(Click on the graph for a bigger version)
Why am I saying Flat when it’s got a couple of degrees rise right at the start?
This chart is the “Change of temperature per year” or dT/yr and the running total of that over time: dT. This is ALL data for North America in the GHCN data set. ( The February 2010 vintage download). Those first few years are interesting, but it is only one thermometer for many years, eventually becoming 2. You have to get to about 1800 or 1825 to get into a decent number, like 20. So now look at that chart again and notice that the dT line is at about zero near 1825. We’ve gone nowhere in almost 200 years. You can see “Eighteen Hundred And Froze To Death” as a down spike after a major volcanic event. You can see some rise out of the Little Ice Age. You can even see a hot 1934 as recorded in our history. Notice also the relatively cold period of 1951 to 1980 where GIStemp put their baseline period. The entire Industrial Revolution is substantially “Flat with a ripple”. And that is what I mean by “Flat” in North America. (We will see a much prettier, but harder to read, version at the very bottom of this posting, that starts time in 1825 just to show how flat it has been).
About all you can really say is that if there is any CO2 effect, it is to compress volatility from the downside putting us at about zero with a hard lid at +1 and ever less down side risk of another Little Ice Age. I can live with that…
North America – Not Quite What You Would Expect
In looking at North America, the dT/dt report is rather interesting. I’ve looked at all the countries of the world as the “tabular reports” (since it takes about a minute to make the table and it takes me about 10 minutes to make a graph out of it, if I’m lucky. It’s much faster for me to screen the tables). Overall, North America is not warming, but there are some parts that do have a “warming signature”. But that “signature” is very oddly distributed. Some countries have it, some countries don’t. And some countries are just dropped on the floor.
I’m going to, for now, just list the countries that are “flat then dropped” in the tables. I could make relatively flat graphs out of these, but what’s the point? They get ‘cut off’ a decade or two in the past and “Global Warming” true believers will simply assert they must be warming even if we have no data. After some more “interesting bits” are explored, I may make a posting with graphs of these guys. For now, you will just have to “trust me” that they are more or less flat when I’m looking at the numbers. The end years are approximations as things sometimes get large data dropouts just before the data ends, and it often ends at an odd month of the year. Countries with the word “NULL” are in the GHCN Country index file, but have NO data.
Country Code, Name, Comments:
408 El Salvador, ends '90 409 Grenada, ends '63 410 Guatemala, ends '85 411 Haiti, ends '93 416 Panama, ends '80 417 Saint Kitts and Nevis, '71-80 only. 426 Antigua and Barbuda, NULL 427 Bermuda, flat but ends in '95 430 Dominica, NULL 436 Saint Lucia, NULL 437 Saint Vincent and the Grenadines, NULL 439 Turks and Caicos Islands, NULL 440 Virgin Islands (USA), ends '80
This leaves some countries with data, but first, just notice that these “cutoff” dates matter. On those dates, “nearby” countries up to 1000 km away can provide “in-fill” data for any missing data in GIStemp. Then, in a later step, a “grid / box” can reach up to 1200 km and be filled in by whatever it finds. While the ‘relationshiip’ will have been determined by the 1951-1980 baseline, the stuff filled in will be based on whatever is happening now. And those are NOT always the same relationship.
As we saw in:
where Bermuda was shown to be quite flat, until it ended in 1995, and where we saw that The Bahamas has a ‘hockey stick’ of about 1.5 C from a change of “Duplicate Number” Flag in 1990. (I’d been calling this Mod Flag, but I’m swapping over to what NCDC has in their docs).
So guess what will happen to all those places that were flat right up until the moment they were dropped?
You guessed it, they will look over at The Bahamas or Jamaica (which gets two lifts from dropouts of old Duplicate Number Flags and additions of new ones):
So even though Jamaica ends up, more or less, back where it started at the beginning of it’s series, the “dropped” places will be filled in with the warming trends from the bottom of the Jamaica baseline period (1951-1980) and The Bahamas Hockey Stick.
So for this reason, I’ve taken to calling this process “The Reveal”.
In a magic act, there are typically 3 steps. There is The Show (or the distractor). This is the “shoving swords into boxes” or the “Put the pretty girl in the box on the table over air”. Then there is “The Trick”. That is tapping the button with your toe that opens the trap door she exits through; or it’s the slanted mirror that makes it look like she is where the swords will pass, but she is really in the “hide” behind that point. The final step is “The Reveal” where you dramatically “open the box and find she disappeared” or “open the door and find she has no sword cuts”. Basically, “The Reveal” lets you see the RESULT of the trick and makes it look like something happened, when it did not.
