Jun 23 2008
Warming on 11 Year Hiatus? Et al.
This is another post on a meme that just won’t die. That would mean it’s about finding a linear fit of a short, noisy time series, right? That’s what started it. Further blogging revealed that ignoring the temperature data from GISS and NCDC, which coincidentally both show warming, probably isn’t a good idea. But nope, that’s not what really got me.
That’s going to be a yearly thing for a while. Last year it was cool to be finding 10 year trends. Now it’s going to be 11 for a bit. And next year it will be 12. For 10 year trends, the excuse was that it was a nice round number. I’d like to hear the excuse for this year. Perhaps, “I needed to include the large El Nino in 97/98 in order to get a negative trend”. Although, “11 is a round number since I count in base 11″ would sound a little better.
Nope, that wasn’t it. What got me was this comment:
I know there are people who love to hate GISS Temp. I understand the arguments both for and against it.
There is no argument against using the temperature data from GISS. None. Period.
To illustrate this, I’ve plotted the temperatures since 1880 from GISS, Hadley Centre, and NCDC. Your job, is to figure out which one is GISS. And then explain why its temperature data are invalid yet the other 2 are perfectly fine to use. I give the answer below, so no peeking.
|
Since the red line starts slightly lower and ends slightly higher that the black and blue lines that one must be from GISS. Right? Since their algorithm warms the present and cools that past. But you’d be wrong. Red is actually from the Hadley Centre. The black line is GISS and the blue line is NCDC.
These 3 time series come from basically the same data; surface stations and sea surface temperature measurements. And while there may be biases in the data, it is not fine to just ignore 2 of them and focus on the 1 that fits.
(No Ratings Yet)
Loading ...Related Posts:
41 Responses to “Warming on 11 Year Hiatus? Et al.”
To reduce spam, comments are automatically closed 30 days after the last comment. If you would like to comment on any closed thread, please use the contact form at the top of this page.
Nathan,
Why have you nothing plotted beyond 2002? Current data exists you know. Following your “it is not ok” line, I’d say that applies to showing current data too, particularly when you criticize the 11 year period, but then don’t plot a good portion of that 11 year period data.
An expanded Y scale might be helpful also. Your graph presentation is compacted. Even so, it is not surprising to see good fits between NCDC, GISS, and HadCRUT data, since GISS and HadCRUT both use the NCDC station data as a base input. Nothing profound there.
It’s plotted through the present. Each horizontal tick is 5 years.
Ah, my bad.
Proof positive I need to get a new (and bigger) monitor.
As I mentioned, “Your graph presentation is compacted.”
Wordpress likes graphics small. I solve that problem on my blog (for other readers and for myself) by offering larger versions of the graph as click throughs.
Being able to see your graph full size would be helpful to prevent such things. Just a thought.
[Reply: Good idea. I'll add a link to a bigger plot soon. Busy with work at the moment.]
[Deleted. See comment policy.]
Nice post- no matter how many times this point gets emphasized, as long as those in the contrarian sphere (no need to name names) continue to cherry pick HadCRUT and Sat trends in order to fit their “warming stopped” frame while bashing GISTEMP, I fear we won’t see an end to it.
Perhaps you and Tamino could combine for a giant mythbusting rundown on the alleged vs. actual differences in the three (or five if you count sat) data series and develop a little more on the ever-shifting goalposts to capture the 98 ENSO high starting point.
The latest is moaning about the baseline period that different groups are using. Your friend Watts and Mc Whatshisname have started. Pielke Sr and Jr are sure to follow. Color Eli unimpressed.
A trend is a trend is a trend is a trend. Regardless of the HUGE problems with the historical and continual adjustments to the GISTemp, NOAA and HadCrut data sets (not to mention UHIs of current data) and the decreasing global coverage of the current surface station data, one cannot deny that the recent trend is negative, whether you regard it as noise or an artifact of ocean temperature cycles.
