 0.77 .

And then over time, a really tight relationship.

So, do Cloud Variations affect the Earth’s Energy Budget? – the title of Dessler’s new paper – His own data says:

holy moley!

Comments
  1. RACookPE1978 says:

    But the GCM’s only assign a single, constant value for all clouds, at all latitudes, for all periods of day and night, for all seasons of the year, across all elevations for all values of humidity and rainfall and percent CO2.

    Right?

  2. HB says:

    That’s impressive!

  3. David says:

    “While we are having no luck finding a good correlation between clouds and temperatures in a feedback sense (the scatters are providing r^2 of 0.02) which indicates there is probably NO cloud feedback either way (and the IPCC calculates that positive cloud feedback might be half of the total feedbacks so that is very clearly in question now) “…

    How many of the other feedbacks are partially interdependent on the possibly false rapid cloud positive feedback?

    The negative feedback on small cloud changes either in amount, type or laditude, could take time as any ocean response would be very lagged. If ocean heat content is a negative feedback to increased clouds, and it is, then this may take time to be realised.

  4. Paul Vaughan says:

    @RACookPE1978 (September 10, 2011 at 11:15 pm)

    You can throw in another important one:
    whether over ocean or land (…and probably lots more too)

  5. Paul Vaughan says:

    Bill Illis often draws attention to interannual correlations at WUWT, but I’ve noticed that he usually limits the data to equatorial bands.

    Anyone know:
    a) What spatial coordinates are summarized in the plots?
    b) where I can find a concise description of the variables plotted (to facilitate avoidance of misinterpretation)?

  6. Brian Hall says:

    Excellent point. Which is, as I grok it, that attempting to tie clouds to temperature is to try for too great a span in the causal chain. But they tie in excellently to the much more proximate measure: “All-Sky Net Radiation”. It then remains to detail the links and relationship(s) between that and temperature.

    Skipping steps in a recipe makes for unpalatable meals.

  7. tallbloke says:

    David: good point. Bart over on climate audit has determined a response time of 4.88 years. I’ll pull that together into another thread here today.

    Paul, Bill is using Spencer and Dessler’s data. He didn’t say he’d limited it to a subset, so I think we can take it as read that it is global.

  8. Paul_K says:

    Hi Tallbloke,
    I suspect that Dessler would argue that Bill Illis’s finding is an unsurprising result. He (Dessler) chose the period 2000-2010 for his analysis precisely BECAUSE, in his opinion, it was ENSO-dominated. His analysis relies on the fact that there were negligible changes in forcing during the period, so the changes in total net flux are dominated by cloud feedback, plus small changes in water vapour feedback and non-cloud albedo changes, which he has sought to account for. I personally think that Dessler 2010 is sunk for a number of reaons, but I am not sure why this result should surprise anyone. Perhaps I am missing something?

  9. tallbloke says:

    Paul, some might argue that ENSO itself is largely a cloud driven phenomenon. See Chief hydrologists posts on Judy Curry’s site.
    The main reason for choosing 2000-2010 is because that’s the period we have reasonably good data for. (CERES)

    See also the new post on Bart’s findings.

  10. Paul_K says:

    Tallbloke,
    I guess no-one doubts that ENSO modulates cloud behaviour. The debate is over the question of whether cloud variation can be caused by phenomena other than temperature-dependent feedbacks, and hence also be a cause of temperature modulation in their own right. But my point is that, in either event, you would expect to see the radiative flux from clouds as a significant portion of the total flux during the 2000 – 2010 period – which is what Bill Illis seems to be showing.

    Bart’s findings are fascinating. I think we need to untangle the issue of the use of seasonally-adjusted anomaly data first. There are major differences in the phase of the annual cycles of solar insolation, temperature, net flux, high-level cloud and low-level cloud. It seems unlikely to me that the “correction” by means of calendar-month anomalies resolves this issue. So, right now, I can’t tell whether a 3 month lag-time (90 deg out of phase) is an artefact of the anomaly calculation or a meaningful signal in the data.

  11. Paul Vaughan says:

    Paul_K (September 11, 2011 at 8:51 am) “I suspect that Dessler would argue that Bill Illis’s finding is an unsurprising result. He (Dessler) chose the period 2000-2010 for his analysis precisely BECAUSE, in his opinion, it was ENSO-dominated.”

