Globe with Atlantic Ocean Currents
As the Thermometers March South, We Find Warmth
Ask any retired person which way to get warmer. They will tell you it’s to head to the tropics.
Well, I couldn’t sleep until I found out if my “eyeball” look at the increase in lower / middle latitude thermometers by year was right. It was.
So what AGW has found is that if you put more thermometers in the tropics you get more warming. Who knew? (Just about everyone retired…) I first discovered this trend in some steps detailed here:
https://chiefio.wordpress.com/2009/08/05/agw-is-a-thermometer-count-artifact/
This chart is the average “thermometer years by latitude band” for each decade of the entire temperature record in GHCN. I took the whole thing, and for each record in each year (a record being one thermometer with temperature averages in each of 12 months – so one “thermometer year”) looked up that thermometers latitude. I divided the world into below Latitude -70 (down toward Antarctica) above Latitude 70 (toward the Arctic) and 20 degree bands in between (-70 to less than -50, -50 to less than -30, -30 to less than -10, -10 to less than 10 or the equatorial band, 10 to less than 30, 30 to less than 50, 50 to less than 70).
So basically you can think of the left side of this table as the south pole, the right side as the north pole, and you can watch the thermometer record start in the North and march south.
The March of the Thermometers
Year, and 20 degree latitude bands, south to north. Thermometer years.
SP – South Pole
SC – Southern Cold
ST – Southern Temperate
SW – Southern Warm
EQ – Equator
NW – Northern Warm
NT – Northern Temperate
NC – Northern Cold
NP – North Pole
SP SC ST SW EQ NW NT NC NP
DecadeLat: 1709 0 0 0 0 0 0 0 1 0
DecadeLat: 1719 0 0 0 0 0 0 0 1 0
DecadeLat: 1729 0 0 0 0 0 0 0 1 0
DecadeLat: 1739 0 0 0 0 0 0 0 2 0
DecadeLat: 1749 0 0 0 0 0 0 1 3 0
DecadeLat: 1759 0 0 0 0 0 0 3 6 0
DecadeLat: 1769 0 0 0 0 0 0 6 10 0
DecadeLat: 1779 0 0 0 0 0 0 9 14 0
DecadeLat: 1789 0 0 0 0 0 0 16 16 0
DecadeLat: 1799 0 0 0 0 0 0 19 16 0
DecadeLat: 1809 0 0 0 0 0 1 24 20 0
DecadeLat: 1819 0 0 0 0 0 1 32 28 0
DecadeLat: 1829 0 0 0 0 0 2 54 48 0
DecadeLat: 1839 0 0 0 0 0 4 74 72 0
DecadeLat: 1849 0 0 1 0 2 6 93 82 1
DecadeLat: 1859 0 0 3 0 2 11 137 92 2
DecadeLat: 1869 0 0 15 0 3 7 173 103 1
DecadeLat: 1879 0 0 27 2 15 20 336 110 2
DecadeLat: 1889 0 0 44 10 18 48 624 184 3
DecadeLat: 1899 0 2 57 26 31 87 1175 309 3
DecadeLat: 1909 0 9 111 61 44 133 1510 382 5
DecadeLat: 1919 0 11 174 124 57 160 1789 479 8
DecadeLat: 1929 0 11 187 145 66 212 1961 545 16
DecadeLat: 1939 0 13 220 180 91 304 2156 713 26
DecadeLat: 1949 0 20 261 259 116 407 2412 887 37
DecadeLat: 1959 9 43 347 453 421 1010 3417 1249 80
DecadeLat: 1969 32 68 466 650 729 1310 4121 1511 105
DecadeLat: 1979 34 85 580 747 661 1269 4204 1511 103
DecadeLat: 1989 25 68 495 605 452 916 3805 1307 82
DecadeLat: 1999 9 32 212 250 224 429 2128 314 27
DecadeLat: 2009 7 20 102 132 159 316 1339 241 17
Fascinating little chart, isn’t it? AGW proceeds at a pace directly correlated with the southern march of the thermometers…
Now the later steps of GIStemp may try valiantly to remove this fundamental bias in the recorded history of thermometers, but it’s just going to get swamped. Yeah, I need to prove it in annoying detail, and I will, but this is just amazing to see. Especially when you remember that we showed that the stable thermometer records show NO warming… (I’ve already demonstrated that the ‘temperature steps’ of GIStemp are an amplifier, so they will make this worse, not better)
https://chiefio.wordpress.com/2009/08/13/gistemp-quartiles-of-age-bolus-of-heat/
Stir in a few “rural” thermometers used to “correct” for an Urban Heat Island effect that are tagged as “cool crops” or “conifer forest” but are really hot thermometers over tarmac at an airport with jet exhaust and I think we have a pretty clear picture what’s going on. (See the bottom quartile thermometer posting if that sentence doesn’t mean anything to you… where the temperature record shows London / Gatwick as “crops”…)
Same thing, as Percentages
Some folks like percentages more than counts. OK, here are the percentage of “thermometer years” (basically, the percentage of the temperature records) for each decade, by latitude band. They are again labled with S for south, N for north, EQ for equator, W for warm, C for cold, and T for temperate