[Reply: I sure as heck can deny it's negative. At best, you can say that within uncertainty limits that the recent 11 year cherry-picked trend is zero.]
I don’t buy into the GISS is outlyer argument. It’s different, slightly different. The rational thing to do is figure out the source of the difference in a clear documented transparent manner. in lieu of that I like lucia’s approach of just averaging all the series.
For me it’s tough because GISS has released more data and source code, albeit abysmally obdurate, while Hadcru plays nixon
with its secret tapes.
Atmoz–
The great uncertainty in trends is the reason I emphasize hypothesis tests rather than the accuracy of the trend itself. The uncertainty bound on the trend are large, but the difficulty is that currently, using a reasonable begin data to tests data from the AR4, the possible trends consistent with data simply do not contain 2C/century projected by the IPCC.
Tamino applied a very similar test to “falsify” nowarming this century in Garbage is forever”.
I should note his reaction to those who suggested the time period might be too short, or there could be any difficulty in doing such a test was rather splenetic.
One may, of course, ask legitimate questions about the validity of any technique, (though possibly not at Tamino’s site after he’s just used the method). But, the idea that one can perform valid hypotheses even when uncertainty is large is well accepted in statistical literature.
So, for example, when I first started on this, your question about ENSO was a valid question, JohnV’s questions about solar were valid, and a number of other questions are valid. But, my response is to see how much difference they might make and report.
I conclude ENSO can’t explain the deviation– but of course, I could be wrong for a variety of reasons. All anyone can do is explain what they looked at.
The fact that these are valid questions does not undermine the idea that one can do hypothesis tests with small amounts of data. The valid questions are: do the assumptions required for this particular test apply?
On the issue of picking start years: I agree that, in addition to selecting favorite (or least favorite) measurement groups, there is the difficulty of selecting start years tailor made to get a specific result.
My standard for testing projections in the AR4 is 2001 for reasons I’ve explained before. I know there are people who don’t like this start date. However, there are a number of factors in favor of this data– but one is that I can’t see the sense testing AR4 projections starting from a date that predates the publication of the TAR. So, that knocks 2000 out.
However, when responding to other bloggers, I do discuss their start dates and what I would find using those dates.
With respect to the 11 year trend: I don’t consider that trend flat. But some people do.
On the GISS thing: I think some of the things GISS does do appear to be demented (to use Steve McIntyre’s term). But, it appears that over the long term, GISS gets the same trend as everyone else.
I believe they will continue to get the same trend over time. In fact.. I believe it enough to bet a regularly publishing blogger with a reasonably decent temperature, and an individual Alexa rank less than 400,000 that the OLS trend for the six months from June-Nov for GISS will fall below that for the average of the four other sets I use. But, I won’t bet money. I’ll bet a dozen homebaked cookies.
With an Alexa rank of 387,276, you qualify!
You would be permitted to chose between “Temperature Anomaly Cookies” (to be described) and Chocolate chip based on the recipe on the Nestle’s package. (Or if you have some dietary issues, we could try to come up with other cookies that ship ok.)
Also, since this is for fun, and think it’s impossible to know which group will have the higher 6 month trend, I’ll take either side of the bet! The purpose is just to highlight who had the larger trend after 6 months and to make a stupid bet since that seems to be essential to good blog-viating.
Seems Atmoz, Eli, thingsbreak are are being a little hard headed here. Making the point that a 7, or a 10, or an 11 year trend is too short or is cherry picking is fair/strong/valid. Especially if somebody stated that “from the year 1998 to 2005 was a major cooling trend!” now that we’re in 2008 that would be a ridiculous thing to focus on.
However, when you are discussing the temperature record to date I dont see anything wrong with saying “temperature has levelled off, cooled or warmed for the last XX years.”
If XX years is too short to be significant OK, but it’s hardly cherry picking to temperatture to date measurements.
