Global Warming from Africa – Contagious Spreading at 100 Miles Per Year

Global Warming Originated in Africa!

I have discovered the earliest “hockey sticks” to date in the data from North Africa. While I still have the middle of Africa to complete, the evidence is quite clear:

Global Warming originated in Africa and is spreading to the rest of the world.

Speed Can Be Calculated

As this is “breaking news” the speed at this time is just an estimate, but a highly precise one. Global Warming spreads at about 100 miles per year over land. The charts and data will be presented below to support this estimate. One very important note: It travels poorly over water. The exact mode of contagion to places like Australia and New Zealand is not clear. It may be that some as yet un-identified tropical vector can carry this infection. Or it might be vacation habits and carting home musty African trinkets.

By inspecting the charts for the North African countries, a rough estimate is that Global Warming has spread from Morocco to Egypt at about 100 miles per year. From there, it looks as though it traveled through Israel to Greece and eventually reached Italy. The earlier onset in Great Britain as compared to Italy may indicate some sea born transmission, though the mode is again very unclear; as the water of the Mediterranean blocked water born transmission to Italy. Perhaps that British penchant for carting off artifacts for the British Museum?

The Graphs

These graphs are all made from the GHCN data set, some of the finest data that NOAA / NCDC could assemble. It’s quality is beyond comparison. Way beyond.

I have applied the dT/dt Monthly Anomaly Processing to it, producing accurate graphs of the change of temperatures over time in each country. As no “fill-in” or “interpolation” or “homogenizing” is done in this process, we were able, for the first time ever, to see the spreading of Global Warming without such confounding factors.

As we look at each graph, there will be in evidence a “pivot point” where a basically flat, or often falling, temperature profile suddenly pitches upward into a runaway fever. A “Tipping Point”. What is truly startling, however, is that this date slowly progresses across the top of Africa, through the Middle East, and into Southern Europe. The “Tipping Point” clearly indicates when the Global Warming Infection arrives in each country.

With that said, here are the countries in order, starting at Morocco and working our way counter clockwise around the Mediterranean. Some countries may be skipped for brevity or, as in the case of Libya, simply because I forgot to make the report for them. (Who thinks about Libya anymore anyway. They are running out of oil, so what use are they? Heck, they don’t even blow things up anymore…) I hope to have an update as time permits.

All of these graphs are fairly large. You can click on them for a better close up view.


Here in Morocco, we have “onset” in 1972, way ahead of the rest of the world!

Morocco Monthly Anomalies and Running Total

Morocco Monthly Anomalies and Running Total


By 1978 or 1979 it has reached Algeria. (Unfortunately, the size of the country prevents accurate calculation of spreading rate from just the two data points, but we will be able to use the longer “baseline” of Egypt a bit later).

Algeria Monthly Anomalies and Running Total

Algeria Monthly Anomalies and Running Total


Tunisia is a bit hard to tell. It has clearly “caught it” by 1981 or so, but there are hints of a rise as early as 1976. Then again, it’s hardly any different from Algeria anyway. Heck, just look at a globe. It’s got Algeria to the West and the South of it. Ought to just make it a department of Algeria. But at any rate, the real steep “onset” is a bit lagged from the rest of Algeria.

Tunisia Monthly Anomalies and Running Total

Tunisia Monthly Anomalies and Running Total


To be added later. If I remember them.


Egypt is at it’s bottom about 1984, then it’s just a “mummy on fire” run up. I make that about 2200 miles in about 22 years, or roughly 100 Miles Per Year. This was calculated from staring at my 9 inch Replogle World Globe, but I stared at it 1000 times so the precision could be calculated to 3 decimal places thanks to the law of large numbers and the mean central limit theory (or some nasty limited theory of averaging lots of lousy things to make perfect precision from them)..

Egypt Monthly Anomalies and Running Total

Egypt Monthly Anomalies and Running Total


Now, I don’t want to accuse anyone of revisionist history, especially anyone who was Chosen By God Almighty, but the Israel data begins in 1862. While I’m pretty sure Israel was created in the 1940s. Rather than make a big fuss about this, I’m willing to just put it down to Divine Intervention. Lord knows the NOAA / NCDC could not have just made something up.

While the low point of the Israel data comes about 1985, this could just be an artifact of the general downtrend that was in place. Clearly by about 1992 the trend has become a Stairway to Heaven.

