South America – Hockey in the Jungle?

South America – Overall Profile

South America - Region 3 - Temperature Change

South America - Region 3 - Temperature Change

There are 15 countries in the GHCN country list for South America. So the puzzle is how to assemble those individual profiles to make this “smooth warming out of a dip” profile.

The truncated country is:

302 – Bolivia, 88-90 has lots of holes, then ends. They, too, produce CLIMAT reports.

And the “by inspection of the tables mostly looking at peaks” FLAT countries are:

305 - Columbia
306 - Ecuador, lots of holes
308 - Paraguay
309 - Peru  (Trying to make the team, though)
312 - Suriname
314 - Venezuela
315 - French Guiana, actually cooling trend
316 - Falkland Islands
317 - South Georgia Islands

Though as we’ve seen in other cases, these can have interesting “texture” inside that overall trend with things like step functions or a drop at the start before the GIStemp baseline that would still leave a ‘warming trend’ in the part of time that GIStemp chooses to use.

Uruguay – Free Basing on the Baseline?

One very fascinating case is Uruguay. At first I thought it was missing, but I’ve since found that I’d left it behind in moving reports from the Linux “Analysis” box over to where I was making graphs. Inspection of the graph is very interesting, and I think it has shown me a new “Trick”.

Uruguay Hair Graph monthly anomalies and cumulative

Uruguay Hair Graph monthly anomalies and cumulative

Notice that blue negative trend cold period? It is EXACTLY on the GIStemp baseline. We pop up to a bunch of stations then, stir them around a bit near the end, and then selectively drop a few of them on the exit. The net effect is to put a nice “dip” in the middle, and a rising trend at the end, with the temperature being roughly the same now as it was in the 1940’s and 1800. Neat Trick.

We still get the 1980 Step Function up and the 1990 “pivot” to a steeper slope. But this is the first time I’ve really paid attention to the baseline dip and I think this the first time I’ve seen the numbers pop exactly on the start and end dates. I’ll have to go back and inspect some of those other graphs and see if any of them do this to. I think I’ll call this the “Little Dipper” (as that’s what it looks like in the graph… ) So the question becomes: To what extent was The Baseline an ‘accidental cherry pick’ of a cool time, and to what extent might it have been “constructed” by thermometer selection…

(Checking the graphs below, it looks like Chile has a “Little Dipper” too, though not as ‘perfect’.)

The Hockey Stick League?

301 - Argentina
303 - Brazil
307 - Guyana (there is that English connection again...)

More than enough to blend in with the surrounding places and smooth the hockey stick… We’ll see as I get graphs made just how much “sticking” each one provides. Inspection of the tables gives an idea what will happen, but not exactly how the picture will look.

Brazil – Hockey in the Jungle Alright

We’ll start with Brazil. It’s large and dominates the middle of the continent. Any evidence for a bit of a Hockey Stick here?

Brazil Monthly Anomalies and Cumulative Chnage Over Time

Brazil Monthly Anomalies and Cumulative Chnage Over Time

( You can click on the graphs to get really big and much more readable ones)

Here we see a common feature, that drop at the start of time, then a long flat phase. I’ve not bothered to break them out for this “Segment by Thermometer Count” graph.

We also have another feature we’ve seen once or twice before. The “Two Step” where we have a “step function higher” in 1980 followed in 1990 by a “ramp higher”. We again get the “pinch” of volatility in the transition just about 1990 and we have a dramatic clipping of the negative going peaks after the pinch.

Something in that changed process of 1990 clips cold peaks out of the data. I’ve left those two segments concatenated as well. Brazil has more of a ‘drift’ of thermometers in and out of the record, so picking clean breaks by thermometer count group is a bit clumsy. This way you can just admire how a long mostly flat history “pivots” in 1980 and makes Hockey In The Jungle.

The other players in The South American League?

Argentina

Argentina Hair Graph monthly anomalies and cumulative

Argentina Hair Graph monthly anomalies and cumulative

A long more or less flat period (though volatile) with a dozen or less thermometers. A “Step Function” up to another slightly tilted segment as thermometers jump to the 50’s and get moved around a little, then a perfect “bullseye pinch” with a small step up but a truly wonderful “warming trend” baked in via clipping the low temperature peaks more than the highs (That ‘white space’ between -1.5 and zero being larger than the white space between 0 and 1.5 after 1990). With the number of thermometers slowly drifting down. Could probably milk this for another decade with careful thermometer removal.

