Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
If you haven't been watching the Roger Pielke Jr train come of the rails and the carriages smashing into each other and exploding, I suggest you look at this post from James Annan:
Roger Pielke has been saying some truly bizarre and nonsensical things recently. The pick of the bunch IMO is this post. The underlying question is: Are the models consistent with the observations over the last 8 years?
Hey, hypothesis testing. First year stats stuff. So Annan carefully explains how it's done.
Marvel at Pielke Jr's response:
All he does is draw some graphs and wave his hands around. Does he calculate the probability of getting such a result if the null is true? Hell no.
Even if the null hypothesis was true, there would be about a 21% probability of observing data this "extreme". There's nothing remotely unusual about it.
And that's using Pielke Jr's cherry picked starting point ...
Update: More carriages exploding here.
Environment
Politics
Comments
Actually it's not Pielke, but rather an engineer (big surprise) named Lucia who's the original source for this heaping pile of poor analysis.
She claims to be teaching herself statistics ... while simultaneously convinced she's on to something that professionals like Gavin Schmidt, James Annan, and the professional statistician who blogs as Tamino can't debunk. Despite their patiently having done so over the past month or so since she first came up with her "I've proven the models wrong with a 95% certainty" baloney.
Watching the show is sort of like watching professional evolutionary biologists try to set a creationist engineer straight on science. Pathetic, in other words.
Posted by: dhogaza | May 16, 2008 8:23 AM
Perhaps the most interesting thing about the latest thread at Pielke's is watching JohnV try to patiently pin Roger down on whether or not he agrees with Lucia's claim of having falsified the IPCC projections to the 95% confidence level.
It's interesting because JohnV began his climate blog commenting career, if you will, at climate audit, as at least a moderately enthusiastic kool-aid drinker.
But the scales quickly fell from his eyes as he began to see just how dishonest that crowd is ...
Posted by: dhogaza | May 16, 2008 8:38 AM
It's also interesting to watch Pielke weasel his way through the subject. At this point, rather than simply state that he doesn't agree with Lucia's assessment, he's claiming that statistical consistency is too weak a test to be meaningful anyway ...
He's steadily moving the goalposts, in other words, to avoid admitting that his past praise of Lucia's work was perhaps a bit premature.
It's also a bit hilarious to see him talk about how the analysis various depending on whether one starts at 2000 or 2001, or whether or not one includes April 2008 data (lucia's first analysis ended in March, as April data wasn't yet available).
He seems to miss that he's only demonstrating how wildly variable weather is over the short term, and how silly it is to compare with projections computed over many model runs (and several models) in order to SMOOTH the equivalent of weather variability churned out in individual model runs.
Posted by: dhogaza | May 16, 2008 8:49 AM
Without wishing to buy into the Annan/Pielke gerfuffle I note a commentator on the Pielke blog criticises the Koutsoyiannis study on the basis its regional samples are not representitive because they do not reflect boundary regions exempt from ENSO and MAO. Which begs the question; are any regions exempt from these phenomena?
Koutsoyiannis's study is over an 18 year model retrospectivity (not the 8 year period which informed the Annan/Pielke business) and uses 100 year temp and rainfall records as a basis for empirical comparison. His study also looks at a sample from every climatic zone. How much more representitive can you be?
Surely, given the current bit of discordant 'weather variability', the test of a model would be whether it predicts any period, of any length, in the future having a flat or declining temp; if it is all up you'd have to suspect that model; which begs the question; when does weather become climate free of weather variability for the purpose of establishing a trend?
Posted by: cohenite | May 16, 2008 9:14 AM
By looking at enough samples from each zone to attain statistical significance, perhaps?
Over a period of ≈18 years, but 30 years is better.
Posted by: luminous beauty | May 16, 2008 9:47 AM
Roger Pielke Jr. says: "If scientists started saying things like:
"In a change that does not disprove global warming computer models, the jet streams that govern weather patterns around the world are shifting their course . . ."
. . . instead of . . .
"In a change that is consistent with global warming computer models, the jet streams that govern weather patterns around the world are shifting their course"
then we'd have an improvement."
And we're supposed to take this seriously? Let's imagine why scientists use the wording they do in the first place:
"in a change that does not disprove the gradualistic or punctuated equilibrium hypotheses of rate of evolutionary change, use of molecular clocks demostrates that 8% of mDNA shifted in the development of the genus Colobarus over the last 4 million years..."
"In a change that does not disprove the Intermediate Disturbance Hypothesis, reestablishment of transient Orca reproductive rates took less time than reestablishment of resident Orca reproductive rates..."
