dMT/dm – A Northern View

Earlier we saw a revised version of the dT/dt code that ‘blended duplicate numbers’ more smoothly. The examples in that post were more Southern Hemisphere focused (largely because I’m fascinated with the southern hemisphere…)

In this posting, I’ll be giving some sample northern hemisphere graphs.

North America

Strange how February and March are falling, while February, April, and July are flat. ( I hope I got those right, the colors are a bit hard to read some times.)

North America blended duplicates cumulative monthly anomalies

North America blended duplicates cumulative monthly anomalies

It is the winter months that really climb. November, December, January. Though, oddly, so does June… Wonder if ‘sun on tarmac’ is greater in June?…

The United Kingdom

That magenta ‘all months’ line has a dead flat trend, though I do note the ‘hockey stick’ ending it gets after the 1990 processing changes. About equal numbers of months falling as rising or flat.

The United Kingdom monthly cumulative anomalies

The United Kingdom monthly cumulative anomalies

Here we have September, October, and November rising, but we now have December, January, and February falling. July and Aug too from the looks of it.

What to make of this mix of conflicting directions? I’d make of it that it’s a bad idea to splice a semi-random set of thermometers together while changing the instruments and the processing. But then, I’ve thought that all along.

Canada

Oh what a MESS! Yeah, Canada is getting “hotter” than other places. But it’s all due to just a couple of months. Look at November, January and February and how they rocket up, then tell me how December, while rising, rises far less? Just screams “instrument and processing problems” to me. Then toss in that September and October are falling while August rises …

Canada blended monthly cumulative anomalies

Canada blended monthly cumulative anomalies

March, April and May are on a tear upwards too, but June and July are basically flat. Then from 1990 on the volatility is basically dead while the volatility to the downside is simply missing entirely.

Mexico

Those major dips look a bit like evidence for a 60 year cycle, in which case we are at a top about to head down. My guess is that it’s a thermometer selection artifact. Looking at individual station temperatures would clear that up fairly quickly. ( In fact, for all the “trends” in these aggregate charts, a look at actual reliable station temperatures would rapidly say if they are real trends in real stations or ‘splice and dice’ artifacts… “Dig Here!” )

Mexico blended cumulative monthly anomalies

Mexico blended cumulative monthly anomalies

January, March and June falling, April, July and August rising? So when is the major vacation travel to Mexican airports? Is there a spike for the April spring break and a drop in June at graduation?

France

We still have the strongly down trending aggregate anomaly line (magenta) that gets “Hockey Sticked” in the end… That’s encouraging (as it shows no major impact from the change to blending the Duplicate Numbers together rather than highlighting the change of Duplicate Number with a reset). We can have some faith that results are consistent, even while the monthly trends are made more clear.

France blended cumulative monthly anomalies

France blended cumulative monthly anomalies

The most striking thing to me is that December / February crossing set. Decembers dropping like a rock, while February rises strongly. January falling slightly while November slightly rises. Frankly, it looks to me like the December data are somewhat wrong, but that the rest of the data has been fudged some too. Or perhaps French airports are just not very busy at Christmas and lots of folks in France take a small vacation to the Islands in February while the tourists head to France for an ‘after the holidays’ getaway ?? … (Hey, it’s a more interesting “story” than the AGW one or than the “bad temperature data” most likely conclusion…

Again of note is the ‘clipping’ of cold going anomalies after 1990. We are now a full 3 C above the W.W.II era. The tops, however, are very much the same recently. (Though I must call attention to the “thinning of the tops” in the baseline interval. Just look broadly at “how much ink” is in the monthly data in that space between 1980 and 1990. Notice the “center of mass” drops from about -1 C to about -2 C, then pops back up right after the end of the Baseline, making the “Hockey Blade”. This, to me, speaks strongly of biasing the baseline interval to the cold side.

This is another case where looking at individual station data from good long lived stations would tell us if France in fact had a cold dip and a big rise; or if they were just being “sticked” in the end and having their baseline sagged… This, IMHO, is a really BIG DIG HERE! It ought to be fairly easy to pick a half dozen well qualified French stations and validate the basic real trends. They would need to either match this aggregate or we would know that the GHCN data set was “cooked”.

Germany

Another confirmation that the aggregates don’t change much in interpretation with this smoother monthly data process. Germany still has a generally dead flat aggregate trend line (magenta) with the present “rise” about the same as past rises. December is falling here, too (So maybe there is something to the colder Decembers in continental Europe? Or at least the processing is consistent.)

