Open Mind

Garbage is Forever

August 31, 2007 · 134 Comments

Some garbage never dies.

A comment on a recent post consisted solely of this link:

http://icecap.us/images/uploads/Lockwood_and_Frolich_Review.pdf

I responded thus:

[Response: I hope you can contribute something other than just a link to a rather amateurish bit of denialist garbage. This paper actually tries to claim that global warming stopped in 1998! This is not a blog to advertize (rather trashy) denialist propoganda.

The Lockwood & Frohlich paper and its implications for possible solar influence are discussed here, here, and here.]

This led to the following:

This paper actually tries to claim that global warming stopped in 1998!

actually no, the paper I cited gives a very balanced view of both PMOD and ACRIM solar cycle analysis

you on the other hand link to yourself

temps since 1998 have decreased, if solar cycles 24-25 continue the weakening trend then cooler temps are in our future

CO2 forcing is on the way out as a major factor

I responded by quoting directly from the paper:

OH YES, ONE MORE THING, THE EARTH STOPPED WARMING

As David Whitehouse noted in a response to the Lockwood/Frolich paper, that the temperatures of the world have leveled off the last decade after peaking in 1997/98.

The implication, or outright statement, that global warming stopped (or reversed itself) after 1998, is one of the all-time favorites of denialists. Let’s take a look at the most recent temperature data. I’ve often (and consistently) said that the “modern global warming era” starts in 1975. Here’s the global average temperature, according to NASA GISS and HadCRU, from 1975:

t1975.jpg

I’ve plotted the two data sets using different y-axes, because they’re on a different scale. That’s because NASA GISS computes anomalies compared to the 1951-1980 baseline while HadCRU uses the 1961-1990 baseline.

Clearly there’s a strong warming trend in both data sets, with GISTEMP data rising at 0.018 +/- 0.003 deg.C/yr, HadCRU at 0.019 +/- 0.003 deg.C/yr. We can also see some some sizeable ups and downs, like the cooling for a few years around 1992 caused by the Mt. Pinatubo explosion, and the strong warming in 1998 caused by el Nino. So we can tell, pretty much by looking, that a few years aren’t going to give us a reliable trend (we can tell by running the numbers as well). That’s why “climate normals” are generally defined using a 30-year timescale, and climate trends are best delineated in that way too.

The 1998 peak is very strong. For the GISTEMP data, 1998 isn’t the hottest year, 2005 is — but for the HadCRU data 1998 is the hottest year on record. Maybe global warming actually did stop in 1998? How would we tell reliably? Let’s look at just the data from 1998 on:

t1998.jpg

It turns out that even if we start at the 1998 peak, the trend is still warming. This is true for both GISTEMP and HadCRU data, and both results are statistically significant. There’s no getting around it: the planet has continued warming overall, since the 1998 el Nino event. The fact that one of two data sets indicates we haven’t broken the 1998 record, doesn’t alter the fact that the trend is still hotter.

We can even look at a very short time span, say, from 2000 onward:

t2000.jpg

Despite the brevity of the time span, there’s still a statistically significant warming trend in both data sets. GISTEMP indicates warming at a rate of 0.028 +/- 0.019 deg.C/yr, HadCRU indicates 0.018 +/- 0.016 deg.C/yr. Note that the time span is so short that these results are far less precise than the 30-year trend; for the trend from 1975 the error range was only 0.003 deg.C/yr, but for the trend from 2000 the error range is +/- 0.019 or 0.016 deg.C/yr. The brief time span of the most recent data, and the strong autocorrelation of temperature time series, combine to make the error range considerable. But even for the brief period since 2000, the trend is still positive, and the estimate is larger than the error range: it’s significant.

A note of caution for those who want to post links to further garbage. This is NOT a holding place for links to denialist propoganda. If you want to link to something that’s relevant to the topic of discussion, that’s great! If you just want to pepper this blog with links to junk, it’ll be treated like spam.

Yes, the claim that “global warming stopped in 1998″ is garbage. But some garbage goes on forever. Even after the 1998 record is broken in the HadCRU data, this garbage won’t disappear; I’m sure that ten years from now, in 2017, somebody somewhere will be declaring that global warming isn’t real — it stopped in 2014!

Categories: Global Warming · climate change

134 responses so far ↓

  • george // August 31, 2007 at 4:24 pm | Reply

    “I’m sure that ten years from now, in 2017, somebody somewhere will be declaring that global warming isn’t real — it stopped in 2014!”

    I’m just glad you said that so none of us has to.

    Last time I said something similar, I got tagged as a “religious alarmist”*

    *I think that means someone who worships fire alarms, but I could be wrong.

  • windansea // August 31, 2007 at 5:53 pm | Reply

    nice pretty graphs, too bad they are based on garbage

    read the last 7-8 threads at climateaudit.org to see why

  • Petro // August 31, 2007 at 7:09 pm | Reply

    windansea claimed:
    “nice pretty graphs, too bad they are based on garbage
    read the last 7-8 threads at climateaudit.org to see why”

    Too bad your evidence in science is collected from absolute junkyard. Tamino’s graphs can be replicated using _any_ dataset by _any_ meteorological office in the world, actually even with the datasets of the Dorks in CA. If you don’t understand the explanations of others, try it yourself!

  • windansea // August 31, 2007 at 8:01 pm | Reply

    It’s too bad Hansen/Mann etc refused to open source their “science”

    the post 2000 correction was just the tip of the iceberg. The “jesters” will soon expose the “adjustors”

    CO2 probably has a minimal effect on our climate, I think the solar cycle theorists will soon be proved correct.

  • windansea // August 31, 2007 at 8:05 pm | Reply

    Tamino’s graphs can be replicated using _any_ dataset by _any_ meteorological office in the world

    I guess you haven’t read the last 7-8 posts at CA

    FYI, the “junjyard” as you call it found a fairly substantial error in the US temp record that was missed by all the experts for the last 6-7 years.

  • BrianR // August 31, 2007 at 9:15 pm | Reply

    the abstract submission deadline for the December AGU meeting is Sept. 6th….the CA folks should present their findings

  • Hank Roberts // August 31, 2007 at 9:32 pm | Reply

    > a fairly substantial error

    Can you state the amount? Do you know how many zeros, and on which side of the decimal point, and if it’s Fahrenheit or Centigrade?

    Just checking whether you’re for real.

  • S2 // August 31, 2007 at 10:30 pm | Reply

    The first time I remember hearing that “Global Warming stopped in 1998″ was back in early 2005.

    I’m not in the same class as Tamino, but plotting (annual, global) temperatures from 1998 to 2004 still shows a positive trend. To get a cooling trend I have to restrict the time series to 1998 – 2001.

    The trouble is that when I keep to the same period of time but start a year later, plotting 1999 to 2002 gives me an eye-watering trend of about +0.63 °C/decade.

    There are three plausibe explanations that I can think of:

    1) My knowledge of statistics is even worse than I thought.

    2) The time scale (just four years) really is too short for any significant conclusions to be drawn.

    3) “Global Warming” really did stop in 1998, but started again in 1999.

    The paper that started Tamino off on this topic concludes with the statement:
    “This year despite the predictions by Jones Hadley Center) as early as January that the year would end up warmest on record, looks to be cooler than 2006, with the record cold Southern Hemisphere winter and a cooling Pacific, and continuing that trend.”

    Whether this year is cooler or warmer than last year is, I think, irrelevant in terms of trends.
    Having said that, even though the Hadley Centre may have been a bit premature with their January forecast I’m fairly certain that 2007 will be in the top three, and certainly warmer than 2006.

  • windansea // August 31, 2007 at 11:00 pm | Reply

    0.15C

    condescend much?

  • nanny_govt_sucks // August 31, 2007 at 11:06 pm | Reply

    Tamino, you’d probably be well advised to hold off on some of this global surface temp trend stuff until the truth on these adjustments at the detail level are revealed. Hopefully the source code will be available soon so that we can find out what is really going on. For more info, see the siberia posts at CA.

  • dhogaza // August 31, 2007 at 11:38 pm | Reply

    Tamino, you’d probably be well advised to hold off on some of this global surface temp trend stuff until the truth on these adjustments at the detail level are revealed.

    NSG, perhaps we should just use the satellite data instead?

    You know, the satellite data that correlates very well with the surface temp record?

  • Heretic // September 1, 2007 at 12:55 am | Reply

    That’s actually a good idea, dhogaza. Tamino, you think you could overlap a graph of the satellite data to the ones you showed?

    [Response: Comparison of GISS global temperature and satellite measurements is here, comparison for just the lower 48 states of the U.S. is here. They're from this post.]

  • windansea // September 1, 2007 at 12:56 am | Reply

    problem with satellite data is it’s less than 30 years

    it does show a cooling trend since 1998 :)

    [Response: False. Despite the fact that satellite measurements (lower troposphere) exclude the arctic (the fastest-warming region of earth), and responded to the 1998 el Nino *much* more strongly than surface air temperature, from 1998 to present the satellite data show no trend. From 2000 to the present the trend in satellite temperature is 0.018 deg.C/yr.]

  • nanny_govt_sucks // September 1, 2007 at 12:56 am | Reply

    dhogaza, what do you mean by “correlates very well”? I think the trend since 1998 for the satellite record is quite a bit smaller than the surface temp trend. Perhaps Tamino can post that up as well for comparison.

    [Response: See the graphs linked to in the response to "Heretic." The correlation is not only quite strong, it's obvious.]

  • DWPittelli // September 1, 2007 at 1:35 am | Reply

    I think the biggest part of the disagreement is a semantic one. If no year since has been as hot as 1998, then a lot of people would say from that that warming has stopped. Others, probably with more knowledge of statistics, draw a trend line from 1998 to the present, and because 1999 and 2000 were a lot colder, this line trends upward, even though the temperatures at the left of the line (1998) are slightly higher than those at the right of the line. But as noted, the denialists, whether versed in statistics or not, likely believe that intermediate temperatures are irrelevant to a judgment as to whether we are warmer now than in 1998, as they see the debate.

    It is silly for either side to say the other is wrong if the debate is about whether “global warming stopped in 1998.”

    Going beyond the purely semantic, I think most smart people on either side would accept that the debate about what “stopped” means is really that of whether the data should lead one to believe that the trend will in future continue upward or not.

    Of course, even in a strongly upward trending line, not every year sets a new record. But one could argue that the claimed trend is sufficiently large, and the annual volatility sufficiently small, that we would not likely see 8 years without a new record, unless the trend is other than we believe. Or, one might argue the opposite (that the volatility is reasonably likely to swamp the trend for 8 years). Fairly simple statistics, which, sadly, I have not done just now.

    From another standpoint, one might explain the 1998 blip from a noted phenomenon such as El Nino, or the 1999 decline from some other events, and explain what the “real” trend is after subtracting all such events. Of course, there will then be a debate about which effects are significant and external to the trend itself.

    While I think Tamino is reasonable in how he describes the period since the beginning of 1998, and right to believe that there is nothing in the record since 1998 which should make us complacent about warming, I also think it is my turn to take the role of saying that he is being a bit over-the-top, in saying “the claim that ‘global warming stopped in 1998′ is garbage.”

  • Heretic // September 1, 2007 at 2:50 am | Reply

    “Temps at the left of the line are slighlty higher than those at the right of the line.”