    Paul_K says (September 11, 2011 at 10:44 am) “I think we need to untangle the issue of the use of seasonally-adjusted anomaly data first. There are major differences in the phase of the annual cycles of solar insolation, temperature, net flux, high-level cloud and low-level cloud.”

    Eminently sensible comments.

    I’ll repeat my request:

    Can anyone provide a direct link to:
    a. plaintext data webpage?
    b. concise description of the variables.?

    [By the way: I'm not following the discussions surrounding this issue in other fora because the discussions are SO far off track. Bill Illis, however, often makes concise, sensible comments.]

    Regards.

  12. tallbloke says:

    Paul V, see my reply on the Bart thread.

  13. Paul Vaughan says:

    Looks like this is turning into a red herring chase. No time for this. Off to work…

  14. tallbloke says:

    “Tallbloke,
    I guess no-one doubts that ENSO modulates cloud behaviour. ”

    Paul, I have no doubt that clouds modulate ENSO behaviour either. As Spencer shows with his phase space plots in SB10, the feedback is bi-directional.

    ENSO isn’t just about SST’s, but also the buildup of heat in the pacific warm pool which hides from the surface temperature record. Hence the 4.88 year response time found by Bart.

    Cloud feedback to solar is important too. I agree with those who point to marine biota producing aerosols which form cloud condensation nuclei responsible for huge cloud sheets over the pacific. This is a response to Solar UV levels. That is likely why ENSO timing is linked to solar cycle timing, as UV varies over the solar cycle much more than total solar irradiance.

    Anyone who believes that cloud amounts only respond to temperature change is ignoring more than half the evidence in my opinion.

  15. tallbloke says:

    Paul V. So sorry I wasn’t able to spoon feed your demands at the rate you wish.

  16. Bill Illis says:

    I believe the CERES data they are using is 60N-60S.

    While there is a very strong ENSO signal in water vapour levels, temperatures and cloud proxies such as OLR, I’m not getting a strong enough signal in the CERES data.

    Perhaps it is too short for the patterns to show up and although there are sometimes changes in the CERES data during an ENSO event, the signs are opposite to that expected. So, naturally I have to put charts up (because a spreadsheet of data does not tell you anything).

    RSS/UAH Temps, Precipitable Column Water Vapour and the ENSO (the water vapour correlation is even stronger going to 1948) [Hence the need to include the ENSO impact in any climate science paper].

    http://img696.imageshack.us/img696/7996/ensotempsvstcwvaug11.png

    Outgoing Longwave Radiation – Global vs the ENSO (which I have been using as a proxy for clouds previously because on a short-term basis, clouds have the biggest influence on OLR and no other cloud datasets are any good (especially the GISS ISCCP data and you can guess why that is).

    http://img683.imageshack.us/img683/9932/ensovs90nsolraug11.png

    Okay now on to the CERES data from Spencer and the ENSO. Some relationship but far from convincing or too short of a timeseries. The LW radiation is more like we would expect it to be. LW first, SW second.

    http://img687.imageshack.us/img687/2944/ceresclwvsenso.png

    http://img6.imageshack.us/img6/5763/cerescswvsenso.png

    Why

  17. tallbloke says:

    Bill, thanks for the extra plots and analysis. I’ve made the graphs visible to aid discussion.

    60N – 60S makes sense, Spencer has the same latitudes for the SST plots he gives from the AMSRE instrument which I think is carried on the same satellite.

  18. Paul Vaughan says:

    Thanks Bill. You ask why.
    Answer: Aggregation criteria affect summaries of spatiotemporal pattern.

    If someone can point out specifically which variables [by name] from the climateaudit dessler & spencer files were plotted and also point to concise descriptions of the variables, that will be much appreciated.

  19. Doug Proctor says:

    The net result is +/- 0.7 W/m2 or so. Unless I read it wrong, the amount isn’t enough to concern the warmists. Part of the situation, yes, but the five year running average would be much less than that, no?

  20. Stephen Wilde says:

    Clouds respond BOTH to bottom up oceanic influences and top down solar influences.

    At times when the solar influence is dominant (on the 1000 year timescale of MWP to LIA to date) ENSO responds to solar effects on the clouds.