Israel Monthly Anomalies and Running Total

Israel Monthly Anomalies and Running Total


Syria gets a “Pivot” about 1992 / 1993 as well, and without the ambiguous lead in. A bit of a fast spread from Israel, but then again only Lebanon is in between them and Lebanon hasn’t been a real country for a few decades now. Just a rest stop for armies and spies as they travel between Israel and the Arab world. But hey, it’s a living and somebody has to do it.

Syria Monthly Anomalies and Running Total

Syria Monthly Anomalies and Running Total


Greece is another one of those slightly ambiguous cases. It stops dropping in about 1983, but does not really ‘take off’ until about 1995. My conclusion is that the islands prevented a rapid spread, but that a minor early infection might have begun from “Certain Greek Practices” involving visits to North Africa by sailors. At any rate, after a slow start, Greece is clearly “catching it” by 1995.

Greece Monthly Anomalies and Running Total

Greece Monthly Anomalies and Running Total


We see in this chart that Italy did not “catch it” until 1999 to 2000. Clearly it is spreading and taking a primarily (though not exclusively) land route.

Oddly, Italy looks like it may have suffered a similar illness for most of the 1800’s. That could explain a lot about Italy… The Fever seems to have broken in the late 1930’s and a fit of “chills” is evidenced. One hopes that the “cure” that gave Italy the “chills” in the late 1930’s is not worse than the disease… This raises the hope of a “cure”, but the specter that this might be a cyclical tropical disease with alternating fevers and chills. At the earliest opportunity I would like to conduct extensive research on the Italian History or that era focusing on medicinal value of foods and beverages in common usage then, starting in a small Villa in Tuscany… To that end, a small grant for a few millions Euros from the UN will be greatly appreciated. The world hangs in the balance and we have no time to delay! Action must be taken now!

Italy Monthly Anomalies and Running Total

Italy Monthly Anomalies and Running Total


As we saw earlier, Gibraltar has not yet caught Global Warming. Now we know why, being isolated by water AGW had to go the long way around the Mediterranean; and now we know that it is only a matter of time…

Gibraltar Monthly Anomalies and Running Total

Gibraltar Monthly Anomalies and Running Total

South Africa

Interesting to note, South Africa has not yet caught Global Warming either. It is unclear at this time if there simply has not been enough time, at 100 M/Yr, for Global Warming to have reached them (as we don’t know the exact epicenter of this epidemic) or if, perhaps, the Sahara has a sterilizing effect on this disease. Only time will tell.

South Africa Monthly Anomalies and Running Total

South Africa Monthly Anomalies and Running Total


Clearly this new information invalidates the CO2 thesis, as CO2 is not contagious and does not have an Out Of Africa root. The causal agent of Global Warming must now be addressed with the utmost urgency. Gaia has caught some kind of African Fever. Finding an antidote, antibiotic, or other suitable treatment must be pushed forward with all possible haste. We in the USA simply MUST devote all the resources of our new Medical Care System to treating Gaia as we are all children of Mother Earth and we must do what is clearly right “for the children”!


OK, it’s now after midnight and a new, non-April First, day.

While I hope folks enjoyed the fun, there is a slightly serious side to all of this. Those graphs are real. I did not doctor the data in any way. There is a small degree of “decision” involved in when to divide the data series between the “shaft” and the “blade” of the hockey stick, but that’s clear when you look at the charts (and I’ve described the ambiguities in the text). So it really is the case that the “Pivot” data does, in fact, come very early in Morocco and very late in Italy.

IMHO it is an artifact of human decisions about when to institute a change of processing of the data series; the “onset” of a new “Duplicate number” or a new thermometer count. Someone had a job, to ‘roll out the new stuff”, and just did it. Not thinking about “foot prints in the snow” nor about “was the job propagating an error in some way?”. So they most likely just started at a convenient point on the African Coast, Morocco, and started moving a pin around the map as they worked through the data series from a country; then moved onto the next one. The “old” duplicate number being for the “historical record” and the new duplicate number indicating the ongoing automated update process with automated rule based “QA” (that may have an unintended consequence…).

So take just a moment to look at those graphs again, and realize that is the real deal; then ask which is more plausible: “CO2 did it slowly accumulating over 100+ years of industrial revolution.” or “It’s an artifact of the data gathering and splicing process”.

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 AGW Science and Background, dT/dt, Favorites, Human Interest and tagged , , . Bookmark the permalink.