Germany blended cumulative monthly anomalies

Germany blended cumulative monthly anomalies

But January is rising and April is plunging… Is it REALLY getting to be ‘horridly cold’ for “Springtime in Germany”?

Once again we see the “thinning of the highs” in the baseline and the clipping of the lows just after. This just looks so… so… ‘manicured”…

Poland

December still plunging, but now January on fire! Is springtime coming to Poland in January? Look at that February, March, April warming! Then it goes flat for the summer. OK, no added warmth, but a less cold springtime. Just don’t try to keep warm in December… So does that match what folks on the ground in Poland have observed over the last 50 years? Surely someone knows an “Old Pole” they can ask.

Poland blended cumulative monthly anomalies

Poland blended cumulative monthly anomalies

Russia – Asian Sector

But if it’s too cold in Poland in December, they can always go to Siberia to warm up. Look at that December rise! November rising nicely too. Just be gone by February when it’s getting even colder than it was in the harsh Siberian Winters of the early 1800s and that spot of trouble the Germany Army ran into in W.W.II. Colder than that has to be mightly cold…

Russian Asia blended cumulative monthly anomalies

Russian Asia blended cumulative monthly anomalies

China

July and October dropping. Many months simply flat. Then some winter months warming. Especially February. Is there something spicial about February in China?

China blended cumulative monthly anomalies

China blended cumulative monthly anomalies

That February has a deep “sag” in the baseline period used by GISS and CRU and pops up just after 1990 is, er, um, “very interesting”. That November does the same, in an overall flat trend, is fascinating…

Japan

Tis a puzzlement. I have no idea what to make of Japan. We have a modestly rising annual trend line (magenta), but with August and November falling. Then there are about 4 flat months (though of them, March is drifting downward)

Japan blended cumulative monthly anomalies

Japan blended cumulative monthly anomalies

I swear it looks to me like alternate months are flat / rising / flat / rising …

Perhaps that Japanese penchant for order, symmetry and structure has gone just a bit too far?

Also of note is that the “cold going peaks” get clipped about 1 C – 1.5 C after the 1990 change of processing. The “hot going peaks” get about 1/2 C of a clip. So we have about a 1/2C – 1C differential hair cut at that processing change.

That could explain a lot…

About these ads

About E.M.Smith

A technical managerial sort interested in things from Stonehenge to computer science. My present "hot buttons' are the mythology of Climate Change and ancient metrology; but things change...
This entry was posted in dT/dt and tagged , . Bookmark the permalink.

13 Responses to dMT/dm – A Northern View

  1. Peter Dunford says:

    Chiefio

    A little edit needed, right at the beginning February is falling, then it is flat…

    Pete

  2. E.M.Smith says:

    @Peter Dunford:

    I had a choice to make. Do I try to describe each month as it wiggles over time, or describe the trend line. I decided to go with the trend line. Yeah, the actual monthly data has a lot more ‘texture’ to it. Yeah, it was a ‘simpler and faster’ choice to just do the trend line. This way it gives you folks more to look for in the graphs, though ;-)

    So I’m not going to edit the text to start describing wiggles, just leave it to comments (and commenters!) to do that ;-)

  3. Alexander says:

    Chiefio,

    As a Russian, I can assure you that German Army had no problems in the Asian sector of Russia – for the simple reason that they never got there, except as POWs. :-)

    Just a minor nitpick.

  4. Keith Hill says:

    OT but first some good news from Oz. In an election year, Labor Prime Minister Rudd, who was voted in partly on his rhetoric of “dealing with climate change, the great moral and economic challenge of our age”, has dropped his plans for introduction of an ETS and/or CPRS until 2013! He has borrowed us heavily into debt and splashed money round like a drunken sailor into poorly targeted and badly managed schemes, so is looking to save A$two and a half billion that would have been spent on compensation.

    Chiefio; for many years up to 2006/07it seemed to be generally accepted , even by the AGW lobby, that there had been no global warming since 2000.
    By 2010, it was being claimed that the last decade was the warmest on record.
    In your examination of all the data, did you find any indication of something strange going on in or after 2006/07?