    Are we looking at the same graph here? Except for 1999-2000 (possible La Nina there?), the global temps post-1998 are higher than pre-1998. This “semantic” confusion is only possible because 1998 was so hot.
    Pitelli, your comment could be interpreted as a suggestion that a true trend can be established only if every year beats the previous, record breaking, year.
    If 1998 had been somewhere between 1997 and 1999, this discussion would not even exist. I’m a little suprised: you recently emphasized that you do not deny that GW is happening but now you seem to say that arguing it is no longer happening has merit. The “semantic” explanation you put forth is exactly one of those convoluted reasoning types that I mentioned and from which you took so much offense.

    The long terms trends on the global scale are what matters and there is no uncertainty there that I can see on the graphs. I note that the global satellite linear regression is trending up even faster than the surface temp measurements.

    Meanwhile, the ice is still melting…

  • tamino // September 1, 2007 at 2:51 am | Reply

    DWPitelli, You’d make a terrific lawyer.

    The difference is not semantic, and the claim that “global warming stopped in 1998″ *is* garbarge. The entire reason we do trend analysis is so as not to be fooled by fluctuations. The essence, the goal, is to discern the real persistent pattern, even when the “ups and downs” are bigger than the pattern itself.

    One of the most salient points is that it really is necessary to look at a long enough time span to reveal what the trend is — short runs are too dominated by the fluctuations — so we apply statistical significance tests to find out whether the answer we get is meaningful. It’s also critically important to realize that deliberately choosing an “up” fluctuation as a starting point is absolutely nothing but classic cherry-picking. As such, it’s fundamentally dishonest. It must really piss off denialists that EVEN IF you start at 1998, you STILL won’t get a downward trend. Not in NASA GISS, not in HadCRU, not in MSU TLT satellite measurements.

    But if you do take a long enough time span to let the trend emerge from the fluctuations, the signal is clear, unambiguous, undeniable. The globe is getting hotter. The very first graph in this post tells the real story. The comparison of GISS and satellite data here tells the real story. Study them well.

  • nanny_govt_sucks // September 1, 2007 at 3:00 am | Reply

    [Response: See the graphs linked to in the response to “Heretic.” The correlation is not only quite strong, it’s obvious.]

    Well, I was talking about the trend since 1998. I wonder if you could post that up, comparing surface trend vs satellite trend.

  • John Mashey // September 1, 2007 at 4:37 am | Reply

    Nanny: didn’t you go through this Nov 2005, http://www.realclimate.org/?comments_popup=217

    I thought William explained it just fine. Was that a different nanny, or is there a cyclic phenomenon where the equivalent question comes back every 2 years?

  • cce // September 1, 2007 at 5:30 am | Reply

    UAH and RSS satellite data warming trends of 0.014 and 0.018 degrees per year respectively. Hadley Centre and NASA instrument records show warming trends of 0.017 and 0.018 degrees. By implication, I think the skeptical claim being made here must be that the UHI effect ended in 1979.

    It’s also funny that when the satellite data showed cooling or no warming trend, that was more than enough evidence that it was all the Urban Heat Island effect. When the UAH algorithm was shown to be flawed (which, unlike the NASA correction, really was significant on a global basis), then it suddenly isn’t long enough to draw any conclusions, unless we are talking about 1998 to 2006, in which case it is long enough again.

  • Petro // September 1, 2007 at 5:57 am | Reply

    windansea’ “a fairly substantial error in the US temp

    Not even “fairly” substantial, but insignificant, see:
    “http://tamino.files.wordpress.com/2007/08/from1975.jpg
    and
    http://www.realclimate.org/index.php/archives/2007/08/1934-and-all-that/

    Even if your friends in CA can not comprehend English, do or read graphs or will not do statistical analysis, it is not too hard for you to learn. You would be surprised to see that you get similar results than main stream science with your own work.

    But yeah, whendo the climate whiners do actual work. It is easier to throw manure in the sidelines.

  • DWPittelli // September 1, 2007 at 12:02 pm | Reply

    “your comment could be interpreted as a suggestion that a true trend can be established only if every year beats the previous, record breaking, year.”

    No. My comment was the suggestion that most people, looking at just 1998 on one end, and 2006 on the other, would conclude that it hasn’t gotten any warmer; that in the recent historical record, “global warming stopped in 1998″. The intervening years would generally not factor into their thinking.

    First, the initial post wasn’t merely that “global warming hasn’t stopped since 1998.” It was that the claim that “global warming stopped in 1998″ was “garbage.” That makes a difference, because use of such language in the face of different interpretations of the data makes you look extreme and perhaps dishonest.

    Second, it makes a difference whether you use the word “stopped” or “trend,” only the latter being explicitly in line with Tamino’s analysis.

    Third, I am trying to be helpful to you here. And yes, I was going to use a lawyer story to do it. Suppose, Tamino, you are in front of a lawyer or Senator. Or indeed, in front of any nonscientist you want to convince to take you seriously. And they ask you, “has global warming stopped since 1998?”

    After you explain all your trend lines etc., they then ask, “OK, just tell me how much warmer it is now than it was in 1998.”

    When you respond, they will think you either a crazy scientist with his head in the clouds, or a liar.

    If you don’t want the masses to think you a liar or a fool, you need to at least avoid rhetoric such as “garbage,” and probably also explain why the outlying 1998 data doesn’t invalidate the up trend. (Explaining annual volatility and the expected time between new annual records might help, El Nino almost certainly would help; but drawing a trend line which trends upward from 1998-2006 because 1999 and 2000 were cold, would not be persuasive in the face of: “so it was getting colder in 1999 and 2000″ or “but it’s still colder than 1998″.)

  • george // September 1, 2007 at 12:50 pm | Reply

    First, even if it were real, the conclusion that a short-term “fluctuation” (leveling-off or down-turn) in a decades-long upward temperature trend means that the trend has “stopped” (or was never real to begin with) is really stretching it, given the natural ups and downs in the temperature (due to ENSO, volcano eruptions, etc)

    Second, if one accepts the basic idea that increasing greenhouse gas concentrations have led to warming (which the vast majority of climate scientists agree on), one also has to accept the idea that the warming can not merely “cease on its own” — ie, without 1) a stabilization of atmospheric greenhouse gas concentrations at some level or 2) some factor that causes the energy input to the system from the sun to decrease.

    Furthermore, even if warming did stop (eg, in response to a volcanic eruption, a drop in solar irradiance, increased cloud cover or something else that caused solar input to the earth’s surface to drop) as long as greenhouse gas emissions continued and the drop in energy input were not long term, the warming would resume after a time.

    It is therefore safe to assume that barring a fairly long-term drop in solar input to the earth climate system, any cessation in warming (if it occurs) will be only temporary.

    In that regard — the critical one — this whole argument that “global warming stopped in 1998″ is more than a little ridiculous. It’s essentially a distraction from the real issue: CO2 increase.

    Respectfully Yours,
    George (aka “Al Armist”)

  • richard // September 1, 2007 at 1:01 pm | Reply

    “Well, I was talking about the trend since 1998. I wonder if you could post that up, comparing surface trend vs satellite trend.”

    Sounds like cherry pickin’ to me.

  • Boris // September 1, 2007 at 2:32 pm | Reply

    “If you don’t want the masses to think you a liar or a fool, you need to at least avoid rhetoric such as “garbage,” and probably also explain why the outlying 1998 data doesn’t invalidate the up trend.”

    But, DW, Tamino is not necessarily talking to the masses, but to people who already have an interest in the science. Further, the people who came up with the “global warming stopped in 1998″ are the contrarian scientists, Bob Carter, Richard Lindzen, and probably a host of others. When these “scientists” have to rely on the ignorance of the masses, well, what do you think that says?

    It;’s a garbage scientific argument and anyone with an open mind–expertise or no–can be convinced of this. We can’t help the willfully stupid.

  • windansea // September 1, 2007 at 3:07 pm | Reply

    [comment deleted because it's nothing but a copy-and-paste of an entry from another blog, and is irrelevant to the topic at hand.]

    [Response: Lots of people come here to discuss. I disagree with some of them, sometimes on a few points (DWPitelli), sometimes on almost everything (nanny_govt_sucks). But they give their own opinions, and at least *try* to comment on something relevant to the post. Despite sometimes extreme disagreement, I've never edited or deleted anything they've said, nor shall I.

    You, on the other hand, have made a habit of simply *regurgitating* denialist propoganda. In fact, it was your comment that spurred *this* post, a comment which consisted of *nothing but* a link to denialist propoganda, and which was irrelevant to the post on which it commented -- despite the fact that if you'd bothered to look, you would have found a relevant post. From what I can see, you don't think for yourself, you don't write your own copy, you just try to use my blog as a "publishing venue" for links and commentary from other sources.

    YOUR "COMMENT" WAS NOT YOUR COMMENT. You're just dumping your garbage in my house. So this is your first and last warning: I will not allow this, and repeated attempts will get you banned. If you've got something to say, say it. If you try one more time to use my blog as a forum for somebody else's posts, you're outta here.

    If you want to regurgitate other people's vomit, GET YOUR OWN BLOG. THIS IS MINE.]

  • DWPittelli // September 1, 2007 at 3:10 pm | Reply

    I don’t think people have to be stupid or even ignorant to believe that 8 years of not reaching the record temperatures of 1998 could reasonably lead to the claim that warming has stopped. Calling them stupid — or their beliefs “garbage” — won’t change any minds.

    I am neither stupid nor unscientific. As it happens, I think “temperatures of the world have leveled off the last decade after peaking in 1997/98″ is a reasonable (although neither complete nor predictive) description of the recent historical record.

    I do expect that temperatures will resume upward, most likely at something similar to the past century’s 1F per century or thereabouts. I see no conflict between the statements, anymore than a statement in 2000 that 1999 temperatures went down was false or invalidates warming.

    I do agree that it is invalid to use the “temperatures peaked in 1998″ claim to imply that warming is done and we have nothing to worry about. Certainly AGW deniers have so used it. I would attack this use of the claim as “garbage” but it is a loser argument to attack the claim itself as garbage.

  • Heretic // September 1, 2007 at 3:39 pm | Reply

    How much warmer is it now than in 1998? That is exactly what is irrelevant. It might not be much warmer now than in 1998, but it is a lot warmer than 1999, or 1997, or 1968, whatever. A lawyer with no understanding of science would ask such a question with the intention of convincing a jury with no more understanding of science. Taking a significant outlier as the start of a trend is total nonsense (or garbage, if you prefer). Garbage is not unusual in the courtroom.

    On the other hand, someone truly trying to understand what’s happening could look at it this way: how close have the past five years been to the exceptionally hot year of 1998? Has any year since been equally or possibly hotter than this exceptionally hot year? The warmth of that exceptionally hot year was linked to El-Nino, did any year since approach the same temperatures, without a significant El-Nino to explain it? Those would be much more relevant questions.

  • BrianR // September 1, 2007 at 3:43 pm | Reply

    windandsea re the Schulte analysis (which you lifted word-for-word from another source)….why change the subject from your concerns above about the satellite data or the topic of tamino’s post?

    It seems whenever tamino generates a plot that is inquired about in the comments, those who don’t like the result then try and change the subject.

    It’s a very curious thing.

    [Response: I guess you saw his "comment" during the brief period of time during which I was writing my response. And you too noticed that it was a verbatim quote of another blog entry, and irrelevant to the topic.]