"In a change that does not disprove the Steady State or Inflationary cosmological hypotheses, and also does not disprove the Big Bang Theory or any theory of Stellar Evolution, supernova G1.9+0.3 has clouds of gas expanding at a rate of..."
I thought Pielke was supposed to be a thoughtful skeptic, but this suggested method of writing is just an idiotic idea.
Posted by: Phoca | May 16, 2008 10:04 AM
But GCMs DO show variability similar to the ENSO effect, etc, so in the real world there's no reason to suspect them on these grounds, at least. That's why the IPCC projections are model ensembles - this has the effect of smoothing out the generated variability, resulting in a steadily increasing LONG-TERM trend.
Now are you beginning to see the silliness? Comparing the short-term temp record with its large weather variability component against a model ensemble which smooths out weather variability doesn't tell us much. You need enough years of data for the signal to swamp the noise.
Posted by: dhogaza | May 16, 2008 10:09 AM
Well, since he can't bring himself to out-and-out state that Lucia's "proof" that IPCC projections are "wrong", after having touted her work earlier, he's degenerated to trying to undermine the usefulness of showing that models are consistent with data altogether, as I mentioned above.
His change in wording is meant to highlight his belief. Remember, he's got this strange notion that if you play by the rules, models can't in practice be falsified.
Posted by: dhogaza | May 16, 2008 10:16 AM
This picture by John V tells quite a lot of falsifying longer time trends (what IPCC is talking about) on the basis of 7 year trends.
link
You could label the points where Lucia should claim global warming is inconsistent with the observations, ie every time the seven year trend goes to zero or below.
Posted by: mz | May 16, 2008 10:18 AM
Pielke:
Pielke again:
...except he missed the step of actually making the argument.
Truly a train wreck.
Posted by: bi -- Intl. J. Inact. | May 16, 2008 11:10 AM
I can't help but wonder if Ti Roger is doing some kind of performance art piece. What laypeople should learn from all this is that there is noise in the system, it's misleading when you look at it as though it has significance, and you need to use enough data to smooth out the noise to see the underlying trend. What's he doing? He's making noise, it's misleading if you look at it as though it has significance, and you need to read enough to be able to "smooth out" his noise to see the underlying trend.
Brilliant!
Posted by: pough | May 16, 2008 11:48 AM
I can't help but wonder if Ti Roger is doing some kind of performance art piece.
I floated this sort of idea a couple of years ago, but ran out of time to make it happen.
Certainly I would have had a tap-dancing character (like both Pielkes), and an obstinate engineer character (Harold Pierce or Willis E at CA [regarding his bash-like-an-engineer methods that completely overturn the science of agronomic yield forecasting]), the clueless shill character (Thapa at DotEarth), etc.
Anyway, truth is stranger than fiction and the real characters are much more entertaining than anything I could come up with or continue to develop.
Best,
D
Posted by: Dano | May 16, 2008 12:04 PM
I think Pielke's point is pretty simple. People like Annan and Schmitt issue these dire warnings about future warming, and then they back them up with huge error bands and wide probability percentages. The end result is that almost anything could happen and these idiot's will claim that their models have not been falsified.
Yesterday I ask Annan to put up or shut up. Annan claims that the temp will rise at .2C per decade. So in 30 years we should have .6C of temp rise. If that number is the correct number, then there should be a 50% chance that there will be more rise and a 50% chance that there will be less. It seems to me that this is an excellent basis for a bet.
I offered Annan a bet of 10,000 dollars that if the temp increases by .6C or more over the 30 years from 2000 to 2030 I would pay him. If the temp rise is less than .6C, then he has to pay me. We will use HadCrut3 data with a linear trend line. If the trend line shows .6C or more Annan wins, if not, I win. That is a simple, straight forward bet. No probabilities, no error bands, no bullshit. Does he believe in the AGW numbers he proposes or does he not. I am waiting to hear from Annan.
So, I am now going to make the exact same offer for the exact same bet to Tim Lambert. Shit or get off the pot Tim.
Posted by: Tilo Reber | May 16, 2008 1:24 PM
This is an outright lie.
And I suspect Tilo knows it.
Posted by: dhogaza | May 16, 2008 1:54 PM
Dano, these people devote their lives to writing rubbish. You can't beat them in their own game.
Tilo Reber:
"I honestly don't know what the proper test is." Yeah, it can't get any simpler than that.