  5. E.M.Smith says:

    @Alexander: OK, good call. I knew that… All I can offer in ‘self defense’ is that I intended to add the Russian European sector graph too, but haven’t done it yet and tend to think of Russia as a whole, not divided by sectors. At any rate, consider that nit harvested ;-)

    One thing that has always fascinated me about Russian history is how so many times folks decide to invade, and just get demolished by the winters. Yet the Russians are living in them every year… Clearly Russians have some winter skills about which others are quite clueless…

    As the next glaciation gets going (sometime in the next few thousand years) I guess we’ll all need to study Russian 8-) and pick up that wisdom…

  6. E.M.Smith says:

    @Keith Hill:

    There is a “small dying” of thermometers in 2006. A discontinuity in the data.

    Also, the changes made in 1990 continue to raise the bottoms on an ongoing basis. (I suspect it’s that QA process that gives control to airports / AWS stations and requires any new data to be conformed to past accepted data, thus giving the opportunity for feedback over time to suppress ever more cold data. The Vjones “digging in the clay” graph that shows ever more data dropouts in winter months over time enters into this too.

    I looked in some detail at the stations dropped then in this posting:

    https://chiefio.wordpress.com/2009/10/22/thermometer-langoliers-lunch-2005-vs-2008/

    Good news about the Australian politics. Just a ‘delay’ at present, but one hopes that as the cooling continues folks will just distance themselves from the ‘issue’… and the whole AGW idea will just collapse of it’s own weight.

    Politicians have a great ability to “distance themselves” from a movement without ‘dissing’ it, once they smell the shift of ‘trendiness’. So he had choices about what to “distance’ from, and decided AGW as the “safer fall guy”. All very good. Yeah, hedged the bets with a ‘get back to you later’, but betting on a resurgence of warming is not going to work out. Not with a PDO flip, Arctic Ice growing, winter storms in April in the Sierras, snow on the Mediterranean cost in winter, etc. etc.

  7. KevinM says:

    “Clearly Russians have some winter skills about which others are quite clueless… ”

    Stow your gun in the tool shed, have plenty vodka on hand, and only get out of bed to wizz.

    Not joking.

  8. Ruhroh says:

    Hey Cheif;

    I was thinking that the way to approach this thing might be to treat each distinct thermometer record as samples of a noisy voltage, and just calculate the RMS noise, 1/F noise, etc.
    I guess there is a need to normalize it per unit time, or whatever, so that there is not intrinsic bias for short or long records.

    But from the conspicuous clipping of the negative peaks in the recent data, it would seem to be possible to show that the ‘value-added’ datasets have different ‘volatility’.

    Maybe one ~good thing about all of this; no need for fancy graphix to deliver or de-spleen the results of the analysis.
    You’ve certainly given plenty of clues to the ‘statistical professionals’ in the audience that they should take a look at the dang data to see if there is some clipping of negative values.

    Anywho, almost done with the great 2+ year duration fire drill.
    OK, don’t quote me on that…

    Hmmm, not obvious how to handle ‘missing’ samples when they are not uniformly distributed… Maybe ‘non-stationarity’ or that other cool term ‘heteroskedacticity’ could be invoked to ‘prove’ that apples are not oranges, or at least reject the null hypothesis of indistinguishibility…

    I guess the general tool would calculate ~power spectral density within (rolling?) weekly or monthly windows?

    Having identified the more fishy and less fishy time epochs, how hard is it to have the computer automatically find and flag the curious compression of volatility despite declining thermometer counts?

    Just repeat the mantra; No Grafix, …
    OK, I gotta get Back to Work…
    PGE is pulling the plug on my respirator again tonight.
    Yikes…
    RR

  9. RuhRoh says:

    Hey Cheif;

    I was thinking that the way to approach this thing might be to treat each distinct thermometer record as samples of a noisy voltage, and just calculate the RMS noise, 1/F noise, etc.
    I guess there is a need to normalize it per unit time, or whatever, so that there is not intrinsic bias for short or long records.

    But from the conspicuous clipping of the negative peaks in the recent data, it would seem to be possible to show that the ‘value-added’ datasets have different ‘volatility’.

    Maybe one ~good thing about all of this; no need for fancy graphix to deliver or de-spleen the results of the analysis.
    You’ve certainly given plenty of clues to the ‘statistical professionals’ in the audience that they should take a look at the dang data to see if there is some clipping of negative values.