  • Boris // September 1, 2007 at 3:49 pm | Reply

    DW,

    I don’t think the statement you give is a reasonable description at all because,

    1. It relies on a cherry-picked year. Begin in any other year–and I mean any–and the whole premise fails.

    2. As Tamino has shown, even if you cherry-pick the year, the trend is STILL positive.

    3. There is a well known reason for 1998’s anomalous warmth–the strongest el Nino ever recorded.

    I don’t think people who buy this argument are stupid, but some people accept it because they want to believe it very, very badly.

    Finally, you didn’t address what I feel is the crux of the “global warming stopped in 1998″ claim and that is that it is promoted by “scientists” who know better, and who are being downright deceptive in their arguments.

  • windansea // September 1, 2007 at 3:51 pm | Reply

    I was just refuting the comment by george that there is a consensus on global warming. [rest of comment deleted]

    [Response: You're not gonna get away with it. It seems to me you're clearly trying to change the subject to someone else's topic, using someone else's words from someone else's blog. I'll be happy to open the pandora's box about "concensus" -- some other day. But for today, you've poisoned the well, that topic is off limits.]

  • richard // September 1, 2007 at 4:58 pm | Reply

    “.. I think “temperatures of the world have leveled off the last decade …. is a reasonable (although neither complete nor predictive) description.. ”

    Based on the graphs above, I’d say it is stretching things quite a bit to characterize that statement as ‘reasonable’. It is instead a statement that appears to be designed to mislead.

  • windansea // September 1, 2007 at 5:56 pm | Reply

    so we can’t respond to other commenters, and anything that refutes the AGW “consensus” is denialist propaganda.

    Solar phase 24 & 25 are going to show who’s boss in controlling the globe’s climate.

    It’s not man.

    [Response: Of course you can respond to others' comments. And so far, there *isn't* anything that refutes AGW.

    What you can't do is use my blog to "publish" somebody else's blog post, in somebody else's words, on an entirely different subject than the one at hand.]

  • dhogaza // September 1, 2007 at 6:16 pm | Reply

    I don’t think people have to be stupid or even ignorant to believe that 8 years of not reaching the record temperatures of 1998 could reasonably lead to the claim that warming has stopped.

    Not necessarily stupid? I agree with you.

    Ignorant? Hopefully. Because if they’re smart and knowledgeable, they’re being dishonest.

    As it happens, the well-known denialists who make the claim are neither stupid or ignorant. Draw your own conclusion.

    I am neither stupid nor unscientific. As it happens, I think “temperatures of the world have leveled off the last decade after peaking in 1997/98″ is a reasonable (although neither complete nor predictive) description of the recent historical record.

    Doesn’t jive with this graph:

    http://data.giss.nasa.gov/gistemp/2005/2005cal_fig1.gif

    Unless, of course, you are cherry-picking 1998 as an endpoint, which, given that we know ENSO causes warming, and that no climate scientist claims that AGW is the ONLY source of warming, and that we know natural variability is significant, makes such cherrypicking “unscientific”.

    Since you’ve told us you’re neither stupid or unscientific, you wouldn’t do that, right?

  • dhogaza // September 1, 2007 at 6:18 pm | Reply

    I was just refuting the comment by george that there is a consensus on global warming.

    The happy campers at deltoid are cheerfully debunking the last effort at proving there’s no consensus here:

    http://scienceblogs.com/deltoid/2007/08/classifying_abstracts_on_globa.php

    [Response: Nothing personal -- you didn't make the mess -- but for today, the subject of concensus is off limits to everybody.]

  • BrianR // September 1, 2007 at 6:39 pm | Reply

    windansea says: “so we can’t respond to other commenters, and anything that refutes the AGW “consensus” is denialist propaganda.”

    From here, it seems you were edited because you went wildly off-topic after tamino directed readers to actual data.

    Even if the post is now drifting off-topic…we can bring it back….you might comment on what you think about these data and tamino’s analysis.

  • Steve Bloom // September 1, 2007 at 7:34 pm | Reply

    As Gavin Schmidt pointed out a couple of years ago in a discussion of this issue over at RC, if one wants to argue that AGW stopped after 1998, to be consistent one has to go back through the record and come up with a scientific explanation for AGW having stopped and started during the three (IIRC) similar periods of flat trend in the last thirty years. Of course the great thing about being a denialist is that you don’t have to be consistent or scientific.

  • Steve Bloom // September 1, 2007 at 7:37 pm | Reply

    Oh, yes: It will be interesting to see how the denialists twist in the wind when the next unambiguous record is set.

  • John Mashey // September 1, 2007 at 8:33 pm | Reply

    I don’t recall tamino explaining the following in one place, although various posts have touched on it. I apologize if this seems awfully basic to some readers.

    SLOPES OF NOISY TIME-SERIES
    OR HOW IS TEMPERATURE CHANGING

    1) One can draw a straight line between any two points, thus ignoring all other data. This is intuitively simple, which is why people do it.
    Also, people are drawn to local maxima or minima (”perceptually salient” is what psychologists would say) and are tempted to pick them as endpoints. So, even if someone is *not* cherry-picking, it is easy to go wrong, and cherry-pickers are glad to help.

    2) One should at least do a linear regression, thus including intermediate data, not just chosen endpoints.

    3) For a simple line, choice of endpoint strongly influences the slope, whereas a regression is much less sensitive, except as the series gets very short.

    4) Here’s a simple exercise to see this, doable with Excel and about 10 minutes.

    A) Pick a simple time-series, like:
    data.giss.nasa.gov/gistemp/graphs/new.Fig.A2.txt
    I did one from 1979-2006, but pick your own.
    B) Put years in A1:A28.
    C) Put temperature anomalies in B1:B28.
    You can chart this with:
    Select A1:B27, Insert>Chart, select X-Y (Scatter), use 2nd style on left; Finish.

    D) In C1, put =SLOPE(B1:$B28,A1:A$28), and then fill down through C27. Each entry in C gives the regression slope from that year through 2006.

    E) in D1, put =(B$28-B1)/(A$28-A1), i.e., this is the line slope from each specific year to 2006.

    F) Ctrl-Select A1:A27, C1:D27, Insert>Chart, select X-Y (Scatter), and use the 2nd style on left; Finish.

    Series1 plots the regression slopes if one started from 1979, 1980 … 2005. The variability tends to increase with later years, because the slope depends on far fewer points.

    Series2 plots the simple ruler slopes, which obviously vary much more from year to year than do the regressions.

    If one WANTS to cherry-pick (say to prove cooling):
    a) use lines, not regressions, so one can pick the preferred starting point (1998 or 2002). Of course, one could pick 2005:2006, which gives a -.08 slope either way … but that’s rather transparently silly.
    b) if forced to talk about regressions, focus on shorter sequences. Here, 2002 is the best (for cooling) since a 1998 endpoint gets balanced by 1999’s low.

    Of course, a chart like this helps select optimal cherry-picking points.

    RESIST TEMPTATION TO LOOK AT MINIMA/MAXIMA OR SHORT SEQUENCES.

  • John Cross // September 1, 2007 at 9:14 pm | Reply

    Steve: the other great thing about being a denialist is that you never have to twist in the wind. All you do is change your name or better yet just ignore it.

    For example, I am still waiting for Nanny to respond to the posts in
    Surface Stations
    .

    Being a denialist means never having to admit you’re wrong.

    Best,
    John

  • dhogaza // September 1, 2007 at 9:47 pm | Reply

    Nothing personal — you didn’t make the mess — but for today, the subject of concensus is off limits to everybody.

    Sorry, I was hoping it was obvious I posted the link to encourage people to take such discussion over to the thread at Deltoid.

  • nanny_govt_sucks // September 2, 2007 at 4:03 am | Reply

    John, I don’t want to give you a half-assed answer because I know you are a stickler for details. Unfortunately I can’t devote any time to a decent response at the moment.

  • DWPittelli // September 2, 2007 at 11:40 am | Reply

    dhogaza,

    I think you are being sloppy (that is the generous interpretation) when you imply my dishonesty

    1) Because you think my calling something a “reasonable (although neither complete nor predictive) description” (parenthetical in original) is the same thing as my embracing it.

    2) And because you apparently failed to read my posts with the minimal thoroughness needed to see I explicitly applied the same caveats to the recent record (e.g., ENSO; statistical significance of outliers to a trend; the trend will probably continue upward in future) as you think refute me.

    But what else is new?

  • Hank Roberts // September 2, 2007 at 2:02 pm | Reply

    Can an opinion be reasonable and wrong? Lots of people reason without understanding statistics and come to conclusions.

    http://scienceblogs.com/stoat/2007/05/the_significance_of_5_year_tre.php

  • What’s the purpose of auditing surface temperature data? « Clastic Detritus // September 2, 2007 at 5:12 pm | Reply

    [...] This is being discussed in much more detail here. [...]

  • dhogaza // September 2, 2007 at 7:00 pm | Reply

    As it happens, I think “temperatures of the world have leveled off the last decade after peaking in 1997/98″ is a reasonable (although neither complete nor predictive) description of the recent historical record.

    followed by:

    Because you think my calling something a “reasonable (although neither complete nor predictive) description” (parenthetical in original) is the same thing as my embracing it.

    I don’t think there’s anything unreasonable about my interpretation.

  • Adam // September 2, 2007 at 8:13 pm | Reply

    “The paper that started Tamino off on this topic concludes with the statement:

    “This year despite the predictions by Jones Hadley Center) as early as January that the year would end up warmest on record, looks to be cooler than 2006, with the record cold Southern Hemisphere winter and a cooling Pacific, and continuing that trend.””

    I suppose pointing out that the forecast was for a 60% chance that this year would be the hottest globally, probably won’t reduce Jones being misquoted forever.

    See: http://news.bbc.co.uk/1/low/sci/tech/6228765.stm

  • windansea // September 2, 2007 at 9:52 pm | Reply

    A note of caution for those who want to post links to further garbage. This is NOT a holding place for links to denialist propoganda.

    that’s why you call it Open Mind right?

    I’m sure that ten years from now, in 2017, somebody somewhere will be declaring that global warming isn’t real

    You may be in for a big surprise. Solar Cycle 24 is late to the dance. NASA panel was split on prediction whether 24 will be strong or weak.

    No reverse polarity sunspots at the right latitudes so far gives more credence to a later and weaker 24. Almost all predict a very weak cycle 25. If some predictions are correct these weak solar cycles will produce a fairly rapid cooling trend. I won’t post links as they might be deemed “denialist propaganda”

    So you may be correct in predicting such comments in the near future, and will have a hard time producing graphs with a warming trend.

  • windansea // September 2, 2007 at 11:12 pm | Reply

    RESIST TEMPTATION TO LOOK AT MINIMA/MAXIMA OR SHORT SEQUENCES.

    usually this is wise, but since the trend since 98 is pretty flat, we may be looking at the top of of a curve which is going to head downward if some solar cycle predictions are correct.

    If we begin a solar induced cooling trend this is good as it will give a nice contrast of the power of CO2 forcings vs solar forcings.

    Too bad there is little chance of global adoption of Kyoto in the near future, as the coincidence at the same time of solar cycle cooling could be used as proof for CO2 reduction viability.

    [Response: Didn't you read the post? The trend since 1998 is NOT FLAT. EVEN IF you start at 1998 (which is cheating for the denialist side), you still get a statistically significant warming trend in both GISS and HadCRU data. That's what this post is about.

    And please don't remind us -- again -- that you expect solar cycles 24 & 25 to bring about a cooling trend. We heard you the first three times.]