S.P S.C S.T S.W EQ N.W N.T N.C N.P
DecLatPct: 1709 0.0 0.0 0.0 0.0 0.0 0.0 0.0 100.0 0.0
DecLatPct: 1719 0.0 0.0 0.0 0.0 0.0 0.0 0.0 100.0 0.0
DecLatPct: 1729 0.0 0.0 0.0 0.0 0.0 0.0 0.0 100.0 0.0
DecLatPct: 1739 0.0 0.0 0.0 0.0 0.0 0.0 0.0 100.0 0.0
DecLatPct: 1749 0.0 0.0 0.0 0.0 0.0 0.0 17.5 82.5 0.0
DecLatPct: 1759 0.0 0.0 0.0 0.0 0.0 0.0 37.0 63.0 0.0
DecLatPct: 1769 0.0 0.0 0.0 0.0 0.0 0.0 38.9 61.1 0.0
DecLatPct: 1779 0.0 0.0 0.0 0.0 0.0 0.0 39.8 60.2 0.0
DecLatPct: 1789 0.0 0.0 0.0 0.0 0.0 0.0 49.7 50.3 0.0
DecLatPct: 1799 0.0 0.0 0.0 0.0 0.0 1.1 54.6 44.2 0.0
DecLatPct: 1809 0.0 0.0 0.0 0.0 0.0 1.8 54.2 44.0 0.0
DecLatPct: 1819 0.0 0.0 0.0 0.0 0.0 1.8 52.3 45.9 0.0
DecLatPct: 1829 0.0 0.0 0.0 0.0 0.1 1.8 52.2 45.8 0.1
DecLatPct: 1839 0.0 0.0 0.0 0.0 0.3 2.5 49.5 47.7 0.1
DecLatPct: 1849 0.0 0.0 0.5 0.0 1.0 3.1 50.3 44.5 0.6
DecLatPct: 1859 0.0 0.0 1.0 0.0 0.9 4.4 55.6 37.3 0.8
DecLatPct: 1869 0.0 0.0 5.0 0.0 1.0 2.2 57.2 34.1 0.5
DecLatPct: 1879 0.0 0.0 5.2 0.3 2.8 4.0 65.8 21.5 0.4
DecLatPct: 1889 0.0 0.0 4.7 1.1 2.0 5.2 67.0 19.8 0.3
DecLatPct: 1899 0.0 0.1 3.4 1.5 1.8 5.1 69.6 18.3 0.2
DecLatPct: 1909 0.0 0.4 4.9 2.7 1.9 5.9 66.9 17.0 0.2
DecLatPct: 1919 0.0 0.4 6.2 4.4 2.0 5.7 63.8 17.1 0.3
DecLatPct: 1929 0.0 0.3 6.0 4.6 2.1 6.7 62.4 17.3 0.5
DecLatPct: 1939 0.0 0.4 5.9 4.9 2.4 8.2 58.2 19.3 0.7
DecLatPct: 1949 0.0 0.5 5.9 5.9 2.6 9.3 54.8 20.2 0.8
DecLatPct: 1959 0.1 0.6 4.9 6.4 6.0 14.4 48.6 17.8 1.1
DecLatPct: 1969 0.4 0.8 5.2 7.2 8.1 14.6 45.8 16.8 1.2
DecLatPct: 1979 0.4 0.9 6.3 8.1 7.2 13.8 45.7 16.4 1.1
DecLatPct: 1989 0.3 0.9 6.4 7.8 5.8 11.8 49.1 16.8 1.1
DecLatPct: 1999 0.2 0.9 5.9 6.9 6.2 11.8 58.7 8.7 0.7
DecLatPct: 2009 0.3 0.8 4.4 5.7 6.8 13.6 57.4 10.3 0.7
Just to put a bit of a finer point on it, if, excluding the poles due to very poor coverage, we add up the “warm areas” of SW, EQ, NW and add up the cold areas (SC, ST, NT, NC) we get the following, by 1/2 century:
YEAR Warm Cold 1839 2.8 97.2 1889 8.3 91.5 1939 15.5 83.8 1989 25.4 73.2
Leaving aside the question of just how do you make a “Global Average Temperature” for comparisons from 1839 or even 1889 with 91.5% of thermometers in the cold north and only 8.3% in the 60 degree band of the planet from near Cairo to Cape Town South Africa ….
We are still left with the fact that we move about 1/4 of the thermometer records from the cold places to the hot places. That not only increases the impact of the hot places, but reduces the impact of the cold places.
Given that the cold places are a physically smaller area, if we “area adjust” our thermometer records by making zones and boxes for our geographical bands, we will even further reduce the impact of the “cold thermometers” on the Global Average Temperature and increase the impact of the “warm thermometers”.
Then there is the problem of the southern ocean.
Even at it’s peak, a decade ago, we had less than 1% of thermometers in the Southern Cold band (not surprising, it’s mostly water). Given that we know the oceans are vitally important to planetary weather, and that the oceans oscillate in temperature with hot and cold zones changing places, we can have little faith that the southern oceans are at all represented properly in the “Global” average temperature. And certainly were not at the start of the GIStemp interval (1880) when we had ZERO in the S.C. band. So exactly what quality is there to the GIStemp “baseline”?
These data do not include the direct antarctic records. I’ll do a similar analysis at a future point with them added, but averaging a dozen or so thermometers in a consistently cold place into a few thousand ought not to change things much…
IMHO, you can zone, grid, box, interpolate, and average all you want and you will still not be able to “correct” these data to accurately show the temperature trends of the planet. The holes are just too large (both in geography and in time). Far better is to simply look at the set of stable thermometers (that we DO have) from the last 150 years. And they show no warming.
There is no global warming. It is a computer fantasy based on The March of the Thermometers.
Update: The GIStemp Zones
GIStemp uses 30 degree zones rather than the 20 degree zones I used here. So what does that percentage table look like done “the GIStemp way”?