So here we have Jamaica with some ‘lift’ during part of the time after the baseline (that happens in the bottom of the blue dip), and The Bahamas with a hard core Hockey Stick from the “Duplicate Number Flag Trick”. And with ‘progressive reveals’ we can slowly spread that “warming” around to more and more grid boxes. The drop of the thermometers matters because that is The Reveal. But you can look at The Reveal all day and never see The Trick. To find The Trick you must first see that The Bahamas Hockey Stick happens, and then find out what changed in The Bahamas. Not the first thing you think of when Haiti drops out in ’93 now is it? (And, of course, the flat / down segment of The Bahamas will “match” the existing part of the flat dropped stations so the only rational conclusion is that they would continue to match after The Reveal. And they must have a valid hockey stick from their in-filled and re-imagined homogenized data…
Other Flat Countries
OK, there are some other countries in the area with data up to the present date and with FLAT temperature series. I’ll list them here (and as time permits, I’ll cook up graphs for them to make it more interesting). Some of these are Dead Flat
(like Puerto Rico) (like Saint Pierre and Miquelon Island) and some have a more complicated but generally flat texture. Well get to all of them over the coming days to weeks. But they are more flat than anything else.
UPDATE 16 March 2010: Turns out Puerto Rico has a bit more interesting texture once you get into the details. I’ll discuss it a bit more down below…
For now just realize that these will be averaged in (in that oh so complicated strange “The Reference Station Method” way that GIStemp uses) with The Bahamas Hockey Stick and the other ‘warming’ countries to make a variety of not-quite hockey stick ‘trends’ in any nearby country that truncated above, and to fill in any missing months for those that still have a temperature series, but with lots of holes in it. The end result is that The Reveal (be it a truncate Reveal or a “holey data” Reveal) will end up making a rising trend any time one of those flat places starts dropping data within 1000 to 1200 km of The Trick stations. So watch these flat stations going forward to see if they develop any of: The Hockey Stick Trick, The Truncate Reveal, or The Data Dropout Reveal.
401 Barbados 405 Costa Rica 406 Cuba 412 Honduras - Has many dropouts 414 Mexico 415 Nicaragua (has a bit of warming) 434 Netherlands Antilles 435 Puerto Rico 438 Saint Pierre and Miquelon Island
(The USA is not listed, even though it is generally flat, as the regions inside the USA have some variation in them. For this reason, depending on what part of the USA you are near, you might pick up a bit of warming or cooling by proxy.)
Saint Pierre & Miquelon Island (France)
I’m beginning to see a pattern here. Places with an Anglo influence have hockey sticks, places with a French influence didn’t get the memo, they were busy with Bordeaux and Brie at the beach… Besides, those silly Englishmen are always making so much noise about something other other, who can keep up? Care for a bit of paté with a crusty baguette monsieur? We can always change the thermometer some other day… besides, the one that is there, she worked fine yesterday, non? Come, a little time with the sun and a bit of le bon vivre, c’est très nécessaire pour la bonne vie…
It does have an odd “windshield chip” bulls-eye in about 1992, though. It would be interesting to see if there was some change of procedure that year or change of Duplicate Number Flag, but no real impact with just one place and one thermometer and maybe even the same one. Anyone want to volunteer for the trip? Remember, you are doing this “For Science” ;-)
ONE thermometer the whole time. Perhaps the same one originally installed years ago. Dead Flat. I can’t think of a better example for why it’s a bad idea to play with the instrumentation in the middle of an experiment than to compare both the Bermuda Islands and Saint Pierre & Miquelon Island to the hockey stick in The Bahamas. IMHO, The Bahamas data needs to be struck from the data set and the old thermometer (or the records from it if still being collected in Bermuda) put back in.
While it starts and ends at about the same place, Puerto Rico turned out to have some interesting details in the fine grained charts. First, the “Hair Graph” of the dT anomaly running total (with all those monthly anomaly dT ‘hairy details’ too ;-) Notice that after the thermometer count crashes, the monthly UP peaks reach about 1.5 C while the DOWN peaks only reach 1-1.25 C. We again have the asymmetrical peak clipping.
So looking at this, we do get a rising trend line to the dT line. That cumulative “change of temperature to date”. Even though we have starts and ends about the same. We also can see a bit of the usual ‘pinch’ of volatility as thermometer change happens. (Thermometer counts in about 1980 and Duplicate Number Flags in about 1991) OK, all the indicia of the “cooked books look”. So what happens when we look at it as distinct segments by thermometer count? Looking at the stable periods, with the time of change between them, gives quite different slopes to the ‘trends’:
With stable instruments we have stable trends. With instrument change we get rising trends. Even that dropping segment recently turns into a rise after The Reveal of the drop to one thermometer about 2005. So the future for Puerto Rico is looking much warmer than the separate flat segments, or the “start to end” flatness, would imply.
