NOAA Adjustments

Anthony Watts has an interesting blink comparisonbetween the current version of history from the GISS and their version of history in 1999.  It is amazing that all of the manual adjustments they add to the raw data constantly have the effect of increasing historical warming.  By continuing to adjust recent temperatures up, and older temperatures down, they are implying that current measurement points have a cooling bias vs. several decades ago.  REALLY?  This makes absolutely no sense given what we now know via Anthony Watt’s efforts to document station installation details at surfacestations.org.

I created a blink comparison a while back that was related but slightly different.  I created a blink comparison to show the effect of NOAA manual adjustments to the raw temperature data.

adjustments

My point was not that all these adjustments were unnecessary (the time of observation adjustment is required, though I have always felt it to be exaggerated).  But all of the adjustments are upwards, even those for station quality.  The net effect is that there is no global warming signal in the US, at least in the raw data.  The global warming signal emerges entirely from the manual adjustments.  Which causes one to wonder as to the signal to noise ratio here.  And increases the urgency to get more scrutiny on these adjustments.

It only goes through 2000, because I only had the adjustment numbers through 2000.  I will see if I can update this.

50 thoughts on “NOAA Adjustments”

  1. Urban effects are only one of many possible biases that need to be corrected for. And, if you look at the data, many urban sites actually do have less of an upward trend in temperature than nearby rural stations, so the urban adjustment is not always downward.

    The other thing to bear in mind is that it is hardly a secret that temperatures in the United States have not risen very much at all, and the hottest year on the record is a statistical tie between 1998 and 1934. This is definitely not true for global temperatures, in which 11 of the warmest years on record have been in the last 13 years.

  2. I am still not clear on how an urban adjustment could possibly result in higher temperatures. You are fundamentally modeling an effect that can only increase the temperature measured at certain locations relative to what would have been measured if that location had not urbanized. I also find it hard to understand how adjustments for an effect that is confined to urban areas would result in changes to rural temperature records.

    “The other thing to bear in mind is that it is hardly a secret that temperatures in the United States have not risen very much at all, and the hottest year on the record is a statistical tie between 1998 and 1934.”

    Could you please define “statistical tie”? Do you mean that the error bars on the raw data are large enough to encompass the roughly 0.3 degC difference between 1998 and 1934?

    Or are you so confident in the various adjustments that you can claim, to a high degree of certainty, that the adjusted value for 1998 is closer to the “real” value than the raw data? Statistically the adjustment itself is suspect, it will introduce error. Of what magnitude is this error compared the magnitude of the adjustment itself? How do we know that this adjustment has adjusted the data enough, or hasn’t adjusted it in the wrong direction. Is there some empirical way to measure the ability of these adjustments to arrive at the “real” data.

  3. Jennifer,

    “The other thing to bear in mind is that it is hardly a secret that temperatures in the United States have not risen very much at all, and the hottest year on the record is a statistical tie between 1998 and 1934. This is definitely not true for global temperatures, in which 11 of the warmest years on record have been in the last 13 years.”

    But the best data is from the us, not only in consistency of device used, accuracy, number of stations, but also in persistence (longest data run from single location). This should lead one to conclude that the US numbers that do not show heating are more likely to be correct. Are the US numbers weighted more heavily?

  4. “Are the US numbers weighted more heavily?”

    I’d have to go back and look again, on either Climate Audit or Watts Up With That, someone pointed out that something like 2/3 of the stations are in the US (maybe it was more). So yes, the US numbers outweigh the remainder of the planet for GISS.

  5. GIGO… with all that’s at stake it would be nice to have a second and third source for this. It sort of has the look of the fox guarding the hen house.

  6. GIGO… with all that’s at stake it would be nice to have a second and third source for this. It sort of has the look of the fox guarding the hen house.

  7. Part of our problem is the language of the trade.
    Negative adjustment = the same thing that happened when Momma shook the thermometer down after checking you for a fever.
    Positive adjustment = the thing that happened when the school nurse left the room and you rubbed the thermometer bulb vigorously against your jeans, in the hope of avoiding that French language exam in 4th period.

    Hansen is literally rubbing the bulb from the 1940’s up to the present.