[chiefio@tubularbells vetted]$ cat Lpct30.decades DecLatPct: 1709 0.0 0.0 0.0 0.0 100.0 0.0 DecLatPct: 1719 0.0 0.0 0.0 0.0 100.0 0.0 DecLatPct: 1729 0.0 0.0 0.0 0.0 100.0 0.0 DecLatPct: 1739 0.0 0.0 0.0 0.0 100.0 0.0 DecLatPct: 1749 0.0 0.0 0.0 0.0 100.0 0.0 DecLatPct: 1759 0.0 0.0 0.0 0.0 89.1 10.9 DecLatPct: 1769 0.0 0.0 0.0 0.0 93.2 6.8 DecLatPct: 1779 0.0 0.0 0.0 0.0 93.9 6.1 DecLatPct: 1789 0.0 0.0 0.0 0.0 94.4 5.6 DecLatPct: 1799 0.0 0.0 0.0 1.1 96.1 2.8 DecLatPct: 1809 0.0 0.0 0.0 1.8 91.7 6.5 DecLatPct: 1819 0.0 0.0 0.0 1.8 89.8 8.4 DecLatPct: 1829 0.0 0.0 0.0 1.9 89.8 8.2 DecLatPct: 1839 0.0 0.0 0.0 2.7 90.1 7.1 DecLatPct: 1849 0.0 0.5 0.0 4.1 89.9 5.5 DecLatPct: 1859 0.0 1.0 0.1 5.2 89.2 4.5 DecLatPct: 1869 0.0 5.0 0.3 2.9 86.1 5.7 DecLatPct: 1879 0.0 5.2 0.8 6.4 82.5 5.1 DecLatPct: 1889 0.0 4.8 1.3 6.9 82.4 4.7 DecLatPct: 1899 0.0 3.5 2.0 6.4 83.8 4.2 DecLatPct: 1909 0.1 5.3 3.4 7.1 79.8 4.3 DecLatPct: 1919 0.1 6.5 5.1 7.1 76.8 4.4 DecLatPct: 1929 0.1 6.2 5.2 8.2 75.5 4.8 DecLatPct: 1939 0.1 6.2 5.7 9.8 72.0 6.2 DecLatPct: 1949 0.1 6.2 6.9 10.9 68.7 7.1 DecLatPct: 1959 0.5 5.2 8.6 18.2 60.6 6.9 DecLatPct: 1969 0.8 5.5 10.9 19.0 57.2 6.6 DecLatPct: 1979 1.0 6.6 11.1 18.0 57.0 6.3 DecLatPct: 1989 0.9 6.7 10.4 15.0 60.5 6.5 DecLatPct: 1999 0.9 6.1 9.3 15.6 64.0 4.1 DecLatPct: 2009 0.9 4.6 8.2 17.8 64.0 4.5
Not nearly so interesting. One is left to wonder if the “30 degree zones” was a cherry pick. The starting single thermometer is no longer in a northern cold zone, but now in the only non-arctic non-equator northern zone. We add some Northern thermometers that end up in the “arctic” band, and with a few wobbles, stay mostly in the 4-8% range. A much more “represented” and less volatile Arctic. The exodus of thermometers from Siberia to Italy and points south disappears into The One Northern Temperate Band.
Moving to the Southern hemisphere, we still have nearly nothing down by the pole. That will undoubtedly be “fixed” by the addition of the Antarctic stations (that I’ll do next) and the Southern Temperate band has a fairly stable 4-6% representation over the life of the records GIStemp keeps (1880+).
That just leaves the (astoundingly wide -30 to +30) tropical band to deal with. It does rise from nothing in the beginning, to about 8% when GIStemp chooses to begin time. Then it ramps up to the 30% range before a drift back down to 24%. This while the Northern Temperate band drops from 100% to 64%. We still see shadows of The March Of The Thermometers, but it is significantly muted. All the migration within the northern hemisphere is hidden. The southern end is devoid of interest, and the only significant artifact is a movement from temperate north to tropical central. Easily handwaved away with an assertion that Grids and Boxes will “fix it up real nice!”…
While not exactly a “smoking gun”, it is clearly “bad form”. Far too much reality is hidden, and the temptation to believe that the data are stable is enhanced too much. The dramatic drop in N.H. Cold zone stations and the rise in N.H. Temperate stations being masked is, all by itself, enough to give pause. The appearance of better N. polar coverage than is warranted also causes a small concern. The masking of the truly dismal S.H. Cold Zone coverage also gives me some pause. Would we ever know from looking at GIStemp zones that the band from -70 to -50 is substantially devoid of thermometers? And what will happen when we use the Reference Station Method to fill in those empty (undoubtedly cold) boxes with fictional temperatures from 1000 or 1500 km north in the warmer band? (There not being many thermometers to the South, either…)
I’m left to conclude that the choice of 30 degree zones is a cherry pick. Perhaps an accidental one. One chosen due to the data in the S.H. being so sparse that a smaller zone was looked at and discarded due to sparse coverage (speculation on my part, but I could see why that might happen). One where the consequences (especially in the context of repeated applications of The Reference Station Method slowly spreading temperatures out to where there are none) of adding fictional warmth to the Southern Cold zone and the Northern Cold zone might have been accidentally overlooked.
But be this a harvested cherry or not, and be it a deliberate act or an accidental one, one fact is clear: 30 degree zones hide too much information about The March Of The Thermometers to warmer lands.
30 degree zones will give a biased view of both thermometer stability and of temperature changes over time. And 30 degree bands hide more than they reveal. IMHO, the 30 degree zone choice is a large “Dig Here!” flag.
Musings from the Chiefio 
E.M., I linked to this post on my sowellslawblog, as your results are fascinating and ought to be spread far and wide. I hope others will do the same.
“Even at it’s peak, a decade ago, we had less than 1% of thermometers in the Southern Cold band (not surprising, it’s mostly water). Given that we know the oceans are vitally important to planetary weather, and that the oceans oscillate in temperature with hot and cold zones changing places, we can have little faith that the southern oceans are at all represented properly in the “Global” average temperature. And certainly were not at the start of the GIStemp interval (1880) when we had ZERO in the S.C. band. So exactly what quality is there to the GIStemp “baseline”?
These data do not include the direct antarctic records. I’ll do a similar analysis at a future point with them added, but averaging a dozen or so thermometers in a consistently cold place into a few thousand ought not to change things much…”
Good discussion, E.M., and while I was reading the part I pulled out here (e-e-e-vil grin!), I thought of G. Schmidt’s statement that he only needs 60 thermometers to determine the GAT. Which 60 would he pick from the above I wonder?
The big problem I see with the “60 is enough” idea is that of a cutoff bandpass. You are basically doing a filter of sorts that will have a cutoff resolution based on “pixels” of temperature.
Now the actual thermometer is a “pixel” in the IR range of about 1 M square (the Stevenson Screen box and air right around it. Everything around it you hope is somewhat related.