Guyana (That English connection again…)

Guyana Hair Graph monthly anomalies and cumulative

Guyana Hair Graph monthly anomalies and cumulative

More than enough to “lift” the neighbor countries via homogenizing, in-filling, The Reference Station Method and all the other forms of contagion built into the “Climatology Programs”. So lets look at the countries that surround these “Team Players”…

One Just MUST Love The French

As we saw in French Polynesia and in the Caribbean, the French do not believe in Global Warming. Ah, to be French, completely liberated from Carbon-Guilt. I simply must look up a Club Med package… While you are looking at this graph, compare it to that Guyana graph. Then think about the fact that these two countries are right next to each other (with Suriname in between) on the equatorial coast of South America. About the size of Montana for both combined (staring at my globe…) So how do two such close neighbors have such different “Global Warming”… Then go look at the Suriname graph…

French Guiana Monthly Anomalies and dT Trend

French Guiana Monthly Anomalies and dT Trend

We have hints of the Step in 1980 but it’s only good for about 1/4 C and we have the “pinch” of near zero volatility, but again, good only for 1/4 C (and that with a lot of slop in it…) Though I must point out that even a cooling place can contribute to “Global Warming”. Notice that deep dip? Look at the dates. About 1951 to 1980. EXACTLY the GIStemp baseline… So when compared to French Guiana in The Baseline, we will find that the rest of French Guiana history has “warming”… So here we have “How to turn 2 C of cooling into 1/2 C of warming” in one easy lesson.

And The Dutch

Such a wonderful trend. And sandwiched right between the French and Guyana.

Suriname Hair Graph monthly anomalies and cumulative

Suriname Hair Graph monthly anomalies and cumulative

Even Socialist Dictators Can Be Guilt Free

Despite clear indications of an attempt to get a rise out of Venezuela, about the best we’ve got is a relationship going nowhere:

Venezuela Monthly Anomalies and Cumulative Change Over Time

Venezuela Monthly Anomalies and Cumulative Change Over Time

We have the “bullseye” pinch of volatility in the monthly anomalies at the 1990+ change, but the best it could do was a slight drift higher. Not enough to really matter. I’d put it at about 1/4 C midpoint to midpoint. We’ll need to watch it, though, and see if enough accumulates over time to matter. After the Pinch we have a reduction in volatility to the downside out of proportion with that to the upside, so there is some “warming” built into the process, just not enough to get to an overall positive change… yet.

Does Columbia Feel The Heat?

Guess they like the climate the way it is…

Columbia Hair Graph monthly anomalies and cumulative

Columbia Hair Graph monthly anomalies and cumulative

Further down The Pacific Coast

Ecuador

Well, over all darned flat.

Ecuador Hair Graph monthly anomalies and cumulative

Ecuador Hair Graph monthly anomalies and cumulative

A nice new trend started from that “bullseye” 1990 change with good slope to it, but still no more than the 1940’s. Yeah, more than the 1951-1980 baseline of GIStemp. Has potential for the future maybe…. So a bit of a stick, but not very impressive… Maybe all the “holes” in the data can get some better “in-fill” from somewhere else in GIStemp.

Peru

I’d have considered putting Peru on The Hockey Team, but it’s definitely just in the C league farm team category. We have a new rising trend, now that we’ve had some thermometers dropped, but they were dropped after most other folks already had their dropping out of the way.

Peru Hair Graph monthly anomalies and cumulative

Peru Hair Graph monthly anomalies and cumulative

We’ve got a great rise out of the “Dip” in that 1951-1980 baseline that GIStemp uses, but then that long flat handle. (Guess that’s why some pruning was needed). Yet through it all, we’re right back where we were in 1943. Sheesh. All right Peru, skate two laps then give me a shootout run at the net. Maybe next year we can put you on the Big League Team…

Bolivia

So much work to get a trend going, then it got dropped?