Radar
You’re missing the point that there is a clear attempt to bias the sign of the trend by continually moving the goalposts to take advantage of the outlying 98 ENSO event.
The “global warming stopped in 1998″ meme is a favorite of the denial sphere, and they are constantly changing the trend length and temp series used to keep it alive.
That the length s also too short to be significant in terms of a longterm trend is a separate issue.
[...] individual bloggers have expressed a strong preference for one particular data set or another. Like Atmoz, I prefer not to drop any widely used metric from consideration. However, because some individuals [...]
“Tamino applied a very similar test to “falsify” nowarming this century in Garbage is forever”.”
Reading the comments section for that thread is priceless! Look how confident all of those McIntyrites were that global temperatures were going to be lowered substantially, and thus disprove current warming! I remember those days fondly. Poor, credulous yobs…
Atmoz, perhaps you right. BUT we are not talking about uncertainty limits as those are highly subjective. Some would say that the uncertainty limits of the surface station datasets are grossly understated given the data manipulation and corruption (ie UHI) that goes along with them. The statements regarding recent negative trends are based purely on published monthly global mean temperatures (ie what Mr. General Public sees) of GISTEMP, HADCRUTv3, UAH and RSS data and using whatever means to establish a trend line over the last 6, 8, 10 or 11 years, take your pick.
Things break….
You are aware that if you repeat Tamino’s “proof” including the colder temperature in the recent data, there is no statistical warming since 2000, right?
That said, it’s perfectly valid to do tests of this sort and see what the current result is. It’s quite likely 0C/century since 2000 is false. However, it is no longer possible to show this using statistics.
The difficulty with short data sets and hypothesis test is that you generally get false negatives meaning you can’t prove things that are false are false.
Right now, the data since 2001 are showing 2C/century, predicted in the AR4 is false, but they don’t show 0C/century is false. Likely, both are false– but unless we can find a way to really deal with the weather noise due to identifyable features (like ENSO), the empirical data won’t exclude both wrong things simultaneously for several years.
This is just “the breaks” in terms of statistics.
But, if you believe Tamino was correctly applying that method last August, and you believed his conclusion was proven based on that analysis, you should believe the IPCC’s projection of 2C/century exceeds the current trend on the real earth based on analysis using the exact same method.
lucia,
I was referring to the comments section of that thread only, and reminiscing about the CA-fanclub furor that was convinced global warming would be overturned due to the slight adjustment of the US lower 48 temps. It was hysterical in both senses of the word.
I don’t find any particular meaning in short term trends for purposes of confirming or denying whether warming has stopped for such short periods. Neither would Tamino, I imagine. However his choice to refute someone’s contention that such a negative trend existed when it did not in no way is subsequently undermined by such a trend existing now. He simply disproved the claim made at the time because it was indeed false.
Thingsbreak–
Well… I didn’t suggest anything is undermined by the flip in the “reject” outcome. I said it wasn’t.
But, if you think Tamino doesn’t think the analysis he did last summer is meaningful, I guess you can think so. I’ve run across other admirers of Tamino who are deciding he didn’t mean many of the things he said about hypothesis testing that post written in August.
It is true the claim of “no statistically significant warming” was indeed false at the time–as could be shown using that method. Equally, the claim the warming trend is 2C/century since 2001 is false at this time, based on the same method.
The second is my claim.
Atmoz: Did you really say this:
“And next year it will be 12.”?!
They will only say this if there it supports “no warming since XXX”. I think you inadvertently predicted cooling!!!!
[Reply: If for some reason the temperatures in the next year jump dramatically (e.g. ENSO) such that a 12 year trend no longer gives a near zero linear trend, I still think some people will use it because it will give the least positive value.]
Thingsbreak,
Point about 1998 is well taken, my brain automatically throws out any article sensationalizing 1998 from either side. And by the way, there was lots of “warmest year on record” hoopla if I recall correctly.