    And lets not forget the fact that it is impossible to cool down a station artificially when that station’s default position is supposed to be alone, out in the elements, well away from shade trees, with the thermometer enclosed from direct sunlight by an aerated box.

  8. Paper:
    Anthony Watts provided an example of artificial cooling. A station covered in melting snow.

    That being said, unless one is somehow magically expanding the air around the station, pretty tough to artificially cool a station.

    Cheers

  9. In the 17th century Galileo, aware that sound has a finite speed, hypothesized by analogy that light also has a finite speed. He stood on one hilltop with a lantern while a partner stood on another hill some distance away. Galileo’s system was to hold a cloth to block his lantern, then swiftly move the cloth away. His partner would look for the lantern signal, then whip away the cloth in front of his own lantern. The time between whipping away his own lantern and seeing his partner’s lantern would be the time for the round trip plus the time for reflex action.
    After measuring the time lapse for one separation, his partner went to a hill much farther away. That would equal the time for light to travel the longer distance plus the reflex time, which would be the same as for the shorter distance.

    Galileo’s plan worked in theory, but needless to say, error measurement in reaction time overwhelmed the measurement of lightspeed. I suspect those trying to measure global temperature change are having the same problems as Galileo.

  10. This comment is somewhat off topic, but given this blog’s history of heated exchanges on pre-1950 global warming, the following tid bit might be interesting:

    “Ed W. Cliver, [an expert on solar cycles,] was just last Friday at GISS giving a seminar on the ‘constant’ Sun and was met with a fair amount of resistance from the AGW modelers who do not like to hear that the Sun was not the cause of the temperature increase in the first half of the 20th century, because that opens the door to admitting of other natural causes, and with too many other causes floating around it becomes hard to say that the increase in the last half of the 20th century is not due to these other [natural] causes.”

  11. An Inquirer – where is this quote from? I don’t believe it represents the views of climate scientists at all.

  12. Jennifer:
    The quote is from Leif Svalgaard who is a colleague of Ed Cliver. (They have written papers on solar cycles.) I do not think that I would label Leif Svalgaard as either a Skeptic or an AGW Pessimist. He does tell skeptics that he does not see enough variations in historical solar ouput to explain variations in climate. At the same time, he does not sound convinced that GCMs to date are adequate to explain variations in climate.

  13. Jennifer, can you please explain your “statistical tie” remark? What statistical criteria are you using?

  14. An Inquirer: implicit in the comment is the notion that every single person involved in constructing climate models wants to come to a certain answer, and will ignore scientific results that they think might contradict that answer. Frankly, it insults my intelligence to expect me to believe such nonsense. Doesn’t it insult yours, also?

    Josh: ‘statistical tie’ means that the numbers are the same to within the errors – ie, the error on one or both is larger than the gap between them.

  15. ‘statistical tie’ means that the numbers are the same to within the errors – ie, the error on one or both is larger than the gap between them.

    Excellent, I was aware of the usual definition, just wanted to know what you meant, as I’ve never seen a clear quantitative analysis of error in temperature measurement or adjustment, so I assumed you were perhaps referring to some other statistical procedure. Now that I know what you mean, where might I find error ranges for both raw and adjusted 1934 and 1998 temperature so that I can validate your assertion. Or perhaps you could provide the numbers to validate your claim here.

  16. You’ve never seen a quantitative analysis of the errors? How hard have you looked? The main GISTEMP page (data.giss.nasa.gov) talks about errors and gives some references. Try reading those.

  17. Jennifer:

    “implicit in the comment is the notion that every single person involved in constructing climate models wants to come to a certain answer, and will ignore scientific results that they think might contradict that answer. Frankly, it insults my intelligence to expect me to believe such nonsense. Doesn’t it insult yours, also?”

    It’s called confirmation bias and it’s a well understood scientific phenomena. That’s why double-blind studies are gold standard and best practice. Are scientists insulted because they are expected to do double-blind studies these days? Of course not. Pity one couldn’t do a double-blind climate model… anyway a climate model is a theory. First you decide what you think causes climate change and then you try to model it. Nothing wrong with that, regardless of how significant you think CO2 should be in the equations. Any scientist who believes in a particular theory can be said to have a ‘bias’. That’s not necessarily a bad thing.