15 Responses to Global Warming from Africa – Contagious Spreading at 100 Miles Per Year

  1. Richard says:

    Looking through all of the graphs the main thing that stands out as being ‘artificial’ is the ‘bullseye’ where they all neck down as you have described.

    Is there any information on why and how this peculuriarity comes about?

    Might it not be worth while using a before and after trend line on all of the graphs showing the relative trends around this obviously artificial cusp point?

  2. suricat says:

    Love it!

  3. Chuckles says:

    I suspect that in the case of South Africa, the global warming got slowed down by having to climb over the map line bumps at the equator and the tropic of capricorn.
    Might be held up at the border as well.
    I’m sure the AGW/CC will get there eventually, as was noted in Copenhagen, it simply requires enough $ to lubricate the path.
    All principles are soluble in cash, and you know they have giraffes there, and if you cross a spherical cow with a giraffe, you’re talking about high steaks.

  4. KevinM says:

    What happened in Syria in 1991? The thermometer count seems consistent at 7, so probably they did not move. Clearly _something_ happened.

  5. E.M.Smith says:

    The “bullseye” is probably a manifestation of the first “shock” of when Global Warming arrives at a country. All the stations take a “reset” on their anomalies as the “Duplicate Number” is reset to a ‘post infection’ status. (This is conveniently done by NCDC, though exactly how they manage to so perfectly map “Duplicate Number”, or as it’s sometimes be called “Modification History Flag” , how they map it so perfectly to the arrival of this infection is still a bit of a mystery. It’s almost as though the know exactly the moment Global Warming will arrive in any one country.

    There are other times when the infection happens slowly and the bullseye is less distinct. Then their are often overlapping records for a given StationID, one with the old and one with the new “Duplicate Number”. I suspect it is an attempt to quarantine the infected thermometers, but it always fails as only the “new” Duplicate Number survives this infective event, though sometimes the old one will live for a 2 or 3 year “overlap” period. You can see these as times of very reduced range of the monthly anomalies, but not forming a perfect “bullseye”. (Though the sometimes precede one).

  6. E.M.Smith says:


    My God Man! You are right! Running my finger over the globe I can FEEL the hump at the equatorial join! Maybe that is why all the thermometers have left the mountains and the only ones surviving are infected, but at low elevation! “Global Warming” can not climb over major elevations like the blue line at the equator, and can only flow around mountain ranges at their low edges, starving out the thermometers that are at altitude! Chuckles, you ought to publish that!


    In 1990 there are 3 thermometer records for each thermometer in Syria with a “0”, “1” or “2” and “3” for each (though one of them has a “5” station 647400800005 is the oddball.

    In 1992, only the “2” survives for most stations, while the “3” survives for 647400300003 and the “5” survives too. It is all the “0” and “1” Duplicate numbers that “die off”. But that THERMOMETER is still there and reporting, it is only this changed “duplicate number” or data “Modification History Flag” that has changed. Undoubtedly the older method of calculating the temperatures from that thermometer died of infection from Global Warming, leaving only the more “Robust” methodology for that thermometer…

  7. Chuckles says:


    Yes I believe Saharan Camel Drivers Monthly is soliciting for papers at the moment, so I’ll have to see if I can make some time to write it up and submit.

    Fascinating story about the bump of the line on the equator in Africa. It was actually built in colonial times by a feller named Deek. Did it all on his own, with his own hands, and 167,000 indentured labourers.

    He was a bit of a Johnny Appleseed type; was forever planting trees wherever he went, and it can be quite difficult to spot it in places, quite well camouflaged, unless you know where to look.

    Guess he was trying to hide the Deek Line, or something like that?

  8. JRR Canada says:

    Thanks that explains everything,including the green zombies.Undead of the creeping plague. Thank for the laugh.

  9. Keith Hill says:

    For me, the two intriguing graphs are Algeria and Syria. Algeria had “Thermometer Birthing” not “Dying”,
    yet still managed the hockey stick right on cue. As Kevin M noted, Syria achieved the same on a level thermometer count so in the case of these two countries, some other “adjustment” must have been done. Another giant ‘dig here’ E.M?