Posted by: bi -- Intl. J. Inact. | May 16, 2008 1:56 PM
Tilo Reber,
You are offering Annan a 1:1 ratio on an event that according to his model is less than 50% likely. That is hardly appetizing.
Since according to you, the trend is zero, why don't you offer him a 1:1 ratio on that? Or at least split the difference and offer him a 1:1 on a .3C rise over three decades?
Posted by: Sortition | May 16, 2008 1:58 PM
Since according to you, the trend is zero, why don't you offer him a 1:1 ratio on that? Or at least split the difference and offer him a 1:1 on a .3C rise over three decades?
Because he doesn't even have the sense of fairness in betting of your average Vegas casino?
Posted by: QrazyQat | May 16, 2008 2:16 PM
"You are offering Annan a 1:1 ratio on an event that according to his model is less than 50% likely. That is hardly appetizing."
Why is it less than 50% likely?
What is wrong with a 1:1 bet. I'm on the other side, and my odds are no better than his.
I'm not the professional climate scientist - he is. If he is not bullshiting us, he should be willing to take the bet.
I also haven't said that the warming trend is zero. I have said that it is zero for the past 10 years. I am willing to make the bet with Lambert and Annan because I believe that their number is too high. Annan is always throwing out bets to people based upon their predictions. It's time that he take one based upon his own.
Annan says:
"over the 30 year time frame there will be strong warming"
And he defends the models when he says this:
"Also, they don't just predict "upward march" but they also quantify it (~0.2C/decade for the foreseeable future)."
Posted by: Tilo Reber | May 16, 2008 2:35 PM
Oh, the odds are 50-50 because Reber is "on the other side". Why care about this "doing the math" thing?
look ma wii dun n33d no edumacashun...
Posted by: bi -- Intl. J. Inact. | May 16, 2008 2:39 PM
Let me explain:
It would have been 50% if you were talking about 30 year into the future. You, however, have loaded the dice by including 8 years in the past - 8 years that have been (according to Annan's model) on the low side - not extremely low, but lower than average. This means that the remaining 22 years have to be higher than average, which means that this event now has less than 50% chance.
Anyway, even with 50%, Annan's expectation would have been zero, according to his model, so your offer would not have been much good even if you started counting at the present rather than 8 years into the past.
Now, you are saying that some warming may be occurring, although .2C per decade is too high. If you are willing to quantify that, you and Annan may find a bet that both of you could embrace. Do you believe in, say, .1C per decade?
Posted by: Sortition | May 16, 2008 2:51 PM
"Oh, the odds are 50-50 because Reber is "on the other side". Why care about this "doing the math" thing?"
Really a bunch of cretins on this site. I didn't say that the odds are 50-50 because I'm on the other side. I said on a 50-50 bet where I'm on the other side, the odds are no more appealing for me than for him.
The 50-50 statement goes like this - if you are not too dumb to follow:
Annan supports a model that claims a .2C per decade rise in temperature. That comes to .6C for 30 years. If the model that Annan supports is correct, then there should be a 50% chance that the temp rise is greater than .6C after 30 years and a 50% chance that it is less.
It's really a very simple bet. The amount of squirming and squealing that I'm hearing tell me that the AGW crowd is all mouth, all politics, and no actual believe in their alarmist crap.
Posted by: Tilo Reber | May 16, 2008 2:59 PM
Where is Annan squirming and squealing regarding your bet, again? I missed that reference. Surely you can link to it?
Posted by: dhogaza | May 16, 2008 3:03 PM
Tilo Reber, did you read my explanation (#20)?
Posted by: Sortition | May 16, 2008 3:05 PM
"It would have been 50% if you were talking about 30 year into the future. You, however, have loaded the dice by including 8 years in the past - 8 years that have been (according to Annan's model) on the low side - not extremely low, but lower than average. This means that the remaining 22 years have to be higher than average, which means that this event now has less than 50% chance."
I don't particularly like that argument. If the trend in temperature is due to the trend in CO2, then the CO2 has been rising for the past 8 years, and any temp rise that is less than that has been masked by "natural events". So if the theory is true, then we should see an upswing in temperature that exceeds .2C per decade when those "natural events" turn in the other direction. Since we are betting on the effects of CO2, any flat portion in the temp trend should be noise, not a permanent built in offset. So we cannot treat the first 8 years as an advantage that will bias the 30 year outcome.
"Anyway, even with 50%, Annan's expectation would have been zero, according to his model, "
Don't understand what you are talking about here. Can you clarify?