    Perhaps if you try to be more measured and take a more balanced approach you will come off as more credible. By trying to argue against everyone for anything you consider immediately objectionable, you sort of hijack your own arguments. Things are often more complex than they appear to be at first glance.

  18. Well I’ve read through Hansen et al 2001 – a source from which you can substantially reproduce the above animated chart above – there are no quantitative error estimates on either the raw data or adjustments. You are the one who made the claim that 1934 and 1998 are a statistical tie, and say you based this claim on the fact that the numbers are the same to within the errors. If this claim is accurate, you must have known the error bounds when you made the claim, and thus should be able to easily provide them here to substantiate your claim. Or you could just throw out random links that don’t answer the question asked.

  19. josh: have you read Hansen & Lebedeff 1987? That’s the one specifically mentioned on the GISTEMP page as having considered errors in detail. In any case, we really don’t need a detailed study of the errors to call 1934 and 1998 a statistical tie. Giss gives the anomalies for both years as +1.24 (http://data.giss.nasa.gov/gistemp/graphs/Fig.D.txt).

  20. This is all just too ridiculous to talk about. Seems the scientists over at NASA/GISS have the intellectual capacity of 5th graders. The Emperor really has no clothes on – the sooner this house of cards known as AGW is exposed to one and all the better. Then we can all get on with tackling REAL enviromental problems. By the way, we all breath out CO2 – is there going to be a cap-and-trade on breathing?

    Enough already.

  21. Please Jennifer, the unadjusted numbers are not exactly the same. They differ by at least 0.2 degC. Do the error ranges for 1934 and 1998 encompass this amount or not?

    Hansen & Lebedeff 1987 is not oddly not available on the GISS website.

  22. josh – it makes no sense to compare two bits of raw data. Why calibrate data, after all? And why are you so concerned about these two particular years?

  23. ‘Jennifer’, you really are a gift to the skeptics. It was you who brought up those two particular years!
    Yes it does make sense to compare raw data. You still have not explained why the recent temperatures need to be adjusted upwards.

  24. Jennifer, you made the comparison between the two years.

    Can you explain why it “makes not sense to compare two bits of raw data”? I personally have performed many laboratory experiments. I do not recall despising “raw” data or somehow seeing it as inferior. There certainly might have been a need to calibrate an instrument – for example a thermistor, as there is a high degree of manufacturing variability. But once my instruments were calibrated I worked with the data they produced, without adjustment.

    “Why calibrate data, after all?”. Indeed. In science, measuring instruments are calibrated. I don’t know what “data calibration” is.

    Are you suggesting that the instruments used to measure temperature out in the field are uncalibrated? I would imagine this would have been done in the factory. Regardless, even if this is so, there is no way to go back and retroactively calibrate those instruments and remeasure the actual temperature at that time. The measurements are what they are.

    So, again, I repeat my question, what are the error ranges on the 1934, and 1998 annual average temperatures, and does the error range encompass the difference between them? Either you can answer this question or you can’t. If you can’t, I ask that you retract your claim of a “statistical tie” between 1934 and 1998.

  25. This discussion is getting us nowhere. If you don’t know what data calibration is, then you have a lot to learn. If you think that raw data and calibrated data are equally useful then, again, you have a lot to learn. Why does the idea of a ‘statistical tie’ between 1934 and 1998 upset you, to the extent that you are issuing intemperate demands for ‘retractions’?

  26. Intemperate? Perhaps you find a simple, straightforward request that you back up your words up with evidence intemperate, but I do not.

    The idea of a statistical tie does not at all upset me. I am more than willing, upon review of the relevant error data, to admit that the difference between the two years is within the bounds of error bars. Does such error data exist? Have you reviewed it?

    As for calibrated data. The data is the data. One can estimate the error in the measurements, one cannot “calibrate” that error away.

  27. Have you looked at the graphics available on the GISS web page? There are error bars on the global temperature plots.

    You’re quite wrong about raw vs. calibrated data. If there are systematic errors that can be quantified, then they obviously can and should be removed. Why do you disagree with that?