The hope is that the siting guidelines make the “pixel” representative of what the surface would be were it devoid of planes, trains, autos, tarmac, houses, tar roofs, … but we know from the suracestations project that that is often just a vain hope.
Further, we have spots where snow is melting into a creek with a sunny rock next to it. We hope that the air temperature a meter or so above the surface does a good job of integrating those different temperatures (but we know from the ocean data that the water does not match the air…)
So take a block 100 miles on a side. It might (using real example from near where I live) have everything from a 50 F fog over cold sea water, to a 65 F urban under fog, to a 105F sunny grassy area to a 115F tarmac at the airport. All at the same time. (SF under fog and SJC in sun, on the same day.)
So just exactly where do you put your “pixel” to capture that IR image? To capture the integral of it?
That, BTW, is one of my major “issues” with the idea of a Global Average Temperature… I dove into a creek one hot summer day only to get an instant splitting headache. Just around the bend, shaded snow was melting into a 32F creek. I was standing in direct sun on sun heated rock at 85+F at least. So just what WAS the temperature there?… And what the average? And if you can’t get a decent answer for how to get an “average for this 100 square meters” how do you get one when you add in the mountain near by (6000 ft up with snow) and the valley 10 miles below (100F in the shade, and there’s no shade). BTW, the place is near Quincy California and the event was real.
So take a globe and look at it real hard. What 60 pixels tell you the ‘average surface temperature’?
The altitudes are fractal (mountains erode as a fractal). The shorelines are fractal. The surface ‘finish’ is checker boarded in all sorts of ways. And it’s dotted with dozens of point heat sources and both point and line cold spots (lakes, rivers, snowbanks, etc.). Then add in some clouds drifting by.
So we have an instrument (thermometer) with both a “small size” and a “large size” cutoff of about 1 M squared. We’re trying to measure a system with a grid size of about 1000 M on a side minimum, to 1000 km maximum. And nature has features that range from a few cm (black stone) to 1000 km (desert sand, snow fields, etc.) in scale. We simply have a dramatic “miss match” of the bandpass of the instrument and the things being measured.
I can see no way to make sense of the idea of a “global average temperature”… (But I can use it as a metastatistic for characterizing GIStemp…)
“So take a globe and look at it real hard. What 60 pixels tell you the ‘average surface temperature’?”
100% agreement with ya, E.M., thus the evil grin when imagining G.S’s [nonsensical] choices.
Personally, I like the geologic ‘temperature’ record. When you see that sea levels have been 100m higher or 100m lower, now that’s REAL climate change. Nope, we don’t know what the temperature was to the hundredth degree, but which is the more useful information; the temperature changed enough for 100m of sea level change even though we have to make a rough guess about what the exact temperature was, or that the temperature is 0.016C warmer this month than the same month a year ago?
You’ve got the sun, the earth, the ocean, and the atmosphere for starters. IMO the focus should be on developing the big picture theories (Milankovitch + plate tectonics, e.g.) using the main players before trying to get into the itty-bitty details. There is no magic bullit to be found in the little details. (Just my $0.02.)
Oh. Weather is another story. IMO, you can’t get detailed enough in studying the weather. We have a long way to go just to understand the weather, and it’s the most important thing to people on a day-to-day basis. (OK. I’m up to $0.04 on this thread.)
Oh heck, go for a whole Nickel!
Yeah, the sea level moving 100 m is a big deal. A fictional 0.02C is not very useful… And the ocean tells us that we’ve had vastly larger movements in temperature with no human involvement at all…
And you are absolutely right. We ought to be looking at rocks from space, volcanic cycles, what causes the periodic Bond Events, etc. Not fooling around with some CO2 fantasy.
Ah, but since the polar regions are warming at an “alarming rate” according to the warmists, any delving into those regions will torpedo your future polar analysis. (Snicker)
On a more serious note, combine the problems inherent in measuring global temperature with splicing multiple proxies in order to demonstrate geological timescale references and you have an instant recipe for misrepresentation.
This junk science, the public’s wide-spread inumeracy and scientific ignorance combined with willfully ignorant politicians like Henry Waxman brought us to a dangerous “tipping point” in public policy. Hopefully we’re not to late to derail this train.
Basically, by “listening to the data” you’ve managed to not only cast doubt on the “A” in AGW, but the “GW” as well. If we can’t accurately measure temperature today, then arguments about tenths of a degree in the past are certainly meaningless (discussion of significant digits notwithstanding)
Great work E.M, I too linked to your work on my blog.
@JKKrueger
Thanks!
Take a look at the update here. First I built the benchmark – that lead to The March Of The Thermometers insights about the “raw” data. Now I’m applying it. GIStemp is not doing well.
One of the things about benchmarks is that you want them to measure outside the ranges of the thing being evaluated. You want information in greater depth than the “black box” covers… because that lets you see the “edge cases” and the places where the device performs as advertised, but did not address the reality.
With GIStemp, the 30 degree zones it applies look to me like either a deliberate “cherry pick” or a terrible error. Not a very pretty choice of options. It hides too much of the reality in the data that we saw in the benchmark.
If there were just one thing that I’d change in the world of Academic-Industrial Science, it would be the way they approach the thesis. The model presented as “right” is at it’s core: Make a thesis and then torture the data trying to prove it is true. IFF you are so unable to torture it enough to support your thesis, then move on to another try.
There is very little room left for what I like to think of as exploratory science. Get on a ship, visit an island with little preconceived notion of what you will find, gather data, and then characterize the data and ask it what story it has to tell- THEN make a thesis and go testing it. That whole front end part (that to me is the heart and soul of science) is left out of the common approach today. Today we start with the thesis, then go hunting for data to support it.
So I suspect that the GIStemp folks had a thesis about zones, boxes, The Reference Station Method et. al. (all the things they have published papers about and lauding) and didn’t bother with that old fashioned step of first letting the data speak for themselves… of “exploring the data”… No Ph.D thesis in that… No peer reviewed paper from saying “I looked at the data”… No Nobel Prize for “The data said foo.” Lets just skip on to “the fun part” where I puff out my chest and tell you how brilliant and worthy My Theory is and what data I found (or created…) to support it.