It really is “All About The Splice” and “Instrument Change”…
The Source of Warming
Just to finish out the set, here are the rest of the countries that are ‘warming’. These have pretty much universally had either a “step function” or a “hockey stick” at the point of a change of “Duplicate Number Flag”. That is, “The Splice” that makes the “The Trick” work. There are none with a smooth rise over time or even a sin wave superimposed on a rising trend as one would expect from a CO2 accumulation signature.
402 Belize (we saw that one where a flat temperature top forms an anomaly rise as the volatility drops and down peaks are clipped.) 403 Canada (looks like a place on fire) 407 Dominican Republic '95 Hockey Stick (a young one!) 413 Jamaica (net it is flat, but has two uplift segments from the baseline lows) 423 The Bahamas (that we've already seen) 424 Trinidad and Tobago 429 Cayman Islands (warming but has a gap from '87 to 2008) 431 Greenland 433 Martinique
So here we have all of 9 countries making all the “fuss”… (what fuss there is).
As we work through the lists, making postings, there may end up being some country where I didn’t read the table of numbers quite right or where I end up changing what list it is in. It’s late, I’m reading little scrawls of red ink on a listing that was cribbed while looking at long lists of numbers, and I could have confused some number with another or some Guyana Giana Dominica Domincan whatever. I don’t think so, but if I stumble on one and fix it, don’t be surprised. I could also find more like Jamaica where the overall movement is small, but their is a dip in the baseline and a rise after.
Now the fun thing to do is put the warming stations side by side with flat stations in the same posting (as we saw with The Bahamas and Bermuda). We’ll be doing that going forward in other postings as we work through all these places.
Back To Cananda
OK, what about that Hair On Fire in Canada?
In this graph there are 12 monthly anomaly series. Each of these is comparing the data in that month to the last valid data in that month and looking at the change. These are averaged together to get the dT/yr (that you see in the summary graphs elsewhere), and that dT/yr is made into a running total of all change to date as the dT line. The “Count” line is the number of thermometer records from Canada and has been scaled by dividing it by 100 so as to fit in the size of this graph scale.
We get about 1/2 C rise coming out of the 1880 cold end of the Little Ice Age and end up running about -2.5 C until 1988 or so (that’s when most of the more interesting new Duplicate Number Flags begin). At that point, notice how the high peaks and the low peaks for the month over month temperature changes (those lines by month) don’t reach nearly as far as they used to? We lose volatility Big Time. Peaks that used to reach +7 C now can barely get to 3.5 C. At the same time, negative peaks that used to reach -6 regularly and even -8 and -9 C on occasion are barely at -2 most of the time and only sometimes make it to the -3 and -4 C range. Further, the number of lines going significantly negative is lower than those going positive.
WIth each asymmetry of those excursions, we’re cooking the history of the Canada Anomaly Total to date.
It is a giant “Dig Here!” to find out what all those Duplicate Number Flags mean for Canada. It could be a change of equipment, or a change of QA process, or? But as you can see from this graph, The Anomaly Process looks at those peak clips and finds a hockey stick. There is a +2 C average Rocket Ride right after The Great Dying of Thermometers Reveal happens (and it’s about 2.5 C total from when the new Duplicate Number Flags were added to the inventory of Canada just before The Reveal).
And it is vastly different in profile from before The Great Dying of Thermometers when the anomaly trend is nearly dead flat.
The process consists of 3 major parts.
1) The Duplicate Number Flag Trick.
SOMETHING is done that merits a new Duplicate Number flag; and the result is a thermometer series that “peak clips” but with a bias to clipping cold peaks more than hot peaks. NCDC via GHCN is the provider of this data and thus of The Duplicate Number flag Trick.
2) The Show.
These records are SPLICED onto the old Duplicate Number records at that location, and they are homogenized into anything within 1000 km that is missing data. That gives the illusion that everything is fine (but leave the tell-tail finger print of that “pinched” area in the monthly dT/yr graph at about 1989-90 (that empty white space just to the left of the plummet of thermometer count in The Great Dying of Thermometers). The monthly data are averaged into yearly averages and seasonal averages and all sorts of things, hiding the ‘pinch’. There are further Gridding and Boxing steps and even a 1200 km “reach” to blend all the data into any boxes that need some. (And merging the “hockey stick” sites with the “flat” sites). GIStemp and CRUT are “the show”.