    Finally, please remind me what point you are making that is relevant to the original post and my comment on it.

  28. About the graph ending in 2000, that has worked out great for me.
    I ran into some idiot who is arguing that there is a Bush pogrom to discredit science.
    So a chart showing that NOAA has been feeding the public multi-decadal bullshit right up to just before Bush took office is perfect.

  29. Jennifer, show me how to artificially cool a whole region through urbanization such that it requires intervention of state agents to correct the colding?
    Explain it.

  30. Jennitist – I am making the point, quite well I think, that you are utterly incapable of backing up your words with facts. If the error bars are available on the GISS web page, I do not know why you have gone to such lengths to avoid quoting the values here. So I will ask the question yet again. Is the difference between the US average temperature for 1934 and 1998 within the error bounds for the measurements of each of those years? You have claimed that it is. Please back up this assertion.

    “You’re quite wrong about raw vs. calibrated data. If there are systematic errors that can be quantified, then they obviously can and should be removed. Why do you disagree with that?”

    If scientists find a systematic error in measurement, they discard the data and re-run the experiment. They do not claim to have divined an adjustment which “corrects” the data. The fact that climatologists cannot go back in time and remeasure, does not exempt them from the rigors of scientific data collection. The data are what they are, they cannot be retroactively corrected and magically become more accurate than the raw data itself.

  31. So your ‘point’ is a simple personal attack, with no relevance to the science of climate studies. I’m beginning to see that such behaviour is not uncommon here, and may even be the norm. Anyway, the anomaly for both years is +1.24. The difference is zero. This is less than the error on each value. Nothing more to say: move on.

    Imagine you are on a plateau 4000+-100m above sea level, you see a mountain and measure its height to be 3000+-100m. Imagine someone at sea level measured the mountain’s height and found it to be 7000+-100m. Your raw data and their raw data are certainly not the same to within the errors – no statistical tie there. By the logic you’ve described above, you’d both have to discard the data and re-run the experiment, and there would be no way you could understand why your results were so different.

    You’re not a scientist, are you josh?

  32. “Nothing more to say: move on.” There is a lot more to say. As pointed out in the original post, they are only the same after Jim Hansen’s massive upward adjustments. Before the adjustments 1934 is about .3 warmer.
    “This discussion is getting us nowhere.” It is certainly getting you nowhere.
    As for the graphics on the GISS web page, they are a complete joke as recently shown with their october-september screw-up.
    And what is the point of your meaningless mountain-measuring analogy? Are you trying to saying that thermometer readings in the 1930s were different from thermometer readings today?
    You’re not a scientist, are you ‘Jennifer’?

  33. Jennifer,
    Hiding behind feigned offense over non-existent ad homs is not a winning strategy. Yet it does seem to be your standard tactic.
    Your analogy to refute the problem with error bars depends on the reader mistaking the 25% difference in your example with the reality of a <1% difference in the real event.
    One of AGW’s big tricks has been to claim meaningful data while wtill being hidden in the MOE.

  34. Scientist: As pointed out in the original post, they are only the same after Jim Hansen’s massive upward adjustments. Before the adjustments 1934 is about .3 warmer. – yes, and a measurement of a mountain from half way up it would only be the same as a measurement from sea level after a massive upward adjustment. Do you believe that all adjustments are inherently wrong, or just some? Please explain which, and why.

    hunter: non existent ad hom? Did you read what josh wrote? I am making the point, quite well I think, that you are… – see how the point related to me, and not to the science of climate in any way?

  35. Jennifer,
    Your continuing to grasp on to the mountain analogy only shows how poorly the AGW promoters are doing.
    We are not talking about ‘half a mountain’. We are talking about measurements within the margin of error.
    Your persistance in ignoring other people’s points, and in simply repeating yours, does not make your lack of support any less obvious.

  36. Jennifer,
    When you start backing up your claims with facts, then his observation might be ad hom. But even if it were, it is certainlynothing compared to what skeptics get treated to daily.
    Living in an AGW echo chamber is, I am sure, comfortable, but it would be good to come equipped to a skeptic site with at least a few facts, good reasoning skills, and skin that is not tissue paper thin.