Oh Well.
On the virtue of benchmarks beyond ‘spec’:
I once worked on the Bradley Fighting Vehicle. (No big, I did the database for the cost accounting system… kind of dull…)
When the “acceptance test” day came, the vehicle was shot with all the “specification” ammo and passed Just Fine. Papers were signed, champaign flowed…
AND THEN, while it was on the field and guns were loaded, they let the guys test it “beyond spec”. It was shot up with ammo way larger than “spec” to find out where it would really break. I won’t disclose “with what” for obvious reasons.
The result was that the Bradley was in a garage in the main engineering area, shot to hell. We were officially told that we were not allowed to look… “even if the door to room foo is open and there is no one checking clearance levels at the door between 10 and 1:30” (or some such). So we all went and took a peek.
The result was a lot of drawn faces and a lot of muttered “people would have died in that”… Shortly afterward, there were a few dozen “enhancements” and “upgrades” proposed to the crew protection facilities and the armor… (Since we all had friends and family or knew someone with friends and family in the military.)
Now, decades later, it’s a pretty nice fighting vehicle. But if we had only used a benchmark that measured “to spec” we would have learned nothing.
What I learned was “Always have your benchmark exceed spec at some point in the testing.”
A parody of this event was made into a movie with Kelsey Grammer titled “The Pentagon Wars”. Great movie and very funny; but they have the facts backassward. There is a scene where the vehicle is shot with “substandard” RPGs and passes, but blows up with Spec ammo. Right idea, wrong specs…
The Bradley took the full spec ammo that was specified (don’t know if that included RPGs then, it was classified). It was some “over spec” special stuff that was the problem. But at any rate, it’s a fun movie and did capture some of the odd bits (in extreme parody) of working in a Pentagon driven program. I would ride into battle in one and not be worried about my “ride”. (And I’d be really glad it had the anti-tank TOW on it… it’s a little tank-ette that takes down big tanks 8-)
And I’d be especially happy that 30 years or so ago somebody shot it up with “over spec” benchmark ammo… to see where it really breaks.
“I was standing in direct sun on sun heated rock at 85+F at least. So just what WAS the temperature there?”
That is why they use anomolies, not absolute temperature. If all you measure is the changes at the same site, then you remove a lot of the issues you mention.
The average temperature doesn’t mean a lot, it’s just a convenient way of saying what the big picture is, like looking at the airspeed of jet. There’s a lot more instruments to look at, with more detailed and varied information, but airspeed by itself is useful to get a simple look at what is happening.
Fascinating analysis.
I’d be careful about accusations of cherry picking. 30 degree bands seems an innocuously reasonable choice for 90 degree hemispheres.
If I were the one doing the bucketing, I might have chosen the major circles of latitude as the dividing lines – +67, +23, 0, -23, -67. And then I would have missed the trend you’ve identified as well, without any intention of obfuscating or cherry picking.
EM Smith,
Fascinating analyses you are performing. The statement that the “thermometers” with the longest records show no warming is a fundementally strong statement.
I can appreciate the southern migration of the thermometers as well, a clear problem. (Perhaps the reason for the dropout in stations is due to thermometers jumping off cliffs during their migration, like lemmings?)
But seriously – if the AGW doctrine is that the specific warming since about 1979 can’t be explained without CO2 / GHG theory while previous warming can – then this particular line of argument of thermometers migrating south wouldn’t seem to be as strong because it looks as if most of the migration took place before this period.
Or have I missed something that would make you think otherwise?
Thanks, and keep up the fascinating work!
@Dan R.
“I can appreciate the southern migration of the thermometers as well, a clear problem. (Perhaps the reason for the dropout in stations is due to thermometers jumping off cliffs during their migration, like lemmings?)”
LOL! Very good imagery.
“But seriously – if the AGW doctrine is that the specific warming since about 1979 can’t be explained without CO2 / GHG theory while previous warming can – then this particular line of argument of thermometers migrating south wouldn’t seem to be as strong because it looks as if most of the migration took place before this period”
I was kind of stuck on something like that too, but then I realized upon a closer look that NH thermometers were dropping out like crazy during the period while the number of SH thermometers seemed to remain (relatively) more stable, eh?
amouse
“I was standing in direct sun on sun heated rock at 85+F at least. So just what WAS the temperature there?”
That is why they use anomolies, not absolute temperature. If all you measure is the changes at the same site, then you remove a lot of the issues you mention.
Nope, not a one is removed. Just moved one layer further into the muck.
So, you want to play with anomalies. OK, then just rephrase the question as “So just what WAS the temperature anomaly there?”.
The basic problem is that if you don’t know the temperature (AND the historical temperature) you can’t know the “anomaly of the temperature”. If it is not possible to decide the temperature in the “(86 F to 32 F) place it is BY DEFINITION not possible to know the anomaly in the temperature.
The average temperature doesn’t mean a lot,
Put a period after that and you’ve got something.
it’s just a convenient way of saying what the big picture is, like looking at the airspeed of jet. There’s a lot more instruments to look at, with more detailed and varied information, but airspeed by itself is useful to get a simple look at what is happening.
Very broken analogy. An airspeed gauge is one of the most direct and reliable things you’ve got. Simple direct measure of what the air is doing relative to the pitot tube. Might need a bit of correction for barometric pressure and temperature, but those are typically available. It’s not an “average” of a bunch of highly variable things.
@DanR
From 1979 or so on, we still have a strong drop in the N. COLD band thermometer count. Peaking right when “global temperature” peaked in 1998. Since then, the count went up (as temperatures “dropped a little”).
@H.R.
You got it! There are two things going on. More in the warm band (up to a point) then uneven “pruning” in the later years.