  • windansea // September 3, 2007 at 1:08 am | Reply

    that last bit was snarky

    [Response: I haven't even come *close* to being as snarky as you have.]

    disclosure: I used to believe in AGW

    little things like amatuer math in Mann-stick, strident “we have to do this NOW!” and failure to be gracious and fully disclose data and methods by AGW proponents (Hansen etc) raised my suspicion.

    Failure of most current climate models to address solar cycle forcings and push CO2 forcings has led me to being a sceptic.

    I look forward to rubber meeting road.

  • windansea // September 3, 2007 at 1:25 am | Reply

    And please don’t remind us — again — that you expect solar cycles 24 & 25 to bring about a cooling trend. We heard you the first three times.

    right…just have not met much rebuttal on that point.

    [Response: There's no such that as either rebuttal or confirmation, since those solar cycles haven't happened yet. Lots of people have tried to predict changes in solar activity, nobody has yet succeeded.]

    EVEN IF you start at 1998 (which is cheating for the denialist side), you still get a statistically significant warming trend in both GISS and HadCRU

    with GISTEMP data rising at 0.018 +/- 0.003 deg.C/yr, HadCRU at 0.019 +/- 0.003 deg.C/yr.

    so diferences of .015 are significant?

    actually in some case they are, a flattening curve would exhibit this.

    [Response: They pass the statistical *significance* tests for trend analysis. They amount to 1.8 and 1.9 deg.C/century. Future warming is likely to be even faster, as CO2 (and other greenhouse gas) cocentrations are on the rise.]

    Let’s have a thread about solar vs CO2 forcings shall we?

    [Response: I've done more than one post about solar forcings, the links are in this post. Did you bother to read them?

    If you want to choose the post topic, get your own blog.]

  • windansea // September 3, 2007 at 2:00 am | Reply

    nobody has yet succeeded.

    actually some have a better than average success in forecasting solar cycles

    quiz: who?

  • windansea // September 3, 2007 at 2:35 am | Reply

    [Response: They pass the statistical *significance* tests for trend analysis. They amount to 1.8 and 1.9 deg.C/century. Future warming is likely to be even faster, as CO2 (and other greenhouse gas) cocentrations are on the rise.]

    that’s nice, mauder minimums will result in equal cooling, so then what?

    Ice ages dominate paleo climate studies, warming trends are anomalies in the geological time frame, endless proxy studies validate this.

    [Response: Methinks windandsea doth protest too much. Way too much. Proof that garbage is forever.]

  • Hank Roberts // September 3, 2007 at 4:43 am | Reply

    Windy, seen this? Understood it?
    http://scienceblogs.com/stoat/2007/05/the_significance_of_5_year_tre.php

  • John Mashey // September 3, 2007 at 6:43 am | Reply

    tamino: sorry if this is getting too OT, but I’ve been looking for one of these, and all the other maunder-hopers keep running away.

    Windansea:
    So, do you want to formulate a 15-year bet to take over to http://www.longbets.org? I’ve explained before why individual years are essentially meaningless, except to the statistically innumerate, so how about 5-year averages (Global mean surface air temperature change) as computed by GISS?

    Start:
    2004 (i.e., 2002:2006, the latest for which we have 5 years): the 5-year average is .55 [the highest in their list so far, so I'm not cherry-picking a low point, I'm giving you every break.]

    End:
    2019 (i.e., 2017:2021), i.e., bet gets settled in 2022 when the numbers come out.

    I’d prefer 20 years, but I’d like to be around for the end of the bet, and I might make 2022.

    From your comments, you seem to think it will get colder. I think it will get warmer, i.e., specifically, that in 2021, the 2019 5-year average will be higher than that shown for 2004 on that chart [i.e., in case there are any adjustments in the meantime].

    We both take our chances with El Ninos/La Ninas, but I’d guess 15 years is long enough to wash them out.

    I picked the last year for which we have 5-year average, and it happens to be the highest (i.e., favorable to you).

    I’ll give you no-change [to two digits].

    I’ll even take my chances with volcanoes, which is a really big concession, as an (unpredictable) volcano like Pinataubo around 2019 could give you a couple years with .2C dip.

    How about it? You join LongBets, propose a bet of this form, we use their process to finalize it, and off we go. Say $1000 each? [I'm already registered there. I'm not interested in proposing it hoping somebody anonymous shows up, but if you propose a bet of this form, and we agree on it via the process, I'll certainly take the bet, and I'm not anonymous.]

    Of course, this is a bet I’d be ecstatic to lose, but here’s your chance.

  • Hank Roberts // September 3, 2007 at 2:44 pm | Reply

    But … Connolley (’Stoat’):

    “the series is definitely going up; that 15 year trends are pretty well all sig and all about the same; that about 1/2 the 10 year trends are sig; and that very few of the 5 year trends are sig.

    “From which the motto is: 5 year trends are not useful with this level of natural variability. They tell you nothing about the long-term change.”

    Arguing over statistically insignificant “differences” is … I don’ t know, there must be a word for it.

  • windansea // September 3, 2007 at 2:52 pm | Reply

    as computed by GISS?

    heh, that’s funny!

  • Heretic // September 3, 2007 at 4:24 pm | Reply

    I’m sure John would accomodate you by using satellite data if you prefer windandsea.

  • Steve Bloom // September 3, 2007 at 8:09 pm | Reply

    I see that our friend Windansea was among the frauditor crowd snarking about the lack of Atlantic hurricane activity on Friday evening. Not so much snarking since then.

    It’s clear enough by now that Windy is an ideological denialist who will be unpersuaded by any evidence until he personally is hit over the head with it.

  • BrianR // September 3, 2007 at 9:23 pm | Reply

    It seems that this thread, which discusses actual data, has lost its steam. Those who have expressed that they have problems with the data tamino shows here have yet to sufficiently explain what those problems are (and, more importantly, their significance).

    They are on to the latest consensus chatter now…amazing how quickly what’s perceived as the most important revelation in debunking AGW can switch from one topic to the next. No need to provide substance…just constantly moving targets of nonsensical chatter does the trick.

  • windansea // September 3, 2007 at 10:02 pm | Reply

    I see that our friend Windansea was among the frauditor crowd snarking about the lack of Atlantic hurricane activity on Friday evening. Not so much snarking since then.

    please post where I snarked about lack of hurricane activity, I posted a link to a study saying current levels of hurricanes are a return to normal.

    I’ll do it for you:

    Not much left to say about this. Instead of being unusually active, it looks like the current hurricane regime is simply a return to more normal conditions, following an unprecedented tranquil couple of decades.

    http://www.worldclimatereport.com/index.php/2007/08/31/more-hurricane-news/#more-262

    love the cute little nicknames you come up with

  • windansea // September 3, 2007 at 10:17 pm | Reply

    Brian must have reading problems

    Those who have expressed that they have problems with the data tamino shows here have yet to sufficiently explain what those problems are (and, more importantly, their significance).

    my first post in this thread:

    nice pretty graphs, too bad they are based on garbage

    read the last 7-8 threads at climateaudit.org to see why

    It would be impossible for me to summarize the gibberish Hansen used to come up with GISS, but I’ll paraphrase it with GIGO

    Besides, I’m not allowed to link what’s called denialist garbage here, so you’ll just have to go read the threads at CA to understand.

  • Hank Roberts // September 3, 2007 at 10:58 pm | Reply

    Sad how the participants over there mostly stir up these new kids who rush over to places like RC filled with their rhetoric and statements of faith, and come a’cropper, instead of teaching them something about analytical thinking first so they could pose useful questions. Canon fodder, as it were.

  • Hank Roberts // September 3, 2007 at 11:04 pm | Reply

    World Climate Report
    http://exxonsecrets.org/html/orgfactsheet.php?id=85

  • windansea // September 3, 2007 at 11:33 pm | Reply

    RE: “frauditors”

    seeing that McIntyre broke the hockey stick and then caused NASA to revise (very quietly) their post 2000 USA GISS temps substantially I’d advise Mr Bloom to look in the mirror and ask:

    “what have I done lately?”

  • dhogaza // September 4, 2007 at 1:29 am | Reply

    seeing that McIntyre broke the hockey stick

    nothing like a little humor on Labor Day!

  • windansea // September 4, 2007 at 2:45 am | Reply

    nothing like a little humor on Labor Day!

    from page 6 AR3 to page 487 in AR4

    pretty funny

    the piltdown man of climate science

  • Heretic // September 4, 2007 at 3:37 am | Reply

    As was mentioned before windandsea, if you’d prefer, just disregard the land data and look at the satellites’. Not enough difference to make any, well ,difference.

  • dhogaza // September 4, 2007 at 3:56 am | Reply

    the piltdown man of climate science

    Right. That’s EXACTLY what the NAS evaluation said, right?

    Oh, wait, Rush know more about science than scientists do …

  • Petro // September 4, 2007 at 4:06 am | Reply

    Hello there windansea! Just a couple of quick questions to scan your scientific reasoning.

    Is there enough evidence on evolution?

    What is your opinion on evolution?

    What sort of evidence would convince you that evolution happens?

    In the last question the word evolution can be replaced with the term global warming.

  • windansea // September 4, 2007 at 4:18 pm | Reply

    please note I am ridiculing Mann’s original 99 hockey stick, his composite with Jones graph in NAS report is more of a boomerang, but still a pathetic attempt to lower temps in MWP and raise them for LIA compared to other proxy composites.

    The NAS report definitely discredits Mann’s claim that 98 was hottest year, temps in 90s hotter than past 1000 years etc. I also like the gentle indirect slapdown they gave to Mann/Hansen etc saying that researchers should improve access to data used in publications.

    Like I’ve said upthread, I hope we get a couple of weak solar cycles so that a fairly direct comparison of CO2 vs solar forcings can be made.

    Petro, I believe in evolution, global warming, global cooling, but am sceptical of alarmist CO2 warming theories.

  • richard // September 4, 2007 at 4:51 pm | Reply

    “..indirect slapdown they gave to Mann/Hansen etc saying that researchers should improve access to data used in publications”

    At least they have publications; peer-reviewed publications in quality journals, at that. Plus there are (as pointed out above) other independent supporting datasets.

    The competing theories to explain the available observations are poorly-supported in the science literature. Blogosphere noise isn’t much of a substitute, I’m afraid.

  • Dano // September 4, 2007 at 6:26 pm | Reply

    “Garbage is Forever” is an excellent title for this entry.

    Look at all the recycled argumentation cycling thru the blogosphere again.

    A suggestion: someone should make a blog that does a timeline of the denialist arguments, and number each timeline.

    Then, we can simply reply to such things as “mann/Hansen data availability” comments with something like: “15″, which would refer to the denialist timeline number.

    Best,

    D

  • windansea // September 4, 2007 at 8:15 pm | Reply

    “mann/Hansen data availability”

    don’t worry, there’s a timeline of excuses being offered for why this simply cannot be done.

  • Hank Roberts // September 4, 2007 at 9:00 pm | Reply

    Wind… Cites, please, for the basis for your beliefs. So far you seem to be just rushing over here with scraps of rhetoric that are tiresomely familiar and long since addressed, as though you wanted to play Warming Bingo. If you want serious discussion, cite your sources and why you consider them reliable.

    PR sites are not reliable, no matter which spoke of the political wheel you consider most important in keeping us moving forward.