But whether you start in 1999, 2000, or 2001 there has been a noticable break from the warming trend. Attributing it to “weather noise” - ok. But who gets to decide that 7 or 8 or 9 (or 11) years is noise affected by ENSO, while 17 years is climate? (Tamino?).
Plenty of peaks and valleys are used in calculations by both sides. Smooth out 1998 warming and 2000 cooling and things will be flat again. In my lowly little engineer brain 10 years out of 58 (”AGW warming since mid century”) is not “insignificant”. It’s interesting and notable.
Atmoz,
I was looking at the data regarding your point that GISS adjustments haven’t caused more slope than the other data sets and something else jumped out of the graph at me.
Reference Hacrut3, GISS, UAH, RSS:
http://www.woodfortrees.org/plot/hadcrut3vgl/from:1979/offset:-0.15/mean:12/plot/gistemp/from:1979/offset:-0.24/mean:12/plot/uah/mean:12/plot/rss/mean:12
UAH and RSS were the outliers at every peak, warm or cool, from 1979 - 2002 when the land/ocean sets suddenly caught up or in the case of Giss, even exceeded the satellites. Is there an explanation for this apparent shift?
[Reply: I'm not sure what "shift" you're talking about. For the most part, the in situ measurements have less wiggle to them that the remotely sensed measurements. But for the latest positive bump, GISS has the highest temperature anomaly. Is that what you mean? If so, I wouldn't read too much into it. GISS has been the highest temp before during a positive wiggle, circa 1990, and the lowest temp before during a negative wiggle, circa 1994. Right now, GISS has the high temp during a negative wiggle. With regards to climate change, these high frequency wiggles are not important. It's the long-term trends that matter. And all of the data show extraordinary agreement.]
Eli, thanks for being unimpressed, you are true to form, phantom rabbit.
Likewise, I’m unimpressed with cowards that don’t put their name to their words, as I the, Pielkes, and MacIntyre do.
Nathan,
Just an FYI after yesterdays’ label misread, I went to COSTCO and got a nice new 24″ LCD monitor. Hey, I can read your graph x axis now!
Glad you like the idea for click through to larger graphics.
[Reply: Whoops. Got busy yesterday and I forgot to make the larger graph.]
Just wanted to say that I enjoy this blog, and especially Lucia’s posts here and elsewhere. I truly wish there were more involved in this debate that could provide objective, dispassionate analyses as she does. What more could one do than to conduct statistical tests, show the results and discuss them in terms of the stated hypothesis! (well, and to show her confidence with such strong statements as cookie bets!) Kudos Lucia and thanks Atmoz for the blog work.
ps - also greatly turned off by those casting personal aspersions who don’t use their own name…I tried to explain this to my nine year old last night…he just didn’t get it and I can’t honestly say I understand either (he used the time honored term of “chicken”)
Atmoz:
Ahh.. yes. That will happen. We are due for an El Nino. The questions are: When will it happen? Are weak El Ninos really correlated with the PDO? Will the next one be strong or weak?
[...] individual bloggers have expressed a strong preference for one particular data set or another. Like Atmoz, I prefer not to drop any widely used metric from consideration. However, because some individuals [...]
ATMOZ,
Don’t forget Lucia is a luker. Very simply, this cold spell will pass.
get ready for great blogging when the next El Nino hits.
A weak short lived El Nino that is.
[Deleted. Comment policy applies to rabbets too.] Come on Tony there are good arguments to be made for just about any choice of baseline period and they have been made, and it don’t make no difference nohow as long as you RTFR.
1951-80 is not bad, global temperatures were reasonably stable, there was good coverage for the surface stations, etc. Sliding the scale is also not a bad idea. You publish, you choose.