Oh, and technically these are “thermometer records” rather than “thermometers” proper. A single instrument might have 2 records with 2 different modification histories… I know, picky picky picky…
One other minor point:
As you get into smaller and smaller time bands, like ‘1979 -1999’, you get higher probabilities of picking up natural cycles like the PDO and ENSO. We had a warming PDO from mid-70s to, well, now. Now we’re cooling off. How much of the “warming” in the ’80s and ’90s was PDO and how much was changing where you stick the thermometer? I don’t know, but it’s pretty darned clear that you don’t need CO2 involved. (For that matter, the “stable records” also say not much warming happened then. I’d look to them for guidance.)
E.M.:
“You got it! There are two things going on. More in the warm band (up to a point) then uneven “pruning” in the later years.”
I forgot for a moment how this fits in with Anthony’s surface station project. It seems a lot of the truly rural stations in the U.S. are getting dropped and the UHI affected or badly sited stations are being kept.
Duncan I’d be careful about accusations of cherry picking. 30 degree bands seems an innocuously reasonable choice for 90 degree hemispheres.
That is why I said “accidental” as an option. A “cherry pick” does not need to be malicious nor deliberate, it is just jargon / shorthand for “picking a starting point or selecting the data / method such that your argument is bolstered”.
I would have missed the trend you’ve identified as well, without any intention of obfuscating or cherry picking.
Note that a “cherry pick” does not require intention! If I were researching car safety and used German statistics that showed large cars significantly safer than small cars, I would be at risk of an accidental “cherry pick” due to the large proportion of Mercedes Benz cars with very early adoption of sometimes quite expensive safety measures (seat belts, head rests, crumple zones, safety cage design, airbags…) not to mention that when a car cost $80,000 and the competition is $20,000 you can put more engineering into the design.
This is something that researchers must guard against at all times. So a German car safety researcher ought to gather Italian, US and Japanese data as well just to be sure that the data are not biased in some way by an “accidental cherry pick”.
When designing a “benchmark” for stress testing something or some code, it is good practice to deliberately pick ranges and sources of data not used by the designer for just this sort of reason. You want to find what are the fundamental truths and what are the artifacts of the data, or the choice of methods. And you want to uncover any “cherry picks” that neither they, nor you, might be aware of at the start of the testing…
Basically, a lot of things seem innocuous and “reasonable”; so those are the ones it is more important to capriciously change in the benchmark and ‘see what happens’…
@H.R.
I’m patiently waiting for the Surface Stations project to wrap up. As soon as it does, I’m going to be asking Anthony for a list of the best rated station IDs and comparing their temperature record profile over time to the benchmarks. I fully expect to find that they show a non-warming signal much like the “best quartile” records. So yes, these efforts are synergistic. I can do a “q.a.” check of sorts on the stations and his “best stations” can help ratify or challenge this methodology.
Mr. Smith,
I’m not sure where on this site you might have covered this, but what would your opinion of satellite data be? Because there is “decent” agreement between Giss, Hadley and UAH/RSS and all show warming.
When I look at the NOAA net adjustments to raw GHCN data (or read your interesting work) I certainly believe that scientific bias has inflated the amount of actual warming, but it seems like it would be hard to make a case for no warming at all.
Thanks
@Dan R.
I don’t really have an “opinion” of the UAH/RSS series. I look at data and see what it says. I look at code and see what it does. I haven’t really looked at the satellites.
One thought does occur to me, though: The satellites were originally calibrated against what benchmark? And they have been checked and adjusted against what standard?
It would surprise me quite a bit to find that they did NOT use the land thermometer data from somewhere like GISS or Hadley…
The other small problem is that the satellite data are from a very short part of the total data series, and for 20 ish years of that series we have a bit “thin” a thermometer count. Then look here:
At the best 3000 thermometers. You see an interesting ‘stability” in the “global average of temperatures” for the year since 1980 with a small “blip” in 2000 and ending with about a 1C total rise over the period. That is not significantly different from the pattern seen in the satellite record, IIRC.
A look above shows that the “invasion” of the warm portions of the globe by thermometers leaving the cold north was more or less stabilized by 1979. with the major ‘oddity’ being a drop in N.C. thermometer percentages in 1999 (coincident with our “hottest year”) and followed by a rebound.
So at a first look (i.e. not a formal analysis) it looks to me like the changes from The March OF The Thermometers had largely reach completion just about the time satellite data started to arrive. THEN I’d guess the two were cross calibrated, and the “drift” between them since then is waved off as “not significant”.
On a purely emotional basis, I’d tend to trust the satellites more; figuring folks probably spend more time on calibration of a multimillion dollar instrument than on a thermometer in a tin can next to the BBQ… I also think the technique of remote sensing will help solve the problem of trying to sample a 1 km (or 100 km) on a side “box” with a 1 m on a side wooden thermometer housing. But again, that is just an emotional / intuitive expectation and is subject to being completely tossed out on any data to the contrary.
Basically, the satellite larger ‘pixel’ size dampens some of my complaints about an “average” temperature for a grid box based on the hope that the air integrates temperatures that might range from 32F to 85F inside a few yards on different surfaces. The IR sensor integrates the IR photons from the surfaces better than the air integrates them? Or the satellite reads a higher up band of air with more time to “integrate” the surface below it?
At any rate, in the absence of a formal review of their data, code, and methods, my opinion is not worth much and any agreement between the data sets might well be due to cross calibration, accident, or “whatever”. Even a stopped clock is right twice a day, and we don’t have a very long climate record via satellites to work with.
The GIStemp “baseline” is clearly outside that record (1951-1980), and what they do to “rewrite the past” is highly suspect as it neatly dodges the whole potential comparison of their present data to the satellite data (had the changes been applied to present data instead of changing the past…). So if the two were cross calibrated in, oh, the late ’70s or early ’80s, then GIStemp stays steady in the present, but re-writes the past (as it does), no problem… see, “we correlate nicely”… but preserve “warming” relative to a rubber ruler past and an odd baseline.