    “The more intermediaries between you and the original text, the further you are from actual evidence and the closer to rumor. …”
    http://www.chicagomanualofstyle.org/CMS_FAQ/new/new_questions01.html

  • windansea // September 4, 2007 at 10:00 pm | Reply

    Hank, do you mean sites like exxonsecrets you posted above?

    just kidding, tell you what, in an hour or two I’ll post some pro solar papers I like and why I became a CO2 denialist :)

  • BrianR // September 4, 2007 at 11:26 pm | Reply

    “I’ll post some pro solar papers I like”

    this is how politicized this has become…as if it’s an abstract philosophical debate to be argued with rhetoric and the right ’smoking gun’ documents…how ’bout we try and summarize the entire body of work, get an idea of what the summation of all the papers are….oh wait

  • Steve Bloom // September 5, 2007 at 12:01 am | Reply

    windy, you can’t even keep *minor* facts straight in your own head. Regarding the NAS report, e.g., Hansen had absolutely nothing to do with it, nor is there anything in the text or any statement from one of the authors that would justify saying that Mann was “slapped down.” As for why the lovely diagram featuring the many hockey sticks wasn’t in the AR4 SPM, it’s because the point made by the studies of recent paleohistory (i.e., hey, we’re not on a rollercoaster like this all the time) had been made effectively in the TAR and there was a large amount of other, newer material that was more important to feature.

    But thanks for letting me know you’re a solar loon so I can just ignore you from now on.

    P.S.: Your friend Lex *has* all the temp data. He’s avoiding processing it because he knows he’ll screw it up badly.

  • Dano // September 5, 2007 at 12:51 am | Reply

    this is how politicized this has become…as if it’s an abstract philosophical debate to be argued with rhetoric and the right ’smoking gun’ documents…

    No, no, no. This is psychology.

    Confirmation bias. “I’ll post papers I like” = I’ll post papers that support my worldview.

    Very common.

    Best,

    D

  • windansea // September 5, 2007 at 12:53 am | Reply

    you are right Brian, this is a waste of time…

    I will just leave you with a 15 year girl’s refutation of CO2 based global warming.

    it’s called Ponder the Maunder

    http://home.earthlink.net/~ponderthemaundercf/id12.html

  • windansea // September 5, 2007 at 1:46 am | Reply

    Steve, “researchers” is plural meaning all

    solar loon?? yes dear..let’s ignore the 800 lb gorilla and put the CO2 horse before the cart.

    Please do ignore me, you do realize you’re an inside joke even in AGW circles right?

    P.S.: Your friend Lex *has* all the temp data. He’s avoiding processing it because he knows he’ll screw it up badly.

    mmmkay..that’s why there are about 10 threads at CA reverse engineering Hansen’s magical adjustments for ROW.

    The fact that his GISS US data had a huge error that went unseen for 7 years of “peer review” is insignificant. :)

  • Hank Roberts // September 5, 2007 at 2:40 am | Reply

    Reading CA, pay attention to Dr. Curry. There may be other real climatologists there, but read the papers and check the footnotes.

    You will find good science if you can avoid being captivated by the PR.

  • windansea // September 5, 2007 at 2:59 am | Reply

    http://solarcycle24.com/

    fast forward to NASA’s Hathaway bactracking from his strong cycle 24 predictions

    recorded 09/04/2007

    oops!

  • windansea // September 5, 2007 at 3:16 am | Reply

    There may be other real climatologists there

    “real climatologists?” the ones who peer reviewed and accepted US GISS previously?

    The most modern and extensive SST network in the world?

    the ones who missed a .15C error in the last 7 years?

    I feel safe…

  • Hank Roberts // September 5, 2007 at 3:43 am | Reply

    Ask your bank, stockbroker, accountant, and tax assessor’s office how accurate they are.

  • windansea // September 5, 2007 at 3:57 am | Reply

    Mashey is close to getting it:

    I find consensus events for El Ninos, with 2nd year of pair given:
    1988: El Nino
    1992: Strong El Nino
    1993: El Nino*
    1995: El Nino
    1998: Strong El Nino
    2003: El Nino

    here comes La Nina and some low amplitude solar cycles, el nino means big ENSO heat transfer to Artic.

    Where does ENSO get hot?

    Pacific temps are cooling (bad data!!)

    CO2 game over.

  • Petro // September 5, 2007 at 4:07 am | Reply

    windansea claimed:
    “Petro, I believe in evolution, global warming, global cooling, but am sceptical of alarmist CO2 warming theories.”

    Ok. So which statments below you doubt?

    1) Atmospheric CO2 level has increased during last hunderd years.
    2) The added CO2 in atmosphere is due to human burning of fossil fuels.
    2) CO2 absorbs heat.

  • EliRabett // September 5, 2007 at 4:18 am | Reply

    Oh yes, Ponder. Eli was quite restrained

  • EliRabett // September 5, 2007 at 4:21 am | Reply

    Oh yeah, Old Man in a Cave has a heads up on a new Esper reconstruction

  • Hank Roberts // September 5, 2007 at 6:13 am | Reply

    > huge
    huge, yet insignificant.

    Pity statistics isn’t taught in the early grades.

  • cce // September 5, 2007 at 6:59 am | Reply

    I particularly like the NRC’s gentle smackdown of the notion that the MWP was both global and warmer than today.

    I also like AR4’s smackdown of McIntyre on page 466, where all of his “hard work” “breaking” the Hockey Stick is dismissed in a few sentences. Of course, all of those other (later) reconstructions showing warmer temperatures today can’t be good for the skeptics’ egos. However, the true slap to the skeptics’ faces is the restatement of those reconstructions in the SPM, extending the TAR’s time period from 1000 to 1300 years. After all that auditing, the goal posts have moved another 300 years out of their reach.

    On another front, NASA’s error accounts for a ~0.001 degree downward correction of global temperature during those 6 years. The trend in the oh-so-accurate UAH satellite record that every skeptic loved for “disproving global warming” was increased 40% after its latest flaw was discovered. That error covered almost the entire globe (except for the poles). But who cares about flaws like that.

  • fergusbrown // September 5, 2007 at 12:11 pm | Reply

    Thanks for the heads-up, Uncle Eli. the Esper is actually a review of the 2002 reconstruction. They had to cool down the MWP a bit. Has SMcI commented yet? Bet he’s looking for the wood-ants in the bark as we speak…

    A side note to all who struggle to reason with denial: it is not a rational position, and has no rational basis, in spite of casual appearances, so reasoning is inevitably going to end up like banging your head against a brick wall. These people simply do not (or will not) understand.

    You might as well reason with your dog. At least it will wag its tail.
    regards,

  • windansea // September 5, 2007 at 2:40 pm | Reply

    They had to cool down the MWP a bit.

    “we have to get rid of the MWP”

    UAH “record” LOL

  • Hank Roberts // September 5, 2007 at 2:46 pm | Reply

    Chuckle. And my dog doesn’t bite me when I feed it something it’s begging for either.

    These kids nowadays ….

  • Dano // September 5, 2007 at 2:54 pm | Reply

    I agree with fergus just above. Denialists will not understand, as they take their identity to be their worldview. “Attacking” their worldview is attacking their identity – hence my ‘confirmation bias’ comment above.

    Best,

    D

  • henry // September 5, 2007 at 4:07 pm | Reply

    Just a few quick questions:

    Why is degrees “C” used instead of degrees “F” in the discussions?

    Because if I’m right in the calculations, doesn’t a .6C rise in temps equal a 33.08F rise?

    Also, what is the “margin of error” for that .6C result? (+/- C)

    [Response: Degrees Celsius are used because that's one of the standard scientific units for temperature measurement (the other is degrees Kelvin), and it's also the common unit used outside the U.S.

    A temperature of 0.6 deg.C is a temperature of 33.08 deg.F -- the large difference is because the two scales have a different "zero point." But a rise of 0.6 deg.C is only a rise of 1.08 deg.F.

    The margin for error in the 0.6 deg.C figure is about +/- 0.1 deg.C.]

  • John Mashey // September 5, 2007 at 4:57 pm | Reply

    Innumeracy often accompanies illiteracy.

  • fergusbrown // September 5, 2007 at 6:35 pm | Reply

    windansea: I look forward to seeing your temperature reconstruction. You do know what temperature is, I presume?
    regards,

  • Eli Rabett // September 5, 2007 at 7:52 pm | Reply

    We work for the lurkers and other innocents.

  • Hank Roberts // September 5, 2007 at 8:14 pm | Reply

    http://www.agu.org/pubs/crossref/2007/2007GL030643.shtml

    By the way, just checking, are all the “windandsea” posts coming from the same IP? I’ve begun to wonder if someone’s trying to parody the kid, posting under the same name.

    Or if it’s coming from Morano’s office.

    Or it’s a bot just throwing bits from the grab bag of talking points. Can’t see any sign of response beyond vague word-matching.

    Talk to us, Windandsea. What are you thinking?
    What are you relying on for your beliefs?

  • ChrisC // September 5, 2007 at 10:46 pm | Reply

    On ENSO and the solar cycle:

    A lot of work has gone into trying to predict ENSO based on various external forcings, such as solar. Piers Corbyn aside, no one has succesfully developed a method for doing so. If they could, they would make work a hell of a lot easier!

    ENSO is a highly non-linear, atmosphere ocean coupled phenomena. Also, ENSO is a regional phenomena. Their are other inter-decadal oscillations that have a greater affect on their regions. NAO, North Indian Ocean Dipole, Southern Annular Mode, Artic Oscillation… These are all highly non-lineral, dynamic processes, that effect climate regionally. ENSO is probably the most important, but is not the only driver of climate on Earth. These phenomena may or may not be coupled, we just don’t know.

    There is no evidence to suggest that ENSO is controlled soley, at least in any simple sense, by the sun.

    A quick question. Every climatologist worth their salt acknowledges that both solar insolation and greenhouse gases effect climate. Why, with denialists, is it an either/or?

  • Steve Bloom // September 6, 2007 at 12:12 am | Reply

    Regarding the NOAA study attributing 2006 U.S. high temps to AGW (linked by Hank above), somebody with a blog should do a post mocking this AccuWeather blog post. Watch the video for the discussion of the study by the “expert” denialist meteorologist. Boiling it down: “Global warming couldn’t have made it hot here last year unless it made it hotter everywhere else too.”

  • windansea // September 6, 2007 at 1:42 pm | Reply

    We work for the lurkers and other innocents.

    don’t forget the children Ghandi

  • fergusbrown // September 6, 2007 at 3:05 pm | Reply

    Jeez, even your ad homs are illiterate!

  • Hank Roberts // September 6, 2007 at 3:06 pm | Reply

    Ok, judging from the spelling error there
    http://images.google.com/imgres?imgurl=http://librarianavengers.org/images/lookitup.jpg&imgrefurl=http://www.librarianavengers.org/worship.shtml&h=403&w=303&sz=70&hl=en&start=2&um=1&tbnid=xYXPV9kmKhz66M:&tbnh=124&tbnw=93&prev=/images%3Fq%3D%2522look%2Bit%2Bup%2522%26svnum%3D10%26um%3D1%26hl%3Den%26client%3Dfirefox-a%26rls%3Dorg.mozilla:en-US:official%26sa%3DN
    he’s likely a kid trying to troll, not a bot.

  • Julian Flood // September 21, 2007 at 10:42 am | Reply

    quote The fact that one of two data sets indicates we haven’t broken the 1998 record, doesn’t alter the fact that the trend is still hotter.
    We can even look at a very short time span, say, from 2000 onward:
    (SNIP) Despite the brevity of the time span, there’s still a statistically significant warming trend in both data sets. unquote

    It is not surprising that things are warming — there is a study (I think it’s on a NASA site but I’ve lost the reference — my folder marked GW is bulging with about two hundred markers and I keep losing things) showing earth albedo. From 2000 to 2004 the albedo dropped by a forcing equivalent of .9 watts/metre^2. The last calculation I have seen of anthropogenic warming caused by CO2 is 2 -3 watts/metre^2 for the entire 20th century.