    At times when the oceanic influence is dominant (on the 60 year Pacific Multidecadal Oscillation timescale) clouds respond to ENSO effects.

    Integrating the two components results in the net outcome at any given time. Sometimes the effects supplement one another as in the late 20th century warming period. Sometimes they offset one another as in the mid 20th century weak cooling period.

    However over the 1000 year period of solar cycling the solar effect is dominant to provide 500 years of upward stepping in the ENSO/PDO signal (LIA to date) and 500 years of downward stepping in the ENSO/PDO signal (MWP to LIA).

    All that remains is to ascertain where we currently are in the 1000 year solar cycles.

    Weak solar cycle 24 suggests we are currently past the peak but it does seem a bit early for that to be the case. However the timescales are only approximate due to the lack of direct observations from more than 150 years ago.

  21. tallbloke says:

    Stephen says:
    “All that remains is to ascertain where we currently are in the 1000 year solar cycles.”

    Pretty near the peak, but the current sudden drop in solar activity is a ‘special case’ (once in ~200 year solar lull) which just happens to coincide with the general topping out of several longer term solar cycles.

    Here’s our speculative but we hope plausible take on it:

    http://tallbloke.wordpress.com/2011/02/21/tallbloke-and-tim-channon-a-cycles-analysis-approach-to-predicting-solar-activity/

  22. tallbloke says:

    Doug: true, but the last decade has been pretty flat temperature-wise and ENSO-wise overall.

    The previous two decades are a different story.

    What caused the drop in tropical low cloud cover from 1980 to 1998, higher temperature, or high levels of solar activity?

    My bet is on the latter.

  23. Ulric Lyons says:

    Stephen says:
    “All that remains is to ascertain where we currently are in the 1000 year solar cycles.”

    It is more like an event series than a cycle, though specific kinds of events do return periodically. The next LIA type cluster of cold events comes in from around 2450AD, and that`s by fine examination of the heliocentric configurations responsible, rather than simple projection of periods.

  24. tallbloke says:

    Thanks Ulric, great to see our two methods agreeing. My simple projection method also has the next LIA centred around the same date. :)

  25. Ulric Lyons says:

    Centered around 2650 AD.

  26. tallbloke says:

    Could well be. I’m seeing a cold period ~2450 and another ~200 years later of about the same intensity.

  27. Ulric Lyons says:

    Much like the LIA, there will be a series of cold events over some 400yrs. With my Planetary Ordered Solar Theory, I can plot the events yearly and even seasonally to identify the coldest winters. I can do this over the last few thousand years, and so doing this centuries ahead is completely feasible.

  28. P.G. Sharrow says:

    Hey guys! I’m a little too old to be concerned with the weather forecast for the next 1000 years. The next 2 decades might be nice. 8-) pg

  29. tallbloke says:

    Good call PG. How about it Ulric? :)

  30. AusieDan says:

    Ulric
    Do you have charts that you can share?
    Thanks

  31. Ulric Lyons says:

    For yearly temp`s, a rather similar profile to 179 years back, so cooling from 2014/15 to 2024, this should be the longer of the cooler periods this century, and then generally warmer from 2025.
    Look at the yearly averages from 1835 to 1846 here: http://climexp.knmi.nl/data/tcet.dat
    The timing of the negative deviations in temperature often falls during N hemisphere winters, with some very poor summers too. The return of Murphy`s winter of 1838 will focus in Feb 2017, as the analogue is roughly 179yrs and 1 month. There is nothing alarmingly different in the configurations to make me think there might be any strong exceptions to good repeats of deviations in temperatures through this colder period (2014-2024). There is the question of how much current ocean temp`s may mitigate cooling ~ my guess is not much.

  32. P.G. Sharrow says:

    @Ulric Lyons says:
    September 14, 2011 at 1:33 am
    “There is the question of how much current ocean temp`s may mitigate cooling ~ my guess is not much.”

    I agree, The oceans can only add energy in the form of water vapor that slows cooling. Lots more rain and snow, then very cold and dry. Your Linked averages indicate reductions of 2+ degrees of C. Quite a difference from the 0.6 degrees of C rise for the last 120 years that the AGW people have been complaining about.
    Need to cover more hoop greenhouses. ;-) pg