    This put me back on the “timing”search which I’ve touched on before. 1988 was apparently the hottest summer in the US for 52 years and no doubt gave James Hansen confidence for his appearance before a US Committee that year (reportedly with windows wide open and air conditioning off on one of the hottest days of the year) ! He did not have the backing of NASA as his former supervisor, senior NASA atmospheric scientist Dr. John S. Theon made clear after he retired. He said Hansen had “embarrassed NASA” with his alarming climate claims and violated NASA’s official agency position on climate forecasting (i.e., we did not know enough to forecast climate change or mankind’s effect on it).

    It is evident Hansen had put his reputation and credibility on the line and when the next four years to 1992 showed sharp cooling in many parts of the world, likely to have been exacerbated by the the June 1991 eruption of Mt.Pinatubo, Hansen must have been put under extreme pressure, not only by his critics within and outside NASA, but also those scientists supporting the AGW hypothesis.

    More and more, I feel this to be a credible background and motive for the matters you’re exposing E.M.

    OOPS ! I see you’ve answered Kevin M on Syria ! Algeria to go. Typing cramp yet? Happy Easter to all from Tassie!

    REPLY: [ Yes, it works both ways. Put in one way, drop the other. Repeat as needed. I think you also see this in the “baseline” intervals where they often drop the temps on thermometer adding going in, then rise on drops coming out. Just like the thermometers in S. America run from the Andes to the beach, and those of Morocco leave the beach with the cold ocean currents and run up into the hot Saharan Atlas Mountains. I’d even go so far as to suspect some degree of ‘tuning’ adds and drops so the average ends up “showing nothing”. There are just some darned curious “too perfect” moments in some of the statistical comparisons that find EXACTLY nothing… that the adds and drops EXACTLY nullify. Just wrong… It just screams some kind of a ‘salami’ technique or an automated widget finding both warming and cooling trend stations then balancing them so that the cooling trends are in one grid box (and average to one spot) while the warming are spread all over (so become more boxes) as a hypothetical way it could have influence (such as Canada where Eureka “warms” half the country, but many ‘cool’ stations are swamped by being on top of each other in one smaller geography). I’ve focused more on disaggregating things rather than study of the “odd balance” when they are aggregated. Another of the “someday” list items…

    The key is STRONGLY in those “Duplicate Number” (as NCDC calls them, and that I thought I saw in the GIStemp code as “Modification History Flag” ) changes at about the 1990 point. In comments under:

    Boballab did a study on Costa Rica. link:

    and chased it down to the specific station changes. Just stellar.

    That’s what needs to be done for each and every “hockey stick” country. Find exactly what stations changed, what processing changed. What places were ‘cooked’ (accidentally or otherwise) and which cooling places got mysteriously truncated. About 4 staff years of work, as a first rough estimate… -E.M. Smith ]

  10. oldtimer says:

    Evidence is now available on-line for the Russell review of the Climategate e-mails here:

    The Climatic Research Unit have submitted their own evidence to the enquiry here:

    Click to access Climatic_Research_Unit.pdf

    Of what will probably be of particular interest to you, they have included evidence on their treatment of data and stations from p45 on, including comparisons with other data sets.

    I have not read it yet; besides which it will mean far more to you than to me.

    REPLY: [ If I get a chance I’ll take a look at their stuff. But I really want to get Europe done first… I set it as a goal to keep motivation focused ( no new idea to be chased until Europe is up… it’s driving me nuts. ;-) but it’s working. -E.M. Smith ]

  11. Kari Lantto says:

    Said Chiefio: “There is a small degree of “decision” involved in when to divide the data series between the “shaft” and the “blade” of the hockey stick”.

    There is an approach that avoids that “decision”. You could use the T (i.e. not dT/dt) and compute
    Tn(t) = T(t)/T(1990) for each station (or country) and plot that for all (many) stations/countries. As the series are now all one (1) in 1990 this will work rather well visually. You could, moreover, try to split these plots by a variable DF (for number of Duplicate flags taking values 0, 1, 2+) for some interesting sets of series. I know the software R would do that for you!

    What one expects is that for

    DF = 0, Tn is increasing
    DF > 0, Tn “smiles”, i.e. is flat or falls to a minimum (close to 1990) and then increases.

    If it works this also takes a step towards attributing the warming to the “data gathering and splicing process”.