Boliva Hair Graph monthly anomalies and cumulative

Boliva Hair Graph monthly anomalies and cumulative

So we have a step function higher as a bunch of thermometers were added, but they were added in the baseline. Doh! Then we get a Great Change Of Trend with that 1980 change, just a Rocket Ride. And it gets dropped on the floor. Doh Doh!

Maybe it’s just “in the bank” waiting to be resurrected when the “lift” from Peru “in-fill” isn’t enough? But what a bummer, such a great new trend wasted. ( Though I’d confidently predict it will be “put back in” Real Soon Now ;-)

Chile

With Argentina right next door serving as a good example, this is the best you can do?

Chile Hair Graph monthly anomalies and cumulative

Chile Hair Graph monthly anomalies and cumulative

So dead flat that I had to make the accumulated “change” Hot Pink (really light magenta) so it would show up. And with a trend line formula of: 0x -0.4

Now that is flat.

Then There Is Paraguay

Don’t really know what to do with that. Warmed, but long ago. Then cooled. But now it’s in a warming trend, but not more than it cooled before. Guess that’s just some oscillation thing going on. Interesting to look at, though, and not what CO2 Theory would predict.

Paraguay Hair Graph monthly anomalies and cumulative

Paraguay Hair Graph monthly anomalies and cumulative

And Those Southern Islands

Falkland Islands Hair Graph monthly anomalies and cumulative

Falkland Islands Hair Graph monthly anomalies and cumulative

Oh Dear! Seems to have been getting a mite cold down south… wonder if that’s why it suddenly ended in the 1980’s?. Wunderground has them at 39 F as I type… Still cold.

But don’t worry, looks like there is a “fix” in progress:

South Georgia Hair Graph monthly anomalies with cumulative

South Georgia Hair Graph monthly anomalies with cumulative

Notice that the “cumulative change” is divided into two segments. Notice further that there is a big chunk of ‘missing time’ in between them. So after 30 years of absence, just as the PDO and a couple of other ocean currents have swapped phase, we have the miraculous return to life of that thermometer… and with a fine warming trend too. (Notice that the “cold spikes” make it only to about 2.5 C to 3 C while the “warm spikes” are almost to 5 C, very odd.)

A splicing we will go, a splicing we will go
High Ho the dairy-O, a splicing we will go…

Advertisements

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.

19 Responses to South America – Hockey in the Jungle?

  1. j ferguson says:

    E.M.
    It looks like we are rapidly reaching the point where we have to figure out how GISTEMP is getting it wrong.

    It might be that “global temperature” is something that the information (temp time series) doesn’t lead comfortably to and that trying to get there with the data you’ve been examining requires too much effort for the data to survive – over cooked.

    But then i think you’ve said this.

    REPLY: [ Well, yes, it does sound familiar 8-)

    I present temperature averages (especially in some of the earlier work were I was trying to see just how much the data change would bias the results, what I call “characterizing the data” – really what my old FORTRAN professor called it a few decades ago) but there are “issues” with averaging them.

    It’s OK to average an intensive variable to find out about the structure of the data but it is a complete waste of time to do it to find out about what the data means.

    A simple example of this is the coins in your pocket. If you double the number of coins in your pocket you CAN say some things of value. You can say you have more money than before the doubling (provided none left…) and you can even say your pocket will be heavier. You can say it will take longer to count your money. But you can not say how much money you have. To do that, you must look at the denominations. (This becomes even more important if you take some coins out and then put some others back in… then you can’t even say that more coins means more money…) Similarly, dumping a pot of hot water into a pot of cold water tells you nothing about the final temperature. For that you need the masses too. And if some of the liquid is not water, then you need specific heats too.

    So right out the gate, thermal energy is an extensive variable (and what we want to know about) but we use temperature as a proxy for it (an intensive variable) and then we try to treat it as something we can “average” and have meaning about either heat or final temperature. And we just can’t. It is a simple fools errand.

    The temperature data, even if clean, would not tell us about heat. If averaged, it would tell even more lies.