Steve Geiger, I bring an unusual perspective to this debate in that much of my experience in what might be called environmental politics has been focused on things like reviewing environmental impact reports. One of the first lessons I learned is to avoid automatically accepting the assumptions of the authors regarding “big picture” stuff like the project descriptions and the selection of alternatives that must be analyzed. Often these choices are used to obfuscate the big picture (or at least portions of it that would be inconvenient to discuss).
You might ask yourself why climate scientists don’t place a lot of weight on the short-term behavior of surface temperatures. A couple of reasons that don’t get much discussion is that they are not a good proxy for heat content since the whole system isn’t being measured (referring to the deep oceans) and that there are other things going on with the ocean-atmosphere system that show no sign of slowing down.
Pulling back a little farther, an even bigger picture comes into view: It is undebatable physics that adding CO2 (and other GHGs) to the atmosphere must warm things. Then, looking at paleoclimate, we can see no sign of a convenient negative feedback that would tend to damp out the warming effects of the CO2 and that the CO2 levels we are headed toward are incompatible with e.g. the continuation of the deep glacial period that is the only climate state our species has ever known.
I should add that this is why the big question in climate science is how fast this process will proceed. Interestingly, as recently as a few years ago it was assumed that the shift into a new climate state could only proceed relatively slowly (on a scale of centuries). Now that’s starting to look like a poor assumption.
Steve Bloom, at the risk of a complete thread hijack, just a couple of items. I agree it is well established that addition of GHGs to the atmosphere will have an initial warming effect. The degree of the effect seems much less well understood. I’m no expert on paleoclimate, but what I have read leaves me thinking that there remains much room for debate about the paleo record (even up through the most recent few thousand years). On the larger scale, it seems intuitive to me that there must be some sort of feedback mechanism that limits heating and cooling phases.
As far as perspective, I have an advanced degree in an engineering discipline, am a practicing professional engineer and commonly encounter very detailed critisims about many facets of my work. From my life experience I believe that you can tell quite a bit about the efficacy of an argument based on how it is presented. A few conclusions I’ve drawn RE the climate blog-o-wars; 1 - the bloggers whom tend to limit personal criticisms and promote exchange-regardless of whether they support or counter their current arguments, gain in credibility in my eyes; 2- bloggers that take detailed measures to present the counter argument to their positions (i.e., like posting verbatim what the other has stated) before presenting their rebuttle also gains a lot of credibility, IMO. Posting anonymously doesn’t take away from ones argument, per se, but is rather anoying when the arguments turn personal. Just my two cents.
Cheers.
“Likewise, I’m unimpressed with cowards that don’t put their name to their words, as I the, Pielkes, and MacIntyre do.”
I agree that you guys are brave to put your names to the things you write.
Steve Geiger, of course climate can’t run away indefinitely. After a while a new stable state is reached as long as the forcings aren’t increased more. My point was that there’s no reason to expect present climate to behave differently in that regard than past climate. And recall that Venus is now in a very stable state (not that we’re at risk of such a thing, of course, although note that paleoclimate gives us good cause to be confident about that).
Re paleoclimate, it sounds as if you’ve mainly heard about our friend the ice sporting implement. Interestingly its importance to this discussion is approximately zero.
For a broad background, Chapter 6 of the AR4 WG1 report is good, but I would also suggest these articles:
The surprisingly fast and violent end of glaciations (new results)
An overview of Cenozoic climate, including what happens when a really big slug of CO2 gets into the atmosphere
How CO2 and temperature track each other over the long-term, and what that tells us about sensitivity
There’s lots more and the foregoing isn’t a perfect selection, but it’s enough to give you an idea. Then probably you should have a look at Hansen’s recent stuff since this material is at the heart of his concerns.
Also of interest is this review article on the current measured movement of the climate system.
Finally, speaking of big slugs of CO2, we have some current prospects in the Arctic Ocean.
As Wally Broecker says, “Climate is an angry beast, and we are poking it with a stick.”
Anyway, this stuff never gets talked about on denialist/”skeptic” blogs. It’s boring, I suppose.