  37. hunter – I have perfectly good reasoning skills. The problem is that reason doesn’t work on unreasonable people. Your last three comments had no scientific content at all, just attacks.

  38. Jennifer,
    Please point out where you last posted ‘scientific content’, rather than assertions.
    Disagreeing with you, is not attacking you.
    By the way, did I miss where you actually deal with Hansen’s quote, instead of your mistinerpretaion of my comment about his quote?
    I look forward to your demonstration of your reasoning skills anytime you wish to use them.

  39. “Well I’ve read through Hansen et al 2001 – a source from which you can substantially reproduce the above animated chart above – there are no quantitative error estimates on either the raw data or adjustments.” So I don’t know where others are getting their error estimates, but if Hansen doesn’t give errors, it’s NOT a scientific report. Measurements are meaningless without an estimate of their accuracy.

  40. No Jennifer. I have not personally attacked you. You have still failed to provide the error bars to back up your claim of a “statistical tie”. This is a fact. It’s not an attack. You on the other hand seem to be perfectly happy attacking my background, rather than simply answering the question. And please stop generalizing about “common behavior” here. You are talking to me – I am not everyone else.

    I am still waiting for an answer to my question, I have grown tired of repeating it, so you will have to look up stream if you have forgotten. As for your claim that the anomalies are exactly the same for the two year – preposterous. The adjusted numbers are not the data. The data is the measured number, plus the error bars. The reason I am interested in this is because I want to know a) if the adjustment falls within the error bars. b) what are the size of the error bars relative to the adjusted and measured anomaly.

    As I have said previously, I am more than willing to admit that there is a statistical tie between the two years. You have yet to provide any scientific evidence of such a claim. Saying “Look here, when you take the raw data, and calibrate it, using a very complex statistical process, you come up with numbers we can add to the measure data, that makes the data identical” is somewhat laughable. It’s statistical naivety at best – statistically you can’t decrease measurement uncertainty without making more measurements. So whatever you are doing with your adjustments, you can’t make the claim that the adjusted number is any better than the original number from a statistical standpoint.

  41. Yes, joshv, you are attacking me, not the data. You’re demanding things of me, accusing me of things, and generally being unpleasant. Is that your intention, or are you doing it by accident? “You are talking to me – I am not everyone else.”, you say. Actually, I am talking to other people as well. This is not a forum just for me and you to chat, josh. And other people are being as unpleasant as you are.

    Can you see why, in my analogy about measuring mountain heights, you would have to take into account the height of the plateau to compare the two measurements? The raw data would be meaningless. In a similar way, until you’ve accounted for differences in the way temperatures are measured, like the time of observation bias, your raw data is not meaningful.

    When I asked you what point you were trying to make, you made it clear your point was only to attack me and there was no wider issue you were interested in tackling. Have you thought of one yet? Why are you getting so upset that 1998 and 1934 were about as warm as each other on one small part of the world? For what reason do you wish to disprove this idea? If you did manage to disprove it, what significance would that have?

    Unless you can be constructive, and discuss the data and the interpretation and not me, then I have nothing more to say. Your fixation on the temperatures in two particular years in one particular part of the globe is not interesting or useful.

  42. joshv,
    Note how Jennifer- and most other AGW believers- resort to false claims of ad hom in order to avoid the issues themselves?

  43. I have stuck to my question and this appears to annoy you. You find it somehow “unpleasant”. You attempt to categorize me with some broad brush of “common” behavior, and I am accused of attacking you? Fascinating.

    “When I asked you what point you were trying to make, you made it clear your point was only to attack me and there was no wider issue you were interested in tackling. Have you thought of one yet?”

    Jennifer, I’ve amply expressed what I am trying to get at. If you haven’t figured it out yet, you are simply not listening. Let me summarize.

    Personally I don’t think anything like proper scientific error bounds exist for measurements of global average temperature. You certainly have not provided any such information. If there are no such error bounds, talking about a “statistical ties” is meaningless. If such error bounds exist, talking about statistical ties might be meaningful. Thought those error bars might include the entire history of measurement, in which case just about everything would be a statistical tie. Or the adjustment which resulted in your claimed tie might itself lie outside of the error bound, which would be a truly fascinating result.