    So I’m very careful to try to explain that when I average temperatures I’m looking for information about the structure of the temperature data set and NOT for information about the actual “Average Temperature” (for the simple reason that ANY average temperature of different things is no longer a temperature… and says less about heat content.) Yet, usually after such a disclaimer, I’ll “go with the flow” and talk about the Average Temperature of some place or another. For some reason “other folks” get confused by this and assert that I don’t know what I’m doing. Oh well…

    So, back to your point. Yes, the data we have is horridly ill suited to finding what is actually happening to the heat balance of the planet. Are we, net, gaining or losing thermal energy? Temperatures will not tell us. One example: You could melt 1,000,000 tons of snow absorbing great quantities of added heat and with a very large heat gain; and do it all with zero change of temperature… Going the other way, we could form 1,000,000 tons of snow and liberate vast quantities of heat (to be lost to space high in the clouds) cooling dramatically on a heat basis, but again with no change of temperature. The “heat of fusion” moves, but does so at one temperature… So it snowed a lot this year. Massive heat loss to space. But a few pounds of air over the ocean gets to ‘outvote it’ in the satellite measurements and folks scream that we’re “warming” when we are losing heat big time. And that’s what happens when you don’t keep track of what is an intensive property and what is an extensive property; and average your intensive variables…

    So somewhere in the bowels of GIStemp is some bit of code doing a really dumb thing inside the context of doing a dumber thing (Global “Average temperature” – the oxymoron) and it makes some very very bad results. (All those A / B blink charts showing homogenizing that makes cold trends warm and changes the past; and all those showing UHI “correction” that goes the wrong way.) And yes, I ought to get back to it someday to figure out exactly ‘what bits’ are messed up.

    But…

    All these charts I’m doing with dT/dt are looking at the INPUT to GIStemp. All these 1990 hockey sticks and 1980 pivot points are in the input data as it comes from NCDC “unadjusted”. Which means there is bigger game afoot than GIStemp. GIStemp is The Show and not the whole Trick. Something about those re-jiggerings of the data at the 1980 and more so the 1990 re-imaginings is horridly cooking the “raw” data. And that’s the Really Big Trick… and thats what really makes the ‘temperature time series’ useless for saying anything at all about actual trends…

    So, find what was changed in how the data were processed by NCDC in 1990 and you find the cause of spurious “global warming” that is not happening. My strongest suspicion right now is that they have something tossing out cold excursions that tosses out more of them than hot excursions. Or, in short, I think their QA processes is likely buggering the data. (Second possibility is that those places all might have installed some new automatic temperature gear, maybe ASOS? And the 1990 date was the ‘commit to electronic’ date. Seems a bit far fetched on a global basis, but hey, worth checking.)

    E.M.Smith ]

  2. P.G. Sharrow says:

    E.M.; you sure hit the nail on the head with the above. The satellite sees heat in the troposphere and we see snow on the ground. The AGW people see global warming and the dumb little people on the ground see a cold miserable winter.
    The Sun transmittes Energy. Temperature of the air 6 feet above the ground is a damn poor proxy measurement of the energy balance of the earth.
    At least you seem to be getting a good handle of the creation of the temperature records, both the real and the “Official”

  3. j ferguson says:

    What you say above is quite clear. Why not consider these things in terms of mass times temperature? Likely it would be very difficult and where would you stop – center of the earth?

    I suspect that you will find that the “heating” is not only in the way step functions in the raw data are dealt with. it will turn out to result from an unfortunate combination of handling step functions (do they ever step down?), station drop outs, uhi, and the changing (is it evolving) constituency of the thermometer population.

    It makes me wonder if some of the more complex, more difficult methods for putting numbers to the trend like say “total heat” were rejected in favor of the old thermometer because everyone understands – or at least thinks they understand it.

    I use term “total heat” loosely guessing it has a specific meaning which I might be splattering a bit.

    The thing that puzzles me most is that if things actually are warming up a bit, maybe a very little bit, why doesn’t it show in so many of the time series? If the warming isn’t happening, and we think it is, then we clearly don’t understand much about it after all.

    there’s a disconnect in here somewhere.

  4. Pingback: UFO storm « TWAWKI

  5. vjones says:

    E.M.,

    Interesting. I am off to compare these with some of my files to see what I can dredge up to add.