3) The Reveal.
The old Duplicate Number flag series are ended. You can inspect them all you want and you will find nothing wrong. You can compare the Duplicate Number flag zero of a kept to a tossed all day and find nothing. And since a dropped station by definition does not have a new Duplicate Number series, you are left with nothing to contest. You can only ASSUME it would have done what it’s neighboring “hockey stick” surviving Duplicate Number flag Trick neighbor is doing now. GIStemp drops some thermometers and NCDC drops others. They both participate in some ways in The Reveal. In particular, GIStemp drops any record shorter than 20 years, so by judicious use of adding gaps, a series can be made to be in GHCN but disappear in GIStemp. Similarly, a short record can, over time, grow to 20 years length and ‘suddenly appear’ to add in Trick records over time. It’s a very complicated and very flowery Reveal…
And that is how the show is done.
A good magician will never tell you how The Trick is done. They will share The Showmanship. Sometimes they will share The Reveal. But never the trick.
But at least now you know were to go looking and how to detect it.
Postscript: For what it’s worth, I’ve looked at the tables for every country in the GHCN data series. Every continent fits this pattern. The dates vary a bit. Some places have a 1980 Small Dying. Others have the 2006 Lesser Dying of Thermometers. There is even a bit of evidence for a 1970 or so Tiny Duplicate Number Trick. Further, it looks like Africa had some play time. As though things were being tried there early on (that is just rampant speculation and could easily be ‘projecting’ too much onto data with odd quirks, seeing faces in clouds – but that is the face I see in those African clouds…)
I will be making postings like this one for each continent and then going pack to put up the individual charts for EVERY country with data. It will take time. There are over 100 countries in the world and I can get through about 4 chart sets a day with postings if I hustle. That would be a month if I did nothing else… So a bit of patience, please. For now, though, just re-read the last few postings and see how it hangs together.
In closing, here is the “Hair Graph” for North America from 1825 to date. It’s pretty hard to read, as the lines are so dense and going nowhere much. There is a very broad coverage so many ‘interesting bits’ are being averaged in with ‘uninteresting’ making it harder to see ‘the action’ too. I had to make the dT line Hot Pink so you could see it behind the “Hair” of months, wandering back and forth about zero.
We notice also that there is not much of a drop in thermometer counts in 1990 with The Great Dying of Thermometers. But we also note, somewhat ominously, that the 2006 Lesser Dying of Thermometers has hit North America particularly hard. Notice how the volatility of the peaks just dies in that far right edge after 2006? I think by now you can figure out what that means… and what part of the world was supposed to be “Warming” most in the coming years in programs like GIStemp. At least until the sun went into a Major Minimum and put the kibosh on things…
As you can see, not much has changed with that Hot Pink dT line. Some wobble back and forth, some volcanoes from time to time. But while some places like Canada and the Caribbean have been doing Magic Tricks, the USA still dominates and it’s not been “Tricked” yet, despite some of the Duplicate Number changes in the mid ’70s and at the ’80-’80 and ’90-’90 transitions. Or maybe we just haven’t had quite enough time from the 2006 Lesser Dying of Thermometers just yet…
Bonus Round! Mexico!
Here is the “Hair Graph” for Mexico. It has an exquisite example of the volatility “pinch” at The Splice point between Duplicate Number flags. Goes to zero and sits there in about 1992-1993. Just perfect. And you can see the ‘lift’ starting in the overlap period of the late 1980s (and inspection of the Mod flags will nail it to the exact date, but from the graph I’d say it’s likely to be about 1987 that the new Duplicate Number flags start) and picking up steam as The Reveal unfolds. Mexico is also interesting in that the thermometer Die Off starts early in about 1980. We saw in another posting that the Mexican thermometers tended to move from the colder more volatile mountains into the more tropical Megathermal Zone and back a little ways from the oceans.
For most of it’s thermometer life, Mexico was running about -1/2 C and was down to -1 C ( on an anomaly basis) when the thermometers were taken out back and shot. But despite all the loss of thermometers, The Duplicate Number trick, moving to the Megathermal Zone… all of that HAS managed to make the slope of the segment from about 1980 to date VERY positive, but it has yet to make it back to the zero neutral point on this anomaly graph. Though we can see that lately they have begun adding thermometers again, and we have nice high warm anomaly peaks and very weak cool ones. So, perhaps, mañana …
I wonder how “cold” Mexico would really be were it not for all the gymnastics with instrument change?…