    Anywho, almost done with the great 2+ year duration fire drill.
    OK, don’t quote me on that…

    Hmmm, not obvious how to handle ‘missing’ samples when they are not uniformly distributed… Maybe ‘non-stationarity’ or that other cool term ‘heteroskedacticity’ could be invoked to ‘prove’ that apples are not oranges, or at least reject the null hypothesis of indistinguishibility…

    I guess the general tool would calculate ~power spectral density within (rolling?) weekly or monthly windows?

    Having identified the more fishy and less fishy time epochs, how hard is it to have the computer automatically find and flag the curious compression of volatility despite declining thermometer counts?

    Just repeat the mantra; No Grafix, …
    OK, I gotta get Back to Work…
    PGE is pulling the plug on my respirator again tonight.
    Yikes…
    Whoops, forgive the doop if I did it.
    Sorry, Crispy Critters here…
    RR

    REPLY: [ For some reason this went to the SPAM queue. No, I have no clue. Nothing I've done with filters. On 'finding fishy compression' the thing that comes to mind for me would be something akin to Bollinger Bands.

    http://www.bollingerbands.com/

    Used with stock as a way to show local volatility changes...

    And yes, that the volatility goes DOWN as there are fewer thermometers is just wrong. A single thermometer can be quite volatile, that's why the start of the graphs is so jumpy, but as more are added, the odds of all of them jumping to an all time high or low at the same time drops; volatility compresses some. Then when thermometer counts drop we have volatility dropping even more? Doesn't add up. That it is an asymmetrical volatility compression with low going peaks clipped more than high going is just dumping gasoline on the fire... -E.M.Smith. ]

  10. Kari Lantto says:

    Winters are getting warmer, but summers not so much. Well that could/must have at least a little bit to do with us humans. I’m not an alarmist and will not advocate less heating of our houses. But surely that must be a (perhaps very small) part of the heating up of the globe during NH-winters. What about SH-winters? Nobody lives down-under?

  11. Kari Lantto says:

    Isn’t this what we mean by heat island effect. Only that we mainly heat up those islands around the thermometers in the winter when we think its too cold for comfort. Sorry about reiterating known facts. Right?

  12. E.M.Smith says:

    @Kari Lantto:

    Yeah, Urban Heat Islands from running the heaters may be part of it. The bigger issue, IMHO, is that “Airports percent”.

    We’ve gone from effectively NO airports (and actually none prior to 1914) to large expanses of tarmac and concrete with tons of kerosene being burned daily. (Almost 100% airports in France and the USA, for example – 92%). The same thermometer that was on a Knights Templar parade ground at Templehof Germany ended up at a major international airport in downtown Berlin by the modern age.

    Nice picture here:

    http://upload.wikimedia.org/wikipedia/commons/a/ae/FlughafenBerlinTempelhof1984.jpg

    So while part of it will have been the home heating for kilometers in all directions, another large part will simply be that a large expanse of black tarmac absorbs more solar heat in the dead of winter than a similar expanse of snow. And airports keep the snow cleared away, while the parade ground did not.

    Further, at the “shoulder months” leading into and out of winter, the effect can be fairly strong. Think of a nice April day. Sun is out. In onc case you stand on a meter of accumulated snow. In the other case you stand on a square kilometer of black tarmac with 747′s on each side burning a few tons of kerosene to get off the ground. Which will be ‘warmer’?

    So we can bat it back and forth about how much of what you see in that picture is tarmac and how much is home heaters in the distance. But in all cases that thermometer is no longer sitting in an open grass field with snows in the winter and little else…

    So yes, we have UHI and it’s a ‘bubble’ over the city and over the airports (sometimes called Airport Heat Island or AHI). Just a ways out of town the temperatures can be up to 4 C lower on any given day.

    IMHO, the failure to properly locate thermometers and properly remove the UHI from the record accounts for more than the measured “warming” of the planet. No CO2 effect needed at all.

    Oh, and I ought to point out that the dTdt graphs are all “UHI included”. It’s a ‘worst case’ warming in that NO allowance has been made for any of the fake warming signals from things like UHI and airport percents. So when dTdt says there has been no or little ‘net warming’ from 1825 or 1930-40 you can pretty much figure that means in reality we have likely cooled off a little bit once you take the UHI and AHI out of the data. But I wanted a clean and unadjusted report at this step so adjustments are clearly visible (when eventually done later).

  13. Casey says:

    Moderator,
    Scratch my previous comment. It’s wrong. Apologies.

    REPLY: [ OK. -E.M.Smith ]

Comments are closed.