    There are good posts at the moment on anomaly calculation at tAV and RomanM’s blog.

    http://noconsensus.wordpress.com/2010/03/17/anomaly-aversion/

    http://statpad.wordpress.com/2010/03/18/anomaly-regression-%E2%80%93-do-it-right/

    Well worth a read, although my eyes tend to glaze over (unfortunately) at the code and maths.

  6. vjones says:

    E.M.,

    Um, I make it two files for Uruguay (313) – Riveras (86350) and Artigas (86330). Only Artigas (makes it into the adjusted dataset) – strong cooling trend, lots of missing months and years.

    The step change for Venezuela in 1931/32 is unexpected – other than that warm spike. Thermometer count is stable at that point.

  7. vjones says:

    Er, actually just found an additional 11 files for Uruguay.

    I was looking at the unadjusted data (downloaded ? early Jan) that does not make it into GISS adjusted when I spotted the other two. The 11 were in the GISS adjusted file. Nine have a strong warming trend; two are cooling overall.

  8. Ruhroh says:

    Hey,
    Do you guys know where that guy got the two versions of ‘raw’ data that he made all of the blinkers from?

    As I understood, he just downloaded the raw data at two different times, one a few months after the other, and the ‘raw’ data was astonishingly different.

    As I understood, it was the purportedly the same ‘raw’ data, with no annoation of why it might have changed so dramatically.

    My day job reverted to a day-and-night job again, so I’ve lost coherency on this topic.
    TIA
    RR

  9. E.M.Smith says:

    @Vjones:

    Thanks for digging into that. Looks like in a “drag and drop” from one machine to another I’d left Uruguay behind (then found it “nil”). On going back to check, found it sitting there…

    And a good thing, too. It has a very interesting new ‘feature’ in it (that I’ve showcased up top).

    OK, I’d said this was being done fast and there would be changes as things got reviewed…. That I’ve managed to knock out this many reports / graphs / postings at all is a bit surprising to me. Hopefully out of all that the Wandering Uruguay will be the only “anomaly” ;-)

    Kind of wondering if everyone is “stunned” by the quantity of it all or just hung over from St. Patty’s day ;-) There is a fair amount of traffic, but very few comments…

    Then again, when I first started looking at this data it was kind of a shocker to me. The consistent “lift” in 1980 and the consistent “pivot” in 1990 is NOT a CO2 signature, but is strongly correlated with when NCDC issue new “analysis and revisions” to the data sets… Just sayin’ …

    So, hope folks are enjoying what they are seeing and not just looking and puzzling…

  10. oldtimer says:

    All good stuff. By the way, this week`s The Economist has a lead article on Spin, Science and Climate Change plus a four page briefing called The Clouds of Unknowing. Among other things, it includes a chart of deviation from World Land Temperature 1850-2009. I will alert the editor to the evidence on your site.

  11. j fergson says:

    would it be possible to analyze the data and arrive at an “unusualness” figure for the “lift” and “pivot” across the data sets?

    This might be a bit like a seismic fault where all the roads come up to the fault on one side and leave it shifted 6 inches to the left on the other?

    I have no idea what forensic tools you might have but it doesn’t seem impossible that the sets could be looked at together and discovered to be similar in terms of randomness except all agree in 1980 and 1991.

    If not a gun, then at least the smoke. A mirror maybe?

    If it isn’t practical to do this, is there a term for the process I’m describing?

    REPLY: [ I started looking at “unusualness” tools. The first of them are the “Segment” graphs that show the change of trend at the decade discontinuity markers. When you start having 80% or so of the graphs with similar “discontinuities” it isn’t some weather oscillation…

    I started working one some new tools, too. That was when I wanted to look at the MIN and MAX separately for Canada and found that Really Bogus January in 2007? IIRC where the MAX was above zero and the MIN was just a degree or so below zero C yet the MEAN was -14 C IIRC. Just bogus. I did look at some other MAX and MIN reports (and really want to chase that down… but discipline… finish this stuff first, then chase the next “shiny thing” ;-) My SPECULATIVE first SUSPICION (have to put that in caps so the AGW True Believes can see it and not accuse me stupid conclusions when I’ve not concluded anything ;-) is that the MAX data look fine, but the MIN has been buggered. A Really Cool Thing would be to make a new MEAN from the MIN and MAX data sets and compare it to the GHCN MEAN. There WILL be differences, and I can just smell that it’s going to be a rich mineral vein to mine… just don’t know if it’s full of gold, silver, or just a bit of copper at this point…