    REPLY: [ Interesting idea. One complication is that while it is 1990 for most places, there is some variation by a year or two. Often for smaller countries it shows up as a line of zeros in the dT/yr for those months. (as the “mod flags” or “duplicate numbers” reset…) but as this tends to be “by country” it would not be an issue for any “by country” or finer grained look. The “R” enthusiasts seem to like it (and it does look like it’s fairly direct and terse). I may take a look at it again sometime (if / when time permits) but right now the idea of going off to learn Yet Another Language is not high on my list… ( I know, perhaps wasting more time by not sharping the axe enough… but every so often you must spend time chopping with the axe you have to get any wood cut. Endless axe shopping warms no one but the axe sellers…) I suspect the most direct thing is just to plot the data “by Duplicate Flag” and / or do a statistical distribution analysis of 0,1 maybe 2 VS. 3+ and see what shows up. Whatever “it” is, it’s a direct function of that change of processing. Picking an individual thermometer and comparing the “before” and “after” raw paper data records with the “after” monthly means would be very enlightening. It would immediately show if there was cherry picking or bogus tossing of “outliers”. -E.M.Smith ]

  12. Kari Lantto says:

    E.M.: “One complication is that while it is 1990 for most places, there is some variation by a year or two.”

    Suppose there is often a minimum in 1990. The beauty of the suggested approach is that when the minimum is instead eg. in 1992 that minimum will show up 1992, and be lower than 1 (one); probably just a little lower than 1. That is what makes me hopeful about the visual effect. The smile will broaden out a bit, say between 1985-1995, thats all.

    The cases when the minimum is 1962 or such should be attributed to completely different forces, and treated separately. So the question becomes how many countries/stations will contribute to a recognizable smile around the birth-year of the IPCC

  13. GregO says:


    You simply rock. You are my hero.

    I’m new to this end-of-times superstitious pseudo-religious nonsense known as AGW – only got into it late last year after Climategate. But as an engineer my first line of reasoning was “what do the thermometers say?” Not an averaged or homogenized or modeled or adjusted or eliminated or fixed-up or government approved or whatever reading; but the real no-lie actual output from the sensor in any unit of choice: 0 to 5V; 4-20mA; 0-500000 K; -100 to + 100 C; I don’t care: Step 1…What are the numbers?

    I don’t have time or the skill set to do what you have been doing; but I appreciate your hard work and am following your blog for further developments.

    Slightly off-topic, have you checked out the Argo project where they have launched buoys world-wide to check ocean temperatures?

    REPLY: [ Thanks for the compliment! What I’m doing is not very hard. You don’t have to do the dT/dt thing in Fortran, for example. I only used it because the rest of GIStemp was written in it and it’s what I’ve been working with lately. I think it could be done in a spreadsheet (the only part I’m not sure of is the “skip -9999 missing data flag” but I’ve seen guys do much more tricky stuff in spreadsheets so I’m fairly sure it can be done).

    Yeah, I had that same Engineers Response of “just give me the dang numbers straight out the machine. I’ll know what to do with them even if it’s measured in BTUs per Fortnight… ” I was “slightly agast” when each time I thought I’g gotten the “raw” data I found out it was just a little “less cooked”… Started at ‘the wrong end’ in retrospect by looking at GIStemp (the last stop on the sausage factory floor… who apply the decorator casings and smoke the suckers…) Then moved up stream to NCDC and found both GHCN “adjusted” and “unadjusted”. OK, easy pick there. Then found out that “UN-adjusted” did not mean “raw” and had in fact been “adjusted” just “not much” with some “QA” in it and a LOT of “Selection Bias”. (So lately a lot of winter months data are missing. Probably because it was too cold and failed “QA” as an outlier (as a first “line of investigation” guess). Then you start to crawly upstream from there to the 100+ B.O. Meteorology on the planet and some of them are “recomputing” their “data” to ship to GHCN at NCDC…. AAARRRRGGGGG! Raw “Data” are NEVER computed. PRODUCT is computed. So before the “raw”meat even gets to NCDC for cooking and grinding, it’s been spiced up and had a couple of odd species ground together at the BOMs of the world. That was a large part of what got me really grumpy about this whole thing. Just how ersatz and slapdash it was BEFORE you got to the “temperature series” programs like GIStemp.

    So at this point, IMHO, what needs doing is to get real RAW data from every country that has it. ZERO “QA”, “value added”, whatever. Then start over from that data AND NONE OTHER. (No contaminated old sausage meat need apply…). IFF there is “global warming” it ought to show up in a clean subset. And anyone can pick a place and start to compile a clean set of numbers. It doesn’t take a lot of technical work, just strong suspicion to check that what is “raw” is really really “raw” and not just “undercooked”. Perhaps even just reading the local newspaper archives (as those numbers are typically reported ‘pre-processing’ as read). One place at a time, we can make a decent record.