    Of course it’s warming. The question is, why is the albedo dropping? Or has it stopped dropping? Palle/ (that’s an e acute btw) has a slide show on albedo showing a forcing of 4 watts/metre^2 equivalent from ‘82 to 2003. Of course it has been warming, it’s simple physics. But why is the albedo dropping?

    Minor changes in albedo dwarf the calculated effects of CO2 increases. -3% albedo change from 82 to ‘03 is nearly double the stuff we’re all worrying about.

    (Add here my swivelly-eyed theory of oil sheen and surfactant pollution suppressing the production of condensation nuclei. It’s _AGW_, it’s caused by Big Oil, it requires changes in the way we use cars — appealing as it does to the half-scientific mind and to the hair-shirt tendency, I don’t see how it can miss. When CO2, solar and GCR theories die, it might be the last man standing. Perhaps I should patent it. More details at my website.)

    JF

  • Global Warming - Page 48 - Volconvo Debate Forums // November 28, 2007 at 4:34 pm | Reply

    [...] since you were looking at the exact same graph in my earlier post. Look familiar? Anyway, here is a much better analysis of what temperatures have been doing over the past nine years. [...]

  • Global Warming - Page 52 - Volconvo Debate Forums // January 4, 2008 at 4:02 pm | Reply

    [...] has seen a leveling off of global temperatures while Anthropogenic CO2 in steadily increasing? False. There has been a clear warming trend over the past decade. Not that it matters, since the trend [...]

  • Carrick // January 5, 2008 at 9:41 pm | Reply

    Um, are you actually using linear regression to estimate the trend lines on data with such large short-term correlated variations? That’s pretty crazy, from a statistics perspective. Try maximum likelihood, not that this is much better given how large the year-to-year variability is, and how strongly correlated the variations in the trend are.

    For L2 (sum of squares) optimization functions, the slope is going to be driven by what happens at the extremes. In this case, clearly the slope of the black curve in this figure is being driven by what happens in the winter of 2000. Move your starting point to 2000, and you’ll find a much flatter curve. Not that this means anything, given how much correlation there is in the data. (L2 is only reliable, when the unfit variability in the data looks like Gaussian white noise.)

    There is other evidence of a global cooling rend, btw: The Earth’s albedo has also been increasing since it’s historic minimum in 1998. The sun has also been systematically cooling since then, so there probably is some truth to the claim that the Earth is currently cooling.

    You might also want to consider the satellite data too. They are immune from a number of problems of ground based data, including spatial under-sampling in remote regions, urban effects, systematic errors associated with a disparate array of temperature sensors, and so forth.

    The sad thing about “Garbage is forever” sort of titles is it labels you as an extremist yourself, unable to evaluate arguments on their merits without personal attacks. Especially given, as it is probably true that we’ve had a short term cooling trend, starting around 2002.

    [Response: Thanks for your comments, but I believe you're mistaken.

    Linear least-squares regression is a perfectly good method to estimate a trend even in the presence of autocorrelated noise, which doesn't have any affect at all on the expected value of the estimated trend rate, it only affects its uncertainty. If one wishes to be precise one can apply modified (rather than ordinary) least squares, which gives the least-variance unbiased estimate of the trend rate for a given data set. Furthermore, the *size* of the variability is completely irrelevant, only the signal-to-noise ratio matters, a fact which follows from basic scaling arguments. For a reasonable perspective on the application of linear regression in the presence of autocorrelated noise see Lee & Lund 2004, Biometrika 91, 240–245. Your claims that it's "pretty crazy, from a statistics perspective" and "L2 is only reliable, when the unfit variability in the data looks like Gaussian white noise" raises serious doubts about your statistical sophistication.

    Also, your claim that "clearly the slope of the black curve in this figure is being driven by what happens in the winter of 2000" is rather nonsensical. This is easily demonstrated by removing the winter 2000 data and recomputing the trend rate (compensated for red noise); the change is far less than the probable error and none of the conclusions which follow changes in any way.

    What's your source for the claim that "The Earth’s albedo has also been increasing since it’s historic minimum in 1998"? Both total snow cover and ice cover have steadily declined since that time, and your statement would seem to contradict albedo estimates from the ERBE satellite measurements. The decrease in solar output is simply part of its regular decrease due to the approaching minimum of the solar cycle, and gives no evidence of being anything different than has been observed for the last many centuries.

    I have indeed considered the satellite data, as evidenced by my latest post (and others). While they are immune to some of the difficulties of surface thermometer measurements they are also prone to numerous other difficulties, as evidenced by the considerable disagreement between different researchers' (UAH, RSS, U. Maryland, U. Washington) temperature time series, compared to the impressive agreement between different research teams' (HadCRU, NASA GISS, NCDC) temperature time series based on surface thermometry.

    The claim the "global warming stopped in 1998," so oft repeated by denialists, really IS garbage, and SHOULD be called by its right name. Many people hope to give it a reputation as simply an alternative viewpoint, but a pile of sh*t by any other name still smells the same.

    And frankly, there is no statistically valid evidence of a cooling trend since around 2002.]

  • Carrick // January 7, 2008 at 5:17 pm | Reply

    Actually it’s easy to show that linear regression based on sums-of-squares (L2) is affected by outliers, especially when those outliers occur near one end of the data sequence. Perform a least squares fit on this series:

    0 2.92
    1 3.05
    2 4.98
    3 6.98
    4 8.95
    5 11.04

    The correct slope without noise is “2″. Note that it’s a straight line with Gaussian white noise, with an outlier at t=0, which I created by adding “2″ to the y value.

    You get a slope of 1.72 using L2 and 1.97 using L1. Obviously L1 is giving a better estimate of the .

    You can see the effect by varying which point to make the outline (that is, add “2″ to the second point, perform a least squares fit and so forth.) Here are the results:

    Outlier_point L2_slope L1_slope
    1 1.72 1.97
    2 1.84 2.02
    3 1.95 2.02
    4 2.07 2.02
    5 2.18 2.02
    6 2.29 2.03

    Note that the slope for L2 is systematically affected by where the appears in the seres.

    It’s actually well known in the physics community that maximum likelihood is a more robust estimator than L2. As I’ve mentioned L2 is perfectly fine if your noise is Gaussian and white. It doesn’t work so well for correlated noise regardless of the underlying distribution. [See The Numerical Recipes in C discussion for further details.]

    A better thing to do with data like this is to use two-year running averages (this filters out the highly correlated annual variations), then keep individual years. Basically all you’re doing in that case is low-pass filtering the data with a cut-off frequency of 1 year, then decimating the time series to remove non-independent data points. Using another filter, e.g. a 4th order forward+reverse Butterworth would be better. Basically you have to use a two-year running average to achieve a Nyquist period of 1 year.

    Even better would be to low-pass with an 11-year cut-off. Then you really are looking at long-range trends (climate rather than weather). Like you, I don’t think you can make too much of shorter term trends.

    Also, your claim that “clearly the slope of the black curve in this figure is being driven by what happens in the winter of 2000″ is rather nonsensical.

    Actually I’ve done it, and it does change the slope If you start the series in 2002, the answers change dramatically: The big dip in 2000 “pulls” the trend line downward. You can get positive or negative trends depending on how you play with the data (making it clear that’s all you’re doing, especially with L2.)

    See my comment above about the effect of outliers as end points for why it matters for L2.

    The bottom line is that you simply can’t use linear regression based on L2 with the raw data series, because there are systematic variations in the data that you aren’t modeling. You either have to model the additional variation, or remove it from the data before applying your model. Bottom line is your statistical analysis is flawed because you are using a not particularly robust estimator on a data set that has large short-term correlated variations in it..

    What’s your source for the claim that “The Earth’s albedo has also been increasing since it’s historic minimum in 1998″? Both total snow cover and ice cover have steadily declined since that time

    The Earth’s albedo is predominantly driven by cloud cover not ice cover. The data I’ve seen are based on EarthShine and you can find the relevant graph here.

    Regarding solar luminance, I am of course aware that solar variation depends on solar cycles, though there are many other cycles besides the 11-year one that most people talk about. So simply going returning to the next maximum doesn’t guarantee that we’ll return to the same peak temperatures we saw in 1998. [I'm not saying it will or it won't, to make that clear, just that the science isn't definite on that.]

    Whether “global warming” has stopped or not… well that depends on what causes it, doesn’t it? I assume we all know that the anthropogenic component of global warming is driving by the water vapor feedback from CO2 increase, but that there are potentially confounding factors due to what affect that has on clouds (thin high clouds = more warming, heavier clouds=cooling). It is possible that we could have entered a period where more heavy clouds are being formed (possibly driven by increased aerosol particles from the increased reliance on coal-burning plants for example), and that we are thus currently experiencing a cooling trend. That says nothing for what it does to the long-term trends however.

    Sorry that I had missed your comments on satellite data. Given the problems with surface data, I would prefer dealing with satellite data using radiosonde data to establish it’s validity. Finally, I’ll just note that disagreement among the various satellite-based temperature series vary between themselves by no more than GISS and HADCrut do. And for the record, I do collect meteorological data as part of my work and have familiarity both with direct (e.g., thermister based measurements in an aspirated volume) and remote sensed (RASS, microwave) methods of temperature measurements….

    [Response: Once again, you're simply mistaken.

    First of all, the slope from a least-squares fit to the first set of data, with your outlier offset applied to t=0, is 1.72 +/- 0.32. So the 95% confidence interval from least-squares regression includes what you call the "correct" value.

    But the most important reason your whole argument is faulty, is that it *assumes a priori that the slope is exactly = 2*, and *assumes a prior that the 1st point is an outlier*.

    Let's suppose, just for the sake of argument, that these data had been collected by God himself. He tells us that he has added measurement errors (because he's omniscient his measurements are without error) just to simulate real-world human conditions, but he assures us that the errors are a stationary process and there are no "outliers". In that case, it's trivially easy to prove (without having to assume that the noise is Gaussian, or even white) that the signal is NOT a linear increase with slope = 2 (I'm confident that I don't have to give you a detailed tutorial on why).

    A proper analysis of this data shows that the residuals from a least-squares (or any linear) fit are *not* a stationary process. This means that either the data don't follow a straight line (possible), or the noise is not stationary (extremely unlikely if this is climate data), or that the first data point is an error and should be discarded (most likely).

    A proper analysis of global temperature data since 1975 shows that it is indistinguishable from a linear trend with stationary autocorrelated noise. In this case, to estimate the trend the *best* method is generalized least squares, and ordinary least squares is unbiased (but not least-variance); this is "well known" in the *mathematics* community.

    I also have to wonder exactly what probability function you're using for your maximum-likelihood calculation. After all, it's very "well known" that if one assumes the noise values are independent, Gaussian, and of constant variance (which you imply are the properties of the data so it would seem to be what you'd assume for a maximum-likelihood calculation), then maximum-likelihood and least-squares regression are *equivalent* (i.e. they give the same estimates of intercept and slope). How then do you get such different results?

    As for "it's well known in the physics community," I've often found that physicists, just because they're really smart people who use a lot of math, seem to think they know mathematics better than mathematicians. Skill as a physicist is not a license to claim superiority in other branches of science.