Regarding your item 1, bloggers with good self-control but hidden agendas can take advantage of that commendable attitude. Sometimes there’s pretty good transparency as regards agendas (Anthony Watts e.g.), but with others it’s less clear. This combined with the long and unpleasant history of fossil fuel industry-funded disinformation efforts (not limited to the blogosphere, of course) can lead people to be suspicious when they see a blog with an excessive focus on details that are purported to undermine the science in some substantial way. Eli in particular has spent a lot of time locating the names of climate change denialists in the tobacco archives, so his ears are acutely tuned to detect this sort of thing.
S. Bloom - in the larger pictures, I don’t really see how the winter-sport-hitting-implement can not be important to our understanding of what’s going on currently.
If some bloggers are tainted by ‘oil money’, or whatever, and providing information they truly don’t believe or are knowingly deceiving others, then that’s despicable. As far as I’m aware I do not read any of those folks’ stuff. I have heard (first person) accounts from folks in academia that they have a much less chance at publishing results that do not show ill effects associated with AGW.
What I really like (and hopefully benefit from) is when people of some standing (and opposing view points) can actually have a dialog via a blog. It seems rare, and usually degrades quickly due to the background chatter, but I’ve caught a few ‘arguments’ that have been civil and provided ideas and counter ideas on some pretty interesting GW-related topics.
Thanks
Why can’t I be the Phantom Rabett? Eli rather likes the idea of gliding through the blogs in cape with carrot,
Little Eli Foo Foo,
Hopping through the forest
Scooping up the cooling mice
And boppin’ ‘em on the head
Down came the good fairy and she said
“Little Rabett Foo Foo,
I don’t want to see you
Scooping up the fooling mice
And boppin’ ‘em on the head.
I’ll give you three chances,
And if you don’t behave
I’ll turn you into a goon!”
. . . . .
Down came the good fairy and she said
“Little Rabett Foo Foo,
I don’t want to see you
Scooping up the Exxon mice
And boppin’ ‘em on the head.
I gave you three chances
And you didn’t behave
Now you’re a Phantom! POOF!!”
The moral of the story is:
HARE TODAY, PHANTOM RABETT TOMORROW
Just out of curiosity, Steve Geiger, why do you think the HS is important to current climate, or perhaps more to the point the climate of the next century or so?
I should add, not wanting to be too mysterious about it, that the reason the HS isn’t very important IMHO is because knowing whether there was an MWP more or less as warm as today doesn’t in itself tell us anything useful about the likely future course of climate. OTOH at one time it was very useful in implicitly answering the ignorant question “How do we know it hasn’t been as warm or warmer before due to natural causes?”
Also, you say “I have heard (first person) accounts from folks in academia that they have a much less chance at publishing results that do not show ill effects associated with AGW.”
This seems odd given the vast and constantly expanding number of papers getting published. Most climate change papers don’t even find effects that could be characterized as ill or not ill. Or do you mean papers that purport to overturn major aspects of the science? Perhaps Atmoz could weigh in on this, but I really don’t see even that as a problem for authors, although I’m sure it is often a problem in terms of getting grants for subsequent work since in the end most such papers are considered to be poor science. Atmoz has blogged on a number of examples; this one is my favorite (and a fine, rare and oh-so-richly deserved scientific flame it was, too).
Steve B. - if the HS is an accurate portrayal of recent temp history, then, of course, the current trends (lacking any other yet identified causes) is much more certainly linked to atmospheric composition, IMO. If current trends are not *that* unique even in recent times, then it is much more likely that our understanding is not as complete as we would like. The question of GHG sensitivity, of course. Now, I assume you knew what my answer would be….so let’s hear your response
Its funny, I strongly agree that scientific consensus should be very hard to overturn. However, I also think perhaps with the theory of AGW that ‘consensus’ was perhaps achieved a bit too easily (and, of course, during a time that temps were/are on the rise). Some of this, I think, might be related to how science is conducted in this day and age. Just a scan of the literature reveals a litany of papers that reach for any relationship to global warming/climate change…the institutional train has left the station and has a lot of momentum, this aspect is divorced from the real science and only a function of how science is currently funded/supported.