At any rate, those are speculations most useful for making scenarios for testing the actual products and far less useful for making conclusions. Dream up the “maybe” and the “could be” and make a test. Then test. THEN evaluate the test results. THEN and ONLY THEN reach conclusions and toss rocks…
I can comment directly on the temperture readings in airports.
Over the 100-year span, sensors currently in airports record a very low raw trend. And, of course, those sensors spent most of their lifetimes in areas that were not airports, as there weren’t any airports in 1900.
However, over the 30-year span, raw data for airport sites shows a dramatic warming for a number of reasons: deregulation, encroachment, and the HO-83 equipment issue, to name three.
Only CRN5 stations (the most poorly sited) record a warming greater relative to the background measurements of the states in which they are located than do airports.
Also, for what it’s worth, USHCN station average trend over the last century (equally weighted, no gridding) for the US is as follows:
Raw: +0.14C
NOAA TOBS adjustment:+0.31C
NOAA FILNET (which includes TOBS): +0.59C
@evanmjones
So we know airports are among “the worst of the worst” in the last 30 years, and we know tht the FILNET adjustment does more to raise the “temperature” than nature does. Hmmphf.
Now if only the record had a “terrain type” of “airport”…
Do you have a list of “Airports by Station ID”? It would be interesting to match such a list against the v2.inv file and see what terrain types it finds for those airports…
In the stuff I’ve got, many airports have no “AIRPORT” in their name. Since location is only xx.yy you don’t get a really fine grid. One could look them all up (all 13,000 station records!) by hand on a map product and then “scroll around” to see if an airport is roughly “there”, but without the xx.yy.zz precision you have the risk of someone saying “But there is a weather station near, but not AT that airport”… I know, it’s highly unlikely an airport would not have a station while something nearby would, but it is a line of attack on just looking up the stationID LAT / LON and seeing if a station is inside that box.
I DO really wonder why there is no “AIRPORT” station type… why they all seem to end up as “CROPS” or “CONIFERS” or trees / fields of some kind when they are in reality hot chunks of concrete, tarmac, and jet exhaust.
UPDATE: I found the “airport flag” in the v2.inv file.
EM. Re the chart you have of % ‘done the GISS way’ with 30deg latitude bands. I’m specifically interested in 90S-64S from 1909 (decade).
DecLatPct: 1909 0.1 5.3 3.4 7.1 79.8 4.3
DecLatPct: 1919 0.1 6.5 5.1 7.1 76.8 4.4
DecLatPct: 1929 0.1 6.2 5.2 8.2 75.5 4.8
DecLatPct: 1939 0.1 6.2 5.7 9.8 72.0 6.2
DecLatPct: 1949 0.1 6.2 6.9 10.9 68.7 7.1
DecLatPct: 1959 0.5 5.2 8.6 18.2 60.6 6.9
DecLatPct: 1969 0.8 5.5 10.9 19.0 57.2 6.6
DecLatPct: 1979 1.0 6.6 11.1 18.0 57.0 6.3
DecLatPct: 1989 0.9 6.7 10.4 15.0 60.5 6.5
DecLatPct: 1999 0.9 6.1 9.3 15.6 64.0 4.1
DecLatPct: 2009 0.9 4.6 8.2 17.8 64.0 4.5
Would you be able to run a grep for me to find the station ID(s) that give that 0.1% in that zone for 1909-1949? I’ve been though the station data and list of stations actually used, which I downloaded and ranked by latitude. The only stations I can find in that latidude band don’t start reporting until the 1950s at earliest.
As an intermediate step in the analysis, I matched the v2.inv data against all of the v2.mean data and brute force constructed a file with each record identified with all station data. It was a big (103 MB), ugly file and the “technique” offended the “database guy” in me, but it has it’s uses… So, these lines are very long and will wrap, but they are station ID, year, tempdata, then the station information. The latitude of interest is -60 to -90:
7018896800001909 10 6 -3 -35 -94 -102 -122 -66 -75 -24 -22 -9 BASE ORCADAS -60.75 -44.72 6 0R -9HIICCO 1x-9WATER A 0
7018896800011909 10 6 -3 -35 -94 -102 -122 -66 -75 -24 -22 -9 BASE ORCADAS -60.75 -44.72 6 0R -9HIICCO 1x-9WATER A 0
So we have two records, both from the same place, with different “modification flags” (the 0 or 1 just before the 1909 year). Eyeballing my dinky 12 inch globe is looks like the South Orkney Islands are at LAT -60.75 LONG -44.72
Here is the data for that station directly from v2.mean:
Base Orcadas – interesting.
The latitude of interest is not -90 to -60 it is -90 to -64 the way GISS does it.
But they are using data from Base Orcadas, a LAT -60 station, to supply the data for LAT -90 to -64 from 1903-1946 when Rothera Point (LAT -67) finally starts.
Naughty Naughty GISS!
I bet Base Orcadas is used for a large portion of ocean grid too – warming rate of +0.019C/year.
OK so Base Orcadas is very cold in 1903-1946 then warms. The reasoning that it is in the sea so should be warmer than the polar continent as is a good proxy station would seem reasonable.
But suppose this. It is (way) off the tip of the West Antarctic Peninsula, which is warming now. Supposing the WAP experiences anomalous cold and was just so then…. I wonder what the ice cores say.
Ellie in Belfast: The latitude of interest is not -90 to -60 it is -90 to -64 the way GISS does it.
In STEP3 GIStemp divides the world into 6 “zones” by latitude. That would be 30 degree bands. Did I miss something in there where they are not equal sized? There was some odd code that looked to be dead code that talked about somebodies “equal area zones”, but I thought it was a past experiment left in as dead code… maybe I need to go look at it again…
But they are using data from Base Orcadas, a LAT -60 station, to supply the data for LAT -90 to -64 from 1903-1946 when Rothera Point (LAT -67) finally starts.
Naughty Naughty GISS!