    As for moving the starting point to 2002, what you said in your earlier post is "Move your starting point to 2000". Surely this was simply a typographical error, but I can only respond to what you say, not what you meant to say, so I kept the starting point at 2000 and eliminated the winter 2000 data you had mentioned. But the fact is, you argument still fails. The slope for GISS data from 2000 to the present is 0.025 +/- 0.018 deg.C/yr, the slope from 2002 to the present is 0.005 +/- 0.026 deg.C/yr. Both are consistent with the value 0.018 deg.C/yr estimated from 1975 to the present. Perhaps you've neglected to compensate for autocorrelation of the noise in estimating error ranges. Or perhaps you didn't bother to compute error ranges at all.

    As for earth's albedo, you're quoting results from the "Earthshine" project, which doesn't measure earth's reflectivity directly but infers it from earth-light reflected by the moon. This method, while interesting and valuable, is absolutely no substitute for direct measurements of reflected light from earth -- and direct satellite measurements contradict the (in my opinion highly speculative) results of the Earthshine project (see Wielicki et al. 2005, Science, 308, 825, as well as this press release from NASA).

    As for whether global warming has stopped or not, it does NOT depend on the cause -- either it has or it hasn't whatever the reason. Statistically valid evidence that it has, simply doesn't exist.]

  • Carrick // January 7, 2008 at 10:10 pm | Reply

    First of all, the slope from a least-squares fit to the first set of data, with your outlier offset applied to t=0, is 1.72 +/- 0.32. So the 95% confidence interval from least-squares regression includes what you call the “correct” value.

    In this exercise we knew the correct (no scare quotes) a priori (we even knew the sigma of the Gaussian noise, which was 0.1).

    The purpose of the exercise was to identify which optimization function did the “best” job in solving the inverse problem in a robust manner. Maximum likelihood had a range of 1.97 to 2.02. L2 was considerably worse, and moreover, was sensitive to the location of the outlying data point.

    Not really sure what your point was regarding stationarity. If there is a problem here, there is 10x the problem in the monthly global temperature data. Really this is standard stuff. I wish you would quit assuming (I guess) that I’m making this up. The Numerical Recipes exposition is a decent one. I use a more mathematically-based text for my work, which I can provide if you are interested. At the moment, I’m assuming that you are encamped in your own notions of how inverse problems are solved, and that further exposition on this is worthless.

    Regarding the changing of the slopes with different intervals, your error bars are basically meaningless, because they require (as I’ve stated) a Gaussian white noise process. The fluctuations we’re seeing are not statistical, and it is a mistake to treat them as if they are. Which is basically what you are trying to do.

    As to albedo measurements from Earthshine being indirect. So what? Why do you think it’s an invalid measure? The physics behind reflection I believe are pretty well understood so to label them as “speculative” is well, just silly. I am aware of your other sources, I suppose you noticed that they aren’t in agreement with each other (this point made in the article you linked).

    As for whether global warming has stopped or not, it does NOT depend on the cause — either it has or it hasn’t whatever the reason

    Missing the point again.

    If we understand (hypothetically) the cause of a short-term slowdown or even reversal in global mean temperature, we’d be in a pretty good place to predict whether we would expect it to resume or not in the future. That’s all I was trying to say.

    [Response: You are no longer simply mistaken. You have proved that you are intransigent, statistically ignorant, and arrogant beyond belief.

    The statement that "your error bars are basically meaningless, because they require (as I’ve stated) a Gaussian white noise process" is more than ample proof of your appalling ignorance. If you were a regular reader of this blog, you'd know that I regularly compute error ranges which take into account the autocorrelation structure of the noise. If you had an in-depth knowledge of statistics you'd know that there are numerous ways to do so. If you were a colleague of mine you'd know that I have in fact advanced (in a small way) the theory of how to do so. I'm guessing that you just don't know how; in fact it seems quite certain because you have professed that error ranges require the assumption of a Gaussian white noise process. Yet you presume to lecture me on the appropriateness of my estimates of the error range, which is pretty solid evidence of your arrogance.

    You haven't given us a clue yet about how the maximum-likelihood method performs so much better than least-squares, when under the stated assumption of a stationary Gaussian white-noise process, the two methods are equivalent.

    I'm guessing that you don't see the point of the noise process being stationary, because you simply don't know what it means.

    As for albedo measurements, the earthshine project measures a reflection of a reflection, so no matter how well understood the physics, the signal-to-noise ratio is vastly less than it is for direct measurements. And as the reference I pointed you to mentions, the Earthshine project estimates are contradicted by the climate itself: "If such changes had occurred at the magnitude of the earthshine data in (4), a global cooling twice the level of the ~0.25 C of the Pinatubo eruption would be expected, even over short time periods."

    I’m not assuming that you are encamped in your own notions, I'm quite certain of it. I meet your type all the time; you're a textbook example of the physicist who thinks he knows better than everybody else about everything, even when he steps out of his field. Further discourse with you is worse than worthless.]

  • Carrick // January 8, 2008 at 12:31 am | Reply

    Stationarity has nothing to do with the problems with using L2 for robust estimation. Non-normality and correlation are the main issues here. Stationarity is necessary if you want to make sense of the error bars, but if the data are non-normal and correlated (the main problem here), you don’t ever get to the point where you converge to a meaningful answer. This is a particular problem for measurements in meteorology (as I’ve had to personally deal with) because the variance typically increases as you increase the duration of the measurement (this is typical of 1/f^n noise, common in meteorology, until you get durations that allow you to probe the “source region”).

    Here’s a good reference on robust statistics for people with an “open mind”. I haven’t discussed the issues myself because the readily available references do such a better job than I could.

    You’re right that I’m “encamped” in a position with respect to the statistics. I’ve been using robust estimators for 20 years now, and wouldn’t go back to L2 for these types of data if my life depended on it!

  • Carrick // January 8, 2008 at 12:45 am | Reply

    Sorry, I seemed to have messed up the link to Wikipedia. This is a good summary statement:

    In statistics, classical methods rely heavily on assumptions which are often not met in practice. In particular, it is often assumed that the data are normally distributed, at least approximately, or that the central limit theorem can be relied on to produce normally distributed estimates. Unfortunately, when there are outliers in the data, classical methods often have very poor performance. Robust statistics seeks to provide methods that emulate classical methods, but which are not unduly affected by outliers or other small departures from model assumptions.

    The Wikipedia reference gives some good references for further reading. Beyond that I would recommend Panofsky and Dutton which discusses some of the statistical issues associated with nonstationarity of meteorological data. Anyway, I’m out of here. Hope somebody found this of use.

    [Response: It appears that you have been inadvertently misleading us. You claimed at the outset to recommend, and have repeatedly referred to, a maximum likelihood estimator. But it appears you were using a maximum-likelihood-TYPE estimator, more properly referred to as an M-estimator. Hence my repeated query about why you got a different result using least squares and maximum likelihood assuming Gaussian white noise, when in fact they are equivalent. All the while you have attempted to ridicule least-squares, in spite of the fact that it too is an M-estimator, albeit a poor choice for robustness against outliers. Once again, I can only respond to what you say -- not what you meant to say.

    Least-absolute-difference is indeed more robust against outliers than least-squares, but at the cost of reduced precision in the absence of outliers. For the sample data you generated it amounts to estimating the slope by using only two of the data points! For example, when the outlier is at t=0 it's equivalent to estimating the slope using only the data at t=1 and t=4; hence your estimate of the slope as 1.97. A considerably *better* way to estimate the slope is to detect the outlier (which is rather easy since it's rather outlandish), eliminate it, and apply least squares. This gives an estimated slope of 1.995, reducing the error by a factor of 6.

    Your continued insistence that non-normality and correlation invalidate the use of least squares would be laughable if you weren't, as I said before, intransigent. You also don't admit the error of your ridiculous claim that one can only compute meaningful error ranges with Gaussian white noise. You just don't know beans about the application of least squares regression in the presence of autocorrelation.

    In terms of real data, your most offensive suggestion was that the temperature data from winter 2000 are somehow "outliers" that should be eliminated, and are solely responsible for the strong warming trend in GISS data from 2000 to the present. In truth those data represent legitimate estimates of the actual temperature during winter 2000; it really was colder then than in succeeding times, and eliminating that data amounts to cherry-picking. But apparently you don't like that data because it undermines your suggestion that we may not have seen confirmable warming since 2000.

    Your choice never to use least-squares again is entirely up to you; you'll be less susceptible to outliers but consequently far less precise in their absence. As for your decision not to return here ... you won't be missed.]

  • Hank Roberts // January 8, 2008 at 2:27 am | Reply

    You’re sure he’s a physicist? Just asking.

  • Jack // January 8, 2008 at 7:01 am | Reply

    Tamino

    It’s interesting that someone who claims to have an open mind obviously does not and calls those who disagree with the tenuous AGW hypothesis “denialists”. You seem to be in state of extreme denial yourself.

    I presume this comment won’t get past your self-serving “moderation” either.

    Never mind – your blog is probably has a destination with continued obscurity.

  • dhogaza // January 8, 2008 at 8:56 am | Reply

    He references “Numerical Recipes in C”, wikipedia and a textbook in support of his notion that he knows more about the subject than a professional mathematician who has done theoretical work in statistics and makes his living doing time-series analysis.

    So my bet’s that he’s a hacker, perhaps one with a BS in physics, who works writing applications to analyze meteorological data (judging from other things he says).

    I think his work is a good candidate for auditing, don’t you? weatheraudit.com, anyone? :)

    His arrogance certainly fits the profile of the software hacker/engineer self-proclaimed expert on science that plagues discussions regarding evolutionary biology and global warming.

    I’m sure he’s off proclaiming victory on some denialist site or another, telling all who will listen that tamino is another “warmer” who doesn’t understand statistics.

    [Response: I doubt it. I suspect he's a working physicist who has made good use of some pretty sophisticated statistical tools and has some strong preferences, but he somehow thinks he knows better than the guys who *invent* these methods. I really do meet these people all the time.

    Speculation on such matters doesn't really get us anywhere.]

  • Martin // January 8, 2008 at 11:24 am | Reply

    Tamino you say about global warming that “statistically valid evidence that it has stopped simply doesn’t exist.”

    Is that not a little strong? I would certainly agree that there is no statistically valid proof that it has stopped, but evidence?

    Using the anaology of a criminal trial, I would certainly agree that the balance of the evidence is still in favour of global warming trend continuing – but that there is no longer proof of this beyond reasonable doubt. That is because there is now some years of evidence to support the proposition that world temperatures have peaked.

    The HadCRU data shows the five year moving average beginning to point downwards, and they predict the very slight downward trend to increase this year. Is that proof that the world is no longer on a warming trend, or that we will not soon exceed the previous high? Certainly not. But surely it is statistically valid evidence against the proposition that we are warming at around 0.2 C per decade? As a statistician you will be able to calculate how many more years of steady temperatures would be needed before that evidence became strong enough to be considered proof of the case that the global warming trend has paused.

    I understand that the precautionary principle may mean that we would be wise to continue to reduce CO2 emissions even if there is no increase in mean global temperatures for many years to come, so it is perhaps just a technical point that I am making.

    [Response: I agree that there certainly isn't proof beyond doubt that global warming hasn't stopped; the data since, say, 2002, are not inconsistent with a reversal of global warming.

    But the data are far more consistent with a continuation of global warming. As this post shows, it's not at all unusual that we'd see as much as a decade of declining temperature, *even if* the trend were uninterrupted at its present rate, given the level of noise in the data. In fact from a statistical point of view it's impossible *not* to.

