First look at filtered Hovmoller 1850 onwards, what does this mean?

Posted: December 19, 2012 by tchannon in Analysis, climate, methodology


I had no intention of doing a post when I remembered late what I had meant to do today but the result is I think important.

This is a hovmuller plot plus two incidentals which come along for the ride, existing code, lots in involved here.

North at top, South at bottom. Month slices side by side vertically.

How do I explain this one, sheesh. Shown these plots a few times. The change here is quickly hacking an existing experimental facility to call a low pass filter for each horizontal line of temperature, in this dataset has them at 5 degree lattitude intervals.

Quick what-if, low pass at about 10 years, set the colour range to +3C.

Worked perfectly first time. (looks over shoulder, what’s the catch?)

Two mysteries.

  • What was going on 1940 to 1960? Does this give a clue on SST?
  • 2000 onwards, why the appearance of cold to the far south?

Far too late, off to sleep. I leave it to any early readers to figure it out.


Pulled update of hadsst2 and crutem3, plotted and made a composite, ouch.


Note: crutem scaling is different

Makes it crystal clear SST dominates. Is my code right? I hope so.


  1. Stephen Wilde says:

    i) 1940 to 1960 (actually right up to 1980).

    Active solar cycles17, 18 and19 caused zonal jets reduced clouds and more energy into the oceans but were followed by less active cycle 20 which caused more meridional jets, more clouds and less energy into the oceans so we see a rather mixed signal from 1940 right through to about 1980

    ii) 1980 to 2000.

    The cumulative effect of strong solar cycles right through from cycle 17 to 23 really got under way once the mitigating effect of cycle 20 has dissipated.Active cycles 21, 22 and 23. More zonal jets, less clouds and more energy into the oceans.

    iii) 2000 to date

    Much quieter cycle 24. Meridional jets again with more clouds and less energy into the oceans.The first signs of cooling appearing in the south. It takes some time for the effects of a run of 7 strong cycles to work out of the system.

  2. Roger Andrews says:


    HadCRUT3 is an area-weighted average of SSTs over ocean areas and SATs over land areas, so SST is bound to dominate.

    Again, however, I have to point out that you are using a VERY heavily-manipulated SST record:

    A Hovmoller plot of the corrections might in fact be more revealing than a plot of the corrected data.

  3. tchannon says:

    Solar? Given the update it looks like back to square one, is dominated by sea temperature.

    Lets whack a bandpass filter on SST, bit over 3y to bit under 8y.


  4. tchannon says:

    Decode/import of ICOADS would take some time, non-trivial.
    Which of these looks a good choice?

    When or if, don’t know.

  5. oldbrew says:

    TC says: ‘What was going on 1940 to 1960?’

    High sunspot numbers?

  6. tchannon says:

    ssn is high then, general fit with ssn is not good, why I am playing down. Any obvious linkage would have been found years ago. An indirect effect is a different matter.

    Late WWII is where a discontinuity exists in the data. so I was wondering whether this gave a clue on why.

    Just tried reducing the filter Q, doesn’t make much difference, try shifting the passband, weird stuff. Can’t really show anything, this is play around time. Have just made some code changes which might help.

    One distinct thing which might be telling, you can see it but less clearly on the plot just above, there is a pale break at 50N or so, more confused in the south with what looks like phase slippage (or time) either side. Is this to do with the circulation cells, where another ground zone exists?
    Give those are large time periods this makes little sense.

    All assumes of course the data is sane. I’m sure I am not time sliding neither does filtering care about longitudinal timing.

    Also possible this is aliasing of something or other. Daily data isn’t going to happen.
    I’m not set up to do 2D filtering.

    Might be that other visualisations would add insight, not easy because this is DIY land off the beaten path.

  7. Doug Proctor says:

    An observation:

    the 20-year cycles of 1880, 1900, 1930 (very weak), 1940 (strong: equal to the non-existent 1930 + “normal” 1940?), then smeared 1980+:


    The heat energy for the 1940 and earlier cycles comes from somewhere either outside the system (the sun) or “buried” inside the system. Where from if not the sun? The oceans? Then a release function, not a new input function?

    Question 2:

    If the 1920/30 non-cycle is in the 1940 cycle, are we seeing some positive interference of SST cycles?

    Question 3:

    It’s been said here that Fourier analysis can give you anything without it meaning anything. Can a deconvolution (??) to remove the earlier cycles by considering out-of-phase constructive and destructive interference of (say) 3 cycles, give us the (constructive) phase in the post 1975 period?

    On a related note: ice to water to vapour are edothermic state changes with large heat uptakes, the reverse, exothermic event happening for the vapour to liquid to solid state changes

    Question 4:

    What happens to global temperatures when the Arctic ice melts and the Antarctic ice grows (and the Antarctic sea cools)?

    Comment and Question 5:

    ASSUMING a closed system that is stable for TSI and planetary radiative releases (IPCC base assumptions) what one looses the other must gain. If the relative, proportional masses involved of water, air and ice, are different but the total energy redistribution is (therefore) energy neutral, air temperatures would not necessily be the same. Couldn’t regional energy redistribution in this scenario cause global temperature “anomalies” without necessitating hidden or extra retained heat?.

    Probably both a common idea and an idea already shot down …

  8. Roger Andrews says:

    Oldbrew: “What was going on 1940 to 1960? High sunspot numbers?

    Doug Proctor: “The heat energy for the 1940 and earlier cycles comes from somewhere either outside the system (the sun) or “buried” inside the system. Where from if not the sun? The oceans?”

    All of the above, I think.

  9. tchannon says:

    Plenty of patterns, looked at the arguably better coverage satellite data as well. For now I think this is one to put on the back burner.

    Perhaps satellite tls is most novel.

    Right now I’m not doing too well with content. Rog will be back so there is no point in starting much.

    Darn rain, bathroom wall is wet… how is that getting in. Rain rain rain for some time to come, seems as though we are 24 hours+. Going to be some pain killers and then crawl, see if I can figure it out. Can’t see anything from the roof.
    .Yesterday apparently there was no sunshine anywhere in the UK.

  10. Doug Proctor says:

    .Yesterday apparently there was no sunshine anywhere in the UK.

    Bet the daytime temperatures were lower ….

    Not that that will affect any records.