The research I alluded to was not on climate per se, but the effects of possibly increased temps on something else. The researcher, a friend of mine from an ivy league university, clearly stated that he was disappointed in his findings to the extent that it would be harder to get published (and possibly future funding along those particular lines) since the effects of the increased temps and CO2 concs tended to reduce the risk of the particular phenomenon he was evaluating.
Thanks
Thans for the reply, Steve G.
GHG changes are not thought to have played anything other than a very minor role (as a feedback) during the pre-industrial global T changes of the last 2000 years. The drivers were some combination of TSI and vulcanism (mostly the latter). So, other than the minor confirmation of knowing to the extent we can that the behavior of GHGs in this minor feedback role is consistent with our understanding of the physics (which goes to the issue of sensitivity), scientific conclusions as to the role played by anthropogenic GHGs in the present warming have to be based largely on other evidence. The physics plus modern observations are actually enough for that, but knowing how GHGs have behaved (as forcings or feedbacks) in the context of past large climate changes is obviously a big help.
Put another way, a warmer-than-present Medieval Warm Period would simply mean that there was some combination of less volcanic dust/more TSI and wouldn’t say much of anything about GHGs. There would be something to your argument if GHGs at levels greater than present had been the remain driver of such events, but that wasn’t the case. (BTW, I should mention that one of the problems in pinning down global temps during that period is because the main player was vulcanism, which is basically a bunch of regional events. Also, our main means of knowing about volcanic activity is ice cores, most of which are poorly located for this purpose.)
The MWP aside, it’s easy to find periods of much greater warmth in paleohistory. For example, the mid-Holocene may well have been a little warmer and the mid-Eemian (about 125000 kya) was substantially warmer (by as much as 2C IIRC). We had better be able to explain why, and in fact we can. OK, those two were easy (Milankovitch cycles), but generally the problem with past climate isn’t explaining how warm or cold it was but interpreting the proxies to figure out what conditions actually were (in terms of both T and forcings). That said, it will be interesting to see how the modelers do with those new abrupt deglaciation results (linked in a prior comment).
I agree that the institutional train has more or less left the station, but it took rather a long time getting up the needed head of steam. Read The Discovery of Global Warming (book-length, I’m afraid) to see how things actually went. The consensus is about 20 years old, BTW. Next year on Hansen Day I suppose I’ll have to have a drink to celebrate its majority.
Re your friend, it sounds as if his problem may have been that he was expecting his results to be more exciting and they weren’t. There are an awful lot of papers finding no impact of AGW on this or that, but they don’t tend to make the covers of the journals. If on the other hand your friend had a paper over-turning important past results, editors love that sort of thing.
(Erratum: In my prior that should be 125 kya, i.e. 125,000 years ago. 125,000 kya things were rather warmer.)
I must admit that I don’t put faith in any of the global trend charts (except the sat based ones), for several reasons. I suppose it is my Engineering background.
Reason #1 - signifcant digits - In the real world of actually making things work we require measurement techics that have 10 times the precision we are seeking in the results. Naturally this is luxury, but the idea that by averaging together many measurements you can somehow determine average temp to a greater level of precision that the initial data is fundamentally flawed.
Reason #2 - Before you can even start looking at measurement results you need to prove out the measurement systems using technics such as gage R&R studies and bias & linearity analysis. A quick look at surfacestations.com makes it clear that the much of the historical record is unredeemably polluted.
Years of experience making statisical predictions that end up getting proven or disproven based on subsequent events (as AGW will be) have shown me that errors are rarely a result of the statisical methods chosen, the problem is generally a result of the quality & quantity of data available for analysis. Of course, the hockey stick shows that methodology can result in meaningless results as well.