    So please Jennifer, answer the following questions if you can:
    a) Do the error bars for the 1934 and 1998 annual US temperature anomaly exist? If so, what are they?
    b) Do the error bars encompass the difference between the 1934 and 1998 anomalies? If so, what other years do they encompass?
    c) If the error bars do not confirm a “statistical tie” before adjustment, how can an adjustment fall outside of the error bounds? Isn’t the adjustment supposed to correct for error?
    d) Are the adjustments made for time of observation exact? If so, please explain what methodology is capable of taking a min/max measurement, made at a particular time of day, and adjusting it to the exact min/max that would have been measured if the reading were taken at midnight? If this is not an exact process (and it isn’t) what error does this adjustment introduce into the measurement?

    These are direct questions about the data and the procedures used to adjust it. Simple, straightforward – they are not an attack on you. I will continue to repeat them until you answer, or you stop responding.

  44. Just sticking to the question, eh, josh? So, when you said “you are utterly incapable of backing up your words with facts“, that was an example of your tenacious adherence to the question, was it? You and one or two others here really can’t seem to see the difference between discussing scientific matters and being rude. Your behaviour has been classless and inappropriate.

    None the less, you seem to be at least attempting, now, to discuss science. So, here are some answers.

    1. GISS discuss errors on their global temperature estimates in pretty much all of their papers describing GISTEMP. Have you read them? Have you noticed the error bars on the GISS charts? What more do you need?
    2. How could they not when the anomalies are equal?
    3. Still you seem not to understand that comparing unadjusted data would not be meaningful. See the mountain analogy, and if you think it doesn’t apply, explain why.
    4. Again, all of this kind of thing is discussed in GISS papers. Maybe you should contact GISS with these questions. Why do you expect me to know their products so intimately?

  45. Jennifer,

    “1. GISS discuss errors on their global temperature estimates in pretty much all of their papers describing GISTEMP. Have you read them? Have you noticed the error bars on the GISS charts? What more do you need?”

    Please answer the question. If you find my pointing this out, again and again to be “rude” I most profusely apologize. But please – answer the question. Do error estimates for the US average annual anomaly for 1934 and 1998 exist, if so – what are they?

    “2. How could they not when the anomalies are equal?”

    The anomalies are not equal. They differ by at least 0.2 degC.

    “3. Still you seem not to understand that comparing unadjusted data would not be meaningful. See the mountain analogy, and if you think it doesn’t apply, explain why.”

    Your mountain analogy is ludicrous. Each of the observers is measuring relative elevation. Most likely they are measuring an angle. And in fact, previous to GPS, such measurements usually involved multiple measurements of elevation angles and distances from different positions – the measurements are subject to error, as are the location of the “known” positions. The trigonometric process of distilling these measurements into an estimated height has nothing whatsoever to do with the process of adjusting climate data. In fact, such a simple mathematical process allows for wonderfully simple error propagation. Our hypothetical surveyors, unlike you, should have no problem quoting the error bars on his mountain measurement.

    Regardless, back to climate data, either the adjustment is within the error bounds of the raw data or it is not. Which is it? If it is within the error bounds, then I assume you are would allow that the error bars on the raw 1998 anomaly are at least +/-0.2 degC? Is that correct?

    “4. Again, all of this kind of thing is discussed in GISS papers. Maybe you should contact GISS with these questions. Why do you expect me to know their products so intimately?”

    You made the claim of the statistical tie, which, given that I cannot find any actual estimates for the error bars on either the raw or adjusted anomaly for the US for 1934 and 1998, means that if you are correct, you must in fact know their products more intimately than I. My sincerest apologies that you find my simple request that you share your information to be so “rude”, “classless”, and “inappropriate”. Clearly, I am out of line.

  46. No, asking the question is not rude. Saying things like “you are utterly incapable…” is rude. You apologised for the wrong thing. Did you do that on purpose? If you can’t see the distinction here, I can’t think that there is any value in conversing with you. How about this – apologise for insulting me, endeavour not to repeat such behaviour, and then we’ll have a nice friendly scientific conversation. What do you say?

Comments are closed.