    And, IMHO, it doesn’t need any more “processing” than can be done in a simple spreadsheet by any bookkeeper on the planet.

    Per ARGO: I’m aware of it. 2 problems: It will only take about 100 years worth of data gathering to have enough data to know we are not just sampling a 50 year cycle poorly. The ocean is a very large place and I’m not convinced we’re able to sample it enough with those buoys to have a Nyquist sample. Maybe, maybe not, I’d need to see the numbers that prove it… (Oh, and how to factor in properly the several hundred to thousands of years time lag for heat to move through the whole ocean…) And then they went and ‘re calibrated’ because things didn’t look right. OK, they had it wrong at least once. Now we’ve got to decide which one was wrong? And is the old data “correctable” or “sausage fixings”? And how do we know the new data is right?

    So at that point I just decided to focus on the land temp data and leave satellites and buoys to others. (And if ANY of the others matches the land data, especially the ‘since 1990’ land data: Be particularly suspicious that something is very wrong in the others…) -E.M.Smith ]

  14. GregO says:


    Thanks for the reply.

    Now to step 2: Check those thermometers (electronic temperature transducers) for instrument biases/errors; but first a little digression on my background and experience.

    I own a small engineering and technical services company (been at it just under 10 years – before that I worked as a mechanical engineer/manager in a variety of industries about 20 years) and over those years we have had requirement to build temperature sensing apparatus “from scratch” so to say.

    On a scale of technical difficulty we estimated these jobs at about a 2 or 3 on a scale of 1-10, 1 being say, a voltage divider and 10 being a 7 motor axes of control DNA testing machine.

    You know what? Those simple little temperature sensing systems were surprisingly tough to get right – more like a 4 to 5. I have a lot of respect for the folks making temperature transducers – it’s actually harder than it looks – not impossibly hard – but you have to be careful, have a passion for uncertainty analysis, and be patient.

    A paper, instructive on the topic of weather station instrumentation, written by Lin and Hubbard of the University of Nebraska: “Sensor and Electronic Biases/Errors in Air Temperature Measurements in Common Weather Station Networks”(Agricultural Research Division, University of Nebraska at Lincoln, Journal Series Number 14228, July 2004) calculates instrument error and as an example of error analysis will cite excerpts here.

    In their paper they pick four commonly used temperature sensor systems and run an error budget on them. It is quite instructive.

    I paraphrase from their paper here:

    Four commonly used air temperature sensors were checked and regardless of microclimate effects, sensor and electronic errors in air temperature measurements can be larger than those given in the manufacturers specifications.

    I quote from their paper here:

    “Three temperature sensors are commonly used in the weather station networks: A thermistor in the Cooperative Observing Program (COOP) that was formerly recognized as a nationwide federally supported system in 1980; a platinum resistance thermometer (PRT) in the Automated Surface Observing System (ASOS), a network that focuses on aviation needs; and a thermistor in the Automated Weather Station (AWS) networks operated by states for monitoring evaporation and surface climate data. Each of these sensors has been used to observe climate data over at least a ten year period in in U.S. climate monitoring networks. The U.S. Climate Reference Network (USCRN) was established in 2001 and gradually and nationally deployed for monitoring long-term and high-quality surface climate data. In the USCRN system, a PRT sensor was selected for the air temperature measurements. All sensing elements in these four climate monitoring networks are temperature-sensitive resistors…”

    Here I paraphrase their paper to line up networks with sensors:
    COOP – Maximum-Minimum Temperature System (MMTS)
    AWS – HMP35C
    ASOS – HO-1088

    Here I paraphrase three of the four temperature sensor error analysis results from their paper:

    MMTS Root-sum-squares (RSS) was above 0.2 deg C and increases for both lower (40 deg C) where error increases to 0.4 deg C.

    HMP35C Sensor with CR10X datalogger error was above 0.2 deg C and increases for both lower (30 deg C) where error increases to 0.4 deg C to 1.0 deg C.

    ASOS sensor error is hard to express in words – graph really helps; but self-heating error ranges from low temp to high temp 0.5 to 0.8 deg C and linearization error (again low to high) of -1.2 to 0.4 deg C.