I would really like to see a trend chart for North America that was developed from the RAW data from rural stations that are class 1 or 2 and are in pretty much the historical setting. I would like the results of each station to be reduced to the deviation from their 1900-2000 average, which could then be averaged to get an overall average. By using anomalies for each station, and only using stations that are in consistent environments i would find the results more acceptable.
I find it absurd that GISS uses stations whose likely error exceeds the signal (anomaly) being studied, and the idea that future temperatures require adjusting the past is without merit, particularly when the assumption is that climate is changing. If the climate is changing, then future data can’t possibly help to “fill in the blanks” in past record.
Dave, it’s interesting how those years of experience in statistics somehow left you ignorant of the law of large numbers, as you amply demonstrated in your point 1. But then maybe they don’t teach that in Engineering school.
Hmm, did Steve G. go away? Perhaps he had no further questions.
Steve Bloom
Yes they do teach that in engineering school but they also teach you that you often need thousands of numbers and you need to plot them all out to actually demonstrate that enough numbers were there so that a normal was achieved. In engineering we don’t just assume things, we prove them. A few hundred items of data in fact is very often not enough - it totally depends on the scatter: The higher the scatter, the less justifiable being the assumption of a gaussian.
I see a lot of climateers making unjustifiable gross assumptions regarding gaussians, including assuming that even a tiny handful of data can be considered a gaussian, and then putting completely bogus confidence intervals on this sparse data. The recent blog argument about wider spreads in the data increasing our confidence in the models tells us exactly how much you guys know about statistics. Did they not bother to teach you what the y-axis on the plot meant? While I’m at it, did they also teach you that it is extremely unlikely that all instrument errors go in the one direction?
And while you are checking those impact assessments, tell us what percentage of them compare the unremarkable regional data with the alarming model results and how many discuss the fact that the models are actually no good at all for regional predictions? This bad practice cannot happen with engineering models because a) people die when you get things wrong, and b) we have several internal and at least one external review of every report we produce.
Atmoz
Regarding why 11 years is important - it’s purely because it is before the 1998 hot year, though why some people make a big issue out of that I fail to see - when doing averaging, it is balanced out by the 1999/2000 cooling and in the GISS plot 2005 apparently was just as hot anyway. I’m glad you admitted it was a flat trend - a lot of people are still in denial about that one. But to tell us an 11 year flat trend is obviously weather noise when you cannot actually say what this noise consists of is clearly hand-waving. And to use that argument while then ignoring the conclusion that a significant portion of the warming up to 1998 (in the non-satellite data) then must have been noise too is obvious bias.
Furthermore Tamino did his trend plot knowing full well he should balance out el niños with la niñas but he chose not to do so. Hence trying to say the statistical trickery is one-sided in this debate is total nonsense. Why are 18 or 30 years the magic numbers? Because by pushing the date out nobody yet has to face up to having been utterly wrong.
The sheep believe it all of course but more independent minds see it as a cheap sales trick. And I say that as someone who does believe we should be moving away from fossil fuels to greener alternatives but who is sick and tired of ego-driven scientists over-hyping the state of the science when they clearly still haven’t much of a clue.
Yes, JamesG, engineers know oh-so-much-more about time series analysis than mathematicians. You betcha.
For a discussion of the issue from someone more qualified than Tamino (in that he knows both the math and the climate science professionally), see here. If after reading it you still have some questions he might even be willing to answer them.
Better yet, he’d like to bet you some serious money with regard to your wacky engineer cooling idea. Oddly enough it’s been hard to find “coolers” willing to put their money where their mouth is, so James will be very pleased to make your acquaintance. Tell him I recommended you, and please do report back and let us know how much you’ve staked.
If you reserve another $50 or so I might be willing to bet you too after I collect on my sea ice bet in a few months. (One bet at a time for this kid.)