    It's kind of like a casino spinning a roulette wheel and letting players bet on black or red -- with 0 neither black nor red so the house wins. The house edge is modest -- a mere 51.4% to 48.6% -- so players believe they have a decent chance to win, and indeed they do. They have a lot of fun playing, and occasionally somebody gets lucky and wins a huge pile of money. This causes quite a stir in the casino, with people gathering to watch the lucky player, drinks all around, raucus cheers for every win, and whispers propagating that "table 3 is hot today!" I say, "statistically valid evidence that table 3 is hot, simply doesn't exist." In fact, if you *didn't* get streaks of luck from players, you'd know that the table is "fixed" and the house is cheating.

    The house edge, though small, is plenty big enough for the casino owner to win in the long run, even after paying employees' salaries, maintenance costs, paying off the mortgage, hiring Wayne Newton to perform, and giving the local mob boss his cut. Casinos actually love it when somebody gets lucky and wins a pile of money. It creates the impression that others can win big, and brings lots more people in to play. All the while, the statistics of the house edge makes them more and more money, more than compensating the payoff to the lucky winner.

    And bear in mind that HadCRU is only one planetary temperature estimate, which I consider less representative than GISS because it doesn't estimate the arctic (the planet's fastest-warming region) as well. In fact both GISS and NCDC estimate that 2005 was warmer than 1998, and GISS is almost sure to place 2007 as the 2nd-warmest year ever, moving 1998 down to *third*. In fact for the climatological year (December through November) 2007 has already moved into 2nd place on the GISS all-time list. Don't get fooled by HadCRU hitting on a ten-year streak! We're betting the house, the car, hell the whole damn planet, and in the long run the odds are strongly against us.]

  • P. Lewis // January 8, 2008 at 3:01 pm | Reply

    but that there is no longer proof of this beyond reasonable doubt. That is because there is now some years of evidence to support the proposition that world temperatures have peaked

    There are always peaks and troughs.

    HadCRU3T 5-year average temperature plot

    ~1915, ~1920, ~1926, ~1932, ~1940, ~1950, ~1960, ~1971, ~1982, ~1989, ~1999 (El Nino artefact), ~2004.

    What do they have in common? Short-term (based on 5-year average) transition through dT/dt = 0 to negative values.

    Causes of dT/dt –> 0 (TSI ~ constant, +/- ~1%)? Eruptions/aerosols, La Nina, El Nino (why? occasionally large outlier, returning to trend), weather effects (up to multi-decadal) due to positive/negative indices of oscillations like NAO, AO, …

    Trend: ~+0.2°C/decade over roughly 40 years. Forecast trend out to at least 2014: ~+0.2°C/decade.

    Analysis of a few years’ figures, besides being statistically dubious climatologically speaking, must take account of short term “sub-climatology noise” when seeking to make attributions.

    Conclusion: there have been multiple periods of short-term cooling or relative stasis for nigh on 100 years, and there will continue to be so. And still the temperature trend is up.

  • George // January 8, 2008 at 3:48 pm | Reply

    Excellent casino analogy, Tamino!

    What makes absolutely no sense about all this talk that “global warming stopped in 1998″ are the facts that CO2 causes warming (ie, it is a greenhouse gas) and that the atmospheric CO2 concentration has continued to increase since then. Even skeptics like Lindzen acknowledge these facts.

    In the absence of some permanent counterbalance (causing cooling), warming from greenhouse gas increase must continue, eventually.

  • Climate Projection at Rearranging the Deckchairs // February 5, 2008 at 12:29 am | Reply

    [...] of 1998 (a record year) as starting point, that claim is garbage. However as Tamino points out: garbage is forever. digg_url = [...]

  • Douglas Watts // March 3, 2008 at 10:06 pm | Reply

    HYPOTHESIS: Nobody has hit .400 since Ted Williams. Therefore, the quality of baseball hitters hasdeclined since Ted Williams.

    See: “Full House” by Stephen Gould for a great exposition on this fallacy.

  • Nick // July 8, 2008 at 3:10 pm | Reply

    So if the length of period improves the estimate, lets go back 1000 years and fit a curve to that. An even better estimate of global warming.

    Of course there is an obvious fallicy involved here, but having a very long base line illustrates it.

    With lots of data the trend is flat. No global warming.

    Now any GW advocate will tell you what the issue is. They will tell you that its all changed. There is a kink when CO2 took off.

    It’s the same as the skeptics, and I’m skeptical. Since 1998, the trend is down or level depending on which source you take. I personally favour RSS because its pretty clear of biases. OK, it doesn’t cover above 85 degrees, but when it comes to the arctic there aren’t alternative measuments either.

    The claim is that there is a kink in 1998, just as the proponents claim there is a kink in temperature before hand causing a rise.

    Another interesting point relates to the drop in temperature since 1998 when measured peak to low. It has to be all natural

    [Response: The trend in RSS data *since 1998* is -0.005 +/- 0.03 deg.C/yr, i.e., the underlying trend is between -0.035 and +0.025 deg.C/yr. Even starting with a record-large el Nino ending with a la Nina, you still don't get a significant cooling.]

  • Paul Hagens // December 9, 2008 at 9:28 pm | Reply

    I am not a climate scientist, but I know that few of the trends in this world, including climate are truely linear. Therefore linear trends fitted on data often give a false impression of the future. This whole fierce discussion on these linear trends to me appears a bit silly if you do not link it to underlying trends and causes. The discussion should not be about the linear trend, but about what drives the temperature variations that are observed. If 1998 was an El Nino year and had such significant influence on global temperature that it has almost canceled the upward temperature trend, the first questions that should be asked is “how does this work, why does it have such a large influence, etc.”.

  • David B. Benson // December 9, 2008 at 11:20 pm | Reply

    Paul Hagens // December 9, 2008 at 9:28 pm — It just so happens that excess CO2 warms the globe in such a way, and humans have been producing excess CO2 is such a way, that an aproximately linear trend is physically correct.

    For the purpose of understanding the trend the smaller, super-imposed, variations are not terribly important, abeit quite interesting.

  • dhogaza // December 10, 2008 at 8:03 am | Reply

    If 1998 was an El Nino year and had such significant influence on global temperature that it has almost canceled the upward temperature trend, the first questions that should be asked is “how does this work

    It redistributes heat from the ocean to the atmosphere

    , why does it have such a large influence, etc.”.

    The ocean holds a lot of energy, and the change in ocean circulation that characterizes “El Niño” pushes a lot of it around.

    ENSO is actively researched, but it’s not particularly interesting in regard to the AGW issue. Natural variability happens. AGW happens. AGW may lead to natural variability making wider swings, but natural variability isn’t going to impact the basic physics underlying AGW at all. Just makes it harder to separate signal from noise.

  • Paul Hagens // December 10, 2008 at 8:12 pm | Reply

    Re: “humans have been producing excess CO2 is such a way, that an aproximately linear trend is physically correct”.
    I do not see your point. What do you mean by physically correct? Is this because CO2 must be the only driver and there cannot be any feedback loop or any other mechanism involved? That seems to easy to me.

  • Paul Hagens // December 10, 2008 at 8:23 pm | Reply

    Re: “ENSO is actively researched, but it’s not particularly interesting in regard to the AGW issue.”
    Why?
    AGW seems to happen. However, it is not the cause, but the result of something else or perhaps many things.
    If ENSO has such a significant effect as to cancel the trend almost completely, than understanding how ENSO influences global temperatures must be worthwhile. Perhaps we have even missed something else, that we haven’t measured yet. For example are global temperatures hiding, large regional differences (between poles or between continents) and what does that mean?

  • Ray Ladbury // December 10, 2008 at 8:42 pm | Reply

    Paul Hagens, Large volcanic eruptions will also obscure the trend. However, their effects, like those of the phases of ENSO are short in duration. It is interesting to be sure. It doesn’t pose a threat, so while it generates interesting research, it doesn’t generate controversy.

  • Ray Ladbury // December 10, 2008 at 8:44 pm | Reply

    Paul Hagens, I would recommend going to the website realclimate.com and clicking on the “start here” button. You will find lots of resources for answering your questions.

    [Response: It's at http://www.realclimate.org/]

  • Hank Roberts // December 10, 2008 at 9:04 pm | Reply

    Try here, Paul. Note the “___ is in the gaps” argument is commonly used, but what science does is making the gaps narrower and brighter all the time; nobody has found a big unexpected fact lurking in a gap in recent memory, in any area where those opposing the science call on the gaps as hiding possible hidden forces.

    Try here:
    A Tutorial on the Basic Physics of Climate Change
    Email | Print

    The following article has not undergone any scientific peer review, since that is not normal procedure for American Physical Society newsletters. The American Physical Society reaffirms the following position on climate change, adopted by its governing body, the APS Council, on November 18, 2007: “Emissions of greenhouse gases from human activities are changing the atmosphere in ways that affect the Earth’s climate.”

    By David Hafemeister & Peter Schwartz

    Abstract: In this paper, we have used several basic atmospheric–physics models to show that additional carbon dioxide will warm the surface of Earth. We also show that observed solar variations cannot account for observed global temperature increase.

  • Hank Roberts // December 10, 2008 at 9:04 pm | Reply

    Link for the APS tutorial:
    http://www.aps.org/units/fps/newsletters/200807/hafemeister.cfm

  • Jim Rennison // January 4, 2009 at 4:43 am | Reply

    The AGW hypothesis rests on ONE IDEA; that CO2 changes the ’set point ‘ of water vapor in the atmosphere. This is the ‘positive feedback that AGW promoters insist dominates. Without this (alleged) effect, the very idea that small changes in the trace gas CO2 could somehow leverage the temperature of the entire atmosphere is simply absurd. But with this understanding, it is plausible, because CO2 then becomes responsible for ~25% of the total greenhouse effect rather than a paltry 2%.

    But , is it true? Is this the way CO2 and water vapor interact? Remember that if the AGW people have this wrong, then the whole case against CO2 collapses.

    [edit]

    [Response: You've given us an outstanding example of crossing the "stupid threshold."

    AGW does not depend on water vapor feedback to exist -- Venus has no water vapor but it's still a hothouse planet due to the greenhouse effect of CO2. The numbers you quote are ludicrous. You're just vomiting stupid ideas you've heard, with no intelligence or critical thinking applied.

    The claim that AGW rests on ONE IDEA which you you can oh-so-easily refute, is one of the dumbest "straw man" arguments around.

    I'm not the least bit interested in hosting your idiotic quote from an interview with an idiot, or your link. The only reason I even allowed any part of your comment is to alert all readers that the "stupid threshold" is now in force. Bye-bye.]

  • Ray Ladbury // January 4, 2009 at 12:53 pm | Reply

    Tamino, Thanks for that save on goal. Jim doesn’t seem to understand that most feedbacks depend on temperature. If you change them for CO2, you change them for ALL forcings–and when you do, your model of Earth stops looking like Earth. I suspect that he is just regurgitating something he read on a denialist website, though I haven’t been able to pinpoint the denialist mothership at which it originated.
    This seems to be a common meme–that somehow the feedbacks are fudged to get a greater warming from CO2. Flat-assed ignorant and stupid.

  • dkodpe // January 27, 2009 at 7:56 pm | Reply

    Tamino

    George Will has jumped on the bandwagon with his 1/17/09 column in Newsweek.

    Here’s a link to a post I made on his temperature trend assessment.

    link

Leave a Comment