    For now, the “Hair Graphs” give a nice indication of data buggering when the volatility of monthly ranges compresses. In the extreme cases it goes to zero (those “bullseye” moments when all the monthly data pass through the same point) on countries that have a 100% change of Duplicate Number (or Modification History Flag – same thing, different name) at a single point in time. So when we have a change of thermometers, we start a new series that has an initial “zero variance to itself” and the average of it with the prior thermometers compresses the range. The following years, though, also have compressed range (that gets more compressed as the old thermometer dies) so I’m fairly sure there is a “processing” difference that is tossing out data (probably extreme lows, thus the upward bias) or a dramatic difference in the character of the station itself (equipment change or location change). It’s possible that the equipment change itself has lead to ‘less lows’, but I think that would have been noticed, maybe…

    That is actually an interesting example of how forensics differs from “climate science”. I WANT to see that ‘artifact’ of the bullseye, it tells me where there is a “Dig HERE!!!” event. They want to erase it via further massaging of the process. I consider that a “hide of the interesting bits” they consider it a “valid correction” (and some times it may be, and if so I’ll eventually get there, but I’m not going to hide my clues as to where to look until I”m done looking ;-) It’s just a different goal. Like the auditor who does not just say “the dollars on the reports match” but looks in the vault and says “You have a receipt for $10,000 in quarters but the vault as $10,000 in dollar bills. SOMEBODY has been moving the money around. What’s the deal?” May be a clerical error. May be sloppy process that let someone swap dollar bills for quarters and not get recorded. Or may just be a Madoff Moment and someone stuck a bag of dollar bills in the vault for the day of the audit… so you dig until you KNOW.

    I don’t know of any Term Of Art for what you describe other than “consistency checks”. It’s looking for the “discontinuity” or the “OOPART” (Out Of Place ARTifact).

    So what I’ve observed so far is that the Duplicate Number (Mod flag) tends to be zero or 1 or sometimes even a 2 for the “good history” and runs up beyond that for the “Blade of the Stick”. Often a “3”, though some places (especially in Canada and New Zealand…) have a short series of runs then settle down with a higher Flag value like a “5” or “6”. Sort of like someone was running experiments until the worked out what would deliver the goods, then rolled it out to the rest of the world (where you get the “2” to “3” transition more consistently).

    So on the “To Do” list is to simply select the Duplicate Number (Mod Flag) 0, 1, and 2 records into one file, and anything larger into a second file, then compare the two trends. That ought to give a very dramatic “Delta Slope” and be a smoking Howitzer… Maybe I’ll do that tonight ;-)

    Part of the “fun” of doing this is that in a normal investigation you have to keep everything close ot the vest so you don’t spook the target and have them destroy evidence. But the evidence for me is the data, and I already have my copy. And changes to the data going forward just become MORE evidence. Gotta love it 8-)

    Oh, and as we’ve seen, email archives tend not to be erasable… That whole FOIA / Sarbox thing…

    Or, put more prosaically: “You can run, and hide, but that’s just more evidence.”

    So a dramatic inconsistency at the decade markers when NOAA “recomputes the data” shows in the data. And when that signal swamps all others (as it does in these “by Country” graphs) you know when, and what agency. That then points to who, and how (that would take looking inside the published and private records of that change: Change of QA, or change to ASOS or?). The only really hard part is motivation, the “why”. It’s very hard to sort out “malice” from “stupidity”; but then again, I don’t really care which it is…
    -E.M.Smith ]

  12. Watchman says:

    On the Falklands, the dropping in 1982/3 might be because there was a little bit of argy-bargy (sorry) in the vicinity that year. War can be so inconvenient for temperature measurements.