This is my reconstruction of what stations would be used based on my belief that they use equal latitude bands. This is not the output from GIStemp. It might be my error if the equal latitude bands are not correct. What is the source for your -64 limit / cut off? (Need to check the actual code against whatever documentation you ran into…)
I bet Base Orcadas is used for a large portion of ocean grid too – warming rate of +0.019C/year.
Well, STEP3 has no choice. As the only station down there then, it WILL be used to warm a 1200 km set of “boxes” around it.
OK so Base Orcadas is very cold in 1903-1946 then warms. The reasoning that it is in the sea so should be warmer than the polar continent as is a good proxy station would seem reasonable.
But suppose this. It is (way) off the tip of the West Antarctic Peninsula, which is warming now. Supposing the WAP experiences anomalous cold and was just so then…. I wonder what the ice cores say.
Reasonable line of examination. It is also located just about in the middle of the “downstream” area where the circumpolar current exits the ‘pinch’ between WAP and Argentina. I would expect that minor changes of trajectory out of the choke point would put it on warmer / cooler sides of the (now spreading as it leaves the choke point) current. There would also be opportunities for differential upwellings / downwellings. You would really need to know how the water their acts as PDO, AMO, ENSO, et. al. flip and flop.
Latitude bands:
64N 44N 24N EQU 24S 44S 64S 90S
-90N -64N -44N -24N -EQU -24S -44S -64S
I was surprised too:
http://data.giss.nasa.gov/gistemp/tabledata/ZonAnn.Ts.txt
Good point about the pinch point.
Damn wordpress removed the spaces. I forgot it would.
REPLY: No problem, I’m getting used to reading things with spaces stripped out ;-) -ems
@Ellie:
See: https://chiefio.wordpress.com/2009/10/22/gistemp-send-in-the-zones/
It is, in fact, the case that GIStemp uses 6 zones of 30 degrees each.
But only in STEP2… STEP3 uses a different set of zones.
The chart to which you posted a link says “anomalies” so it is after the STEP2 process and most likely based on the product of STEP3. That would be the “grids and boxes”. So at this point my expectation is that STEP2 uses 6 zones and smears things around inside those 6, then later STEP3 use those 6 zones to make different zones as it makes the “gridded boxes” (that can be up to 1200 km from the “thermometer” that has a real temperature…) to make new “anomaly zones”. If that is correct, then GISS is, in fact, smearing a station at -60 degrees to look like it has data in the -64 to -90 zone, when in fact it has a coarser grain of zones from an earlier step. Just Bizarre.
Nice bit of work. It certainly provides food for some serious thought.
Monitoring changes in the number of high, medium, and low latitude reporting stations and changing the range of these bins is a very crude way to estimate the possibility of bias. The amount of sunlight striking the surface of the earth varies roughly with the cosine of the latitude.* According to Boltzmann’s law, equilibrium temperature varies with the fourth root of the incident radiant energy.** So one could try to derive a useful mathematical relationship (linear?) explaining how temperature varies with the fourth root of the cosine of latitude.*** Then you could look at the list of stations reporting at any time in the past and use this mathematical relationship predict the mean global temperature GISS would report if they were merely averaging all reporting stations AND the changing latitudes of the mix of stations was the only factor producing “climate change”. This would give you an estimated of the maximum bias that might be obtained from changes in station location.
About 85% of the variance in the mean annual temperature of US cities correlates linearly with latitude (and about half of the residual correlates with elevation. Moving 100 miles south raises the average mean annual temperature about 1 degC
* When one takes into account the tilt of the earth’s axis, incident radiation is 96% of vertical radiation at the equator (rather than 100% using the cosine), 83% of vertical rather than 87% (cosine) at 30 degN, 48% of vertical rather than 50% (cosine) at 60 degN, 27% of vertical rather than 26% (cosine) at 75 degN, and 12.5% of vertical rather than 0% (cosine) at 90 degN. (When one takes the fourth root, none of these differences are important except >75 degN. At 90 degN, where the cosine predicts an equilibrium temperature of 0% of vertical (the equator) while correcting for tilt gives an equilibrium temperature 60% of vertical (the equator).
** Of course, energy flows from the tropics to the poles. If losses and gains are proportional to the temperature differential, the relationship will still be linear.
*** Given the tremendous difference produced by an ocean at the North Pole and a continent at the South Pole and the relative amounts of land and sea in the two hemispheres, I’d perform separate fits for the two hemispheres.
@Frank:
Yes, it is a crude cut. This was one of the earlier “discovery” approaches. The “finding out where to look” step.
I’ve now got a fairly in depth set of analysis done with “by Latitude”, “by Altitude”, “by Longitude”, and some hints for “by distance to water” as a useful approach. See:
Taking all these to “the next step” of measuring the bias impact is something I’ve just barely started upon. First up will be a benchmark of GIStemp, then the estimate of bias, then the measured impact. But it will be a while since I’ve got a lot on the “to do” list ahead of it.
The source code to look at the files is posted here, so you can do any particular study that suits your fancy. This is a “Ya’ll Come” party…
-E.M.Smith ]
E.M.
After reading and re-reading Frank’s note, I’m unconvinced that what you have done is crude – simple, maybe, but not crude. Frank’s description of the math for how temperatures come to vary by latitude seems useful for some purposes, but not necessary for yours.
The analyses you did were of temperatures “reported” – as best can be divined.
It matters not how they got that way due to the geometry of solar incidence. They are what they are and if the centroid of the population of thermometers has morphed from the higher latitudes to the lower, then clearly the “mean” temperature derived from that population will rise, without any rise in the individual records at those stations.
I think you had it right. Frank’s stuff is sound – just not applicable here.
REPLY: [ Perhaps crude was the wrong word. Inelegant? Lacking “nuance”? It’s just a direct “meat and potatoes” first screen. It says, “Look here”, and did it’s job. It pointed me to the more refined “by latitude” reports later. Whatever words mean that… But yes, nothing “wrong” with it, it’s just not “the final step”. -E.M.Smith ]
Pingback: The Blackboard » Timeline of “The march of the thermometers Meme”
Pingback: March of the Thermometers | Digging in the Clay