    For the USCRN PRT I quote from their paper:

    “The RSS errors associated with using the CR23X datalogger and the USCRN PRT sensor were from 0.2 to 0.34 deg C. Results indicate that the error from CR23X datalogger was a major source of error for the USCRN PRT sensor. It should be noted that this result does not really imply the performance of CR23X, but it reveals that the USCRN PRT sensor is not suitable for the CR23X datalogger if a higher accuracy is required in the USCRN network. This is because the half-bridge circuitry in the USCRN PRT on utilizes a very small portion (668-816 mV) of the full-scale input range (+/- 1000mV). The analog output signals only use 7.4% of the full-scale input range in the USCRN PRT measurements. This suggests that to obtain a higher accuracy from the CR23X, the USCRN PRT should be improved for its signal sensitivity. In addition, the self-heating power dissipated in the USCRN PRT sensor is relatively large for a standard PRT (1000 ohms) sensor.”

    Sorry about the extensive citations – not my point to review their paper – just want to show that even a doing a “sitting at your desk” kind of analysis of temperature sensors bottom up you may find that there is more error (really should be called uncertainty…) than the mfg lists in the spec sheet.

    Of course there is more; in my company we call it “deployment” and it refers to putting systems in service. On-site calibration; user abuse; environmental factors (electrical noise for example; and in the case of AGW, UHI with temperature sensors abandoned in asphalt parking lots with AC exhaust blowing on them…) long-term sensor drift; on and on.

    And our Climatology High Priesthood believes they are measuring planetary temperature. Really. I have my doubts. Call me skeptical.

    REPLY: [ I’ve got no problem with long detailed responses (but I’m sure nobody has noticed ;-) As long as things are advancing understanding, I’m all for it. (It’s only folks posting loads of ‘insult’ stuff that will get snipped.) So feel free to add Real Science to the discussion, even “in abundance”. That’s kind of the whole point of this place, IMHO.

    OK, I’ve given short shrift to the whole sensor thing by just sort of bundling under “instrument change like the ASOS”… but I do keep wondering how much of that 1990’s sudden change is the rollout of electronic sensing at airports. And, as you noted, the sensors claimed accuracy is not present over the whole range. I could easily see about 1/2 C or 1/3 C of that “loss of hair to the downside” being from the instrument used in one way or another (and that would be most / all of it, at some sites). I’ve also had a suspicion that perhaps during the ‘transition’ the electronics were calibrated and ‘spliced’ to the original LIG (or the LIG readings “conformed” to the electronics as reported in whole degrees) then a year or two after the LIG is gone, the “standard” is brought around and we get that 1/2 C or so “jump” as the electronics are “re-calibrated” It’s just darned suspicious that the volatility goes way down at the ‘overlap’… then after the overlap ends, we “step up”. One of the virtues, IMHO, of the old LIG thermometer is that it CAN’T be “re-calibrated” in the field. You get a consistent series from it over it’s lifetime.

    AND as you pointed out, electronic things change over time. Dopants drift in junctions, resistors age, capacitors shift, etc. So you calibrate it “as per guidelines” at, say, 20 C on a nice day for re-calibration, but don’t detect that the -20 C performance is now nowhere near the ‘as shipped’ curve. You calibrated a POINT or two, but not the CURVE. I’d like to see a selection of deployed sensors pulled back to a cryo-lab for curve mapping after 10 to 15 years in the field… but that would put a “splice” in the data where the sensor was swapped…

    I also have this nagging suspicion that feeding electricity into a box is going to have a differential impact on very low excursions… You may be dumping your heat OK for your calibration at +25 C, but as you try to move from -28 C to -30 C that tiny bit of excess energy in the box is going to “mean more” than it does moving from 30 C TO 28 C where you are much more effectively radiating and convecting… I really do question the validity of ALL the electronic readings from places that have excursions to below about -20 C as the “electronics heat island” is going to be worse and the calibration issues extreme.

    But I have to “pick my battles”, so I’m not taking on those issues personally. Feel free to post comments or contribute on that area (it interests me a great deal and I think it’s very important!) even if I don’t have the “stuff” to approach it. (If folks want a thread on it, I’d even post a ‘generic’ instrument posting, just my contribution would be small…) -E.M. Smith ]

  15. Pingback: Damn Lies, Statistics, Perspective, Graphs, and Background Pics « Another View on Climate

Comments are closed.