    REPLY:[ Most likely. I usually try to first just present “The Facts” and then come back for the speculative parts. FWIW, one of THE major benefits of doing the “Temperature Average” series ( that “climate scientists” don’t like because it isn’t they way they see the world) is that is shows you what the data says directly. And one of the biggest things it pointed out was just how much “war” and “revolution” are present in the data. A far stronger signal than CO2 could ever generate. You get the cold 1940’s spike in the Pacific as the hot islands went off line. You get a massive drop out in Indonesia when they “had issues” in about 1990? somewhere around there. Siberia and the end of the cold war. Latin American stations that ‘start time’ after their revolutions were over. Across the board, the “temperatures of the planet” tell you about social disorder more than they tell you about “global warming”. So unless you find a way to take that “Social Unrest Bias” out of the data, you will “have issues” with finding a valid signal of “global warming”… Yet actually looking at the data to see “What signal is in here? And what signal not so much?” seems to just cause traditional “climate scientists” to have a conniption. But it’s a standard forensic approach. Bring nothing to the data, just ask it what it has to say. Then follow the trails, both large and small… -E.M.Smith ]

  13. vjones says:

    Hmm, good catch on Uruguay.

    Please would you check French Guiana again? I find only 4 stations (plus 4 series for Cayenne/Rocha) but no data from ~1910-1950. (not sure if the following html will work properly.):

    Saint Georges   3.9 N 51.8 W 315814080000 rural area 1961 – 2010

    Cayenne/Rocha   4.8 N 52.4 W 315814050000 37,000 1891 – 1991

    Cayenne/Rocha   4.8 N 52.4 W 315814050001 37,000 1956 – 1990

    Cayenne/Rocha   4.8 N 52.4 W 315814050002 37,000 1971 – 1980

    Cayenne/Rocha   4.8 N 52.4 W 315814050003 37,000 1987 – 2010

    Kourou   5.2 N 52.8 W 315814030000 rural area 1971 – 1980

    Maripasoula   3.6 N 54.0 W 315814150000 rural area 1961 – 2010

    Saint-Laurent   5.5 N 54.0 W 315814010000 rural area 1961 – 2010

  14. vjones says:

    Do check French Guiana (as per comment in moderation) – also looked in my downloaded file of GHCN v2.mean, but you might want to snip part of my comment.

  15. j fergson says:

    E.M.

    Looking at what in radio amateur talk is called a waterfall display where you see a fairly broad width of the band you are receiving and can see the signals here and there as vertical lines, maybe like a spectrograph (if I have the word right).

    In this case you would be seeing the signals clearly midst the noise. The temps aren’t noise but the steps and pivots are not temperature signals, but they are signals, just not temperatures.

  16. j fergson says:

    I see first sentence in above post isn’t .

    What I was trying to suggest was that step and abrupt inflexion points or pivots are signals but different from temperatures and would display well on a spectrograph.

  17. e.m.smith says:

    @vjones:

    When I look at the French Guiana graph I see a jump from 1909 to 1951 so I think it’s the right stuff. ( I probably need to emphasize that gap in the text). Also note that the thermometer count is marked as “Count / 2” to scale it to fit on the page. So that ending “2” is 4 thermometers. ’71-80 has 5 but that’s probably just multiple Duplicate Numbers (Mod Flags) for some places as they run in parallel.

  18. vjones says:

    OK. I was just puzzled to see a line in the period 1910-1950. Although there is one active thermometer, there are no actual temperatures in my download of the GHCN v2.mean and GISS station data has only 9999 missing months for that period. I just wanted to make sure you hadn’t found some thermometer I had missed.

    REPLY: [ Yeah, it can be a bit “odd”, but I don’t correct or remove or fill in missing data. A block of years all full of missing data flags will show up on the chart (though as a place of zero volatility – those ‘bullseye’ points where all the monthly ranges head to zero). The dT/dt method was designed to ‘span’ the dropouts but not to delete them. So it carries a value from a prior month as the ‘last month’ until it finds a future valid value for that month no matter how many years it takes (why this graph is not discontinuous at that long break, but takes a big movement). My purpose was to not “correct” that out, but to observe what, if anything, it might have to say. Yeah, that “forensics” mindset as opposed to the “climate science” one… -E.M.Smith ]

  19. A C Osborn says:

    E M, I am also surprised by the lack of Comments, if the other posters are like me they are just too busy reading your comments and Graphs as well as being slightly stunned.
    I think this is wonderful work and I just hope that you get recognition for it.

Comments are closed.