"If you are willing to quantify that, you and Annan may find a bet that both of you could embrace."
Nope. Like I said, Annan is constantly throwing out bets based upon other people's predictions. If he expects people to take those bets, he should be willing to take one based upon his own.
I'm a computer scientist, he is a climate scientist. That should be edge enough for him.
Posted by: Tilo Reber | May 16, 2008 3:14 PM
of course we can. It's no different than on betting on coin flips after the first 8 have been heads, for instance. The most probable outcome of the next 22 flips is 11 heads, 11 tails, or a total of 19 heads and 11 tails for 30 flips.
So a bet that the 30 flips end up 15-15 after the first 8 turn heads is not a 50-50 bet.
Would you offer Annan a 30 year bet using January 2008 as the starting point? Just curious ...
Posted by: dhogaza | May 16, 2008 3:27 PM
Annan's model doesn't work that way. According to the model, the temperature does not have an unobserved deterministic trend that the temperature hovers around. Rather, the temperature is performing a random walk, with an upward drift. So, on average the temperature becomes hotter by .2C every decade, but if by chance one decade was below average, this lowers the expected temperature from that point on. There is no rebounding expected.
Not being a climate scientist myself, I cannot tell you authoritatively why this is the way the model is set up. My understanding is that it is not a matter of accumulating CO2 over those 8 years, but of accumulating energy over those years. If due to some random effects not as much energy has been accumulated in the atmosphere during that time (say, it was absorbed in the ocean or reflected to space), then that energy is gone for ever and will not suddenly re-appear.
If you are unable to commit to less than, say, .15C/decade then there may not be any observable difference between your predictions and those of Annan, at least not over a period of 30 years. In that case there would be no bet where you both think you will win.
Posted by: Sortition | May 16, 2008 3:34 PM
If Tilo Reber insists on his bet, then I'd like to propose my own bet. Here it is:
This month is May 2008, right? So we can make a bet based on the 30-year trend of monthly global temperature anomalies (following HadCRUT3v) starting from June 1978 and ending in May 2008. If the trend is below .2ºC per decade, then Tilo wins and gets to post a victory message, and I'll admit to wearing baggy pants; otherwise, I get to post the victory message, and Tilo admits to wearing baggy pants.
Want to take the bet, Tilo?
Posted by: bi -- Intl. J. Inact. | May 16, 2008 3:41 PM
"Where is Annan squirming and squealing regarding your bet, again? I missed that reference. Surely you can link to it?"
I don't see that Tilo was refering to Annan with this comment, more likely to responses from "the AGW crowd" on this site....
Just a guess.
Posted by: Betula | May 16, 2008 3:53 PM
dhogaza "of course we can. It's no different than on betting on coin flips after the first 8 have been heads, for instance. The most probable outcome of the next 22 flips is 11 heads, 11 tails, or a total of 19 heads and 11 tails for 30 flips. "
Dumb comparison dhogaza. The coins have no memory. The temperature trend does. It's called the CO2 effect. The whole thing that we are betting on here is that effect. The rise in CO2 continued unabated for those 8 years. And the trend that we are betting on is supposedly caused by the CO2. This means that this year we have 8 years worth of CO2 rise that has not manifested itself in temperature. But if the theory is true, it must do so because it is there and because it's greenhouse effect is, in fact, the trend. The extra CO2 should be like a coiled spring, pushing temperature up that much faster when the natural variations that are holding it down are removed or reversed.
Posted by: Tilo Reber | May 16, 2008 3:53 PM
"Annan's model doesn't work that way. According to the model, the temperature does not have an unobserved deterministic trend that the temperature hovers around."
If the trend is due to CO2, then the temperature rise should be relative to the increase in CO2.
"There is no rebounding expected."
Why not? If the extra CO2 is being added, then why wouldn't you get it's effect?
"If due to some random effects not as much energy has been accumulated in the atmosphere during that time (say, it was absorbed in the ocean or reflected to space), then that energy is gone for ever and will not suddenly re-appear."
If it is reflected into space, then the CO2 isn't doing the job it is advertised to do. If it is absorbed by the oceans, then it will reappear as those warmer oceans release energy into the atmosphere.
In any case, as CO2 is added, it cuts the amount of heat that the earth radiates into space. The earth responds by getting hotter. As it gets hotter it radiates more heat so that more get's through the added CO2. Eventually a balance is reached according to the total greenhouse effect. The temperature rises to seek that balance point. So the CO2 that we add now, and whose effect that we do not see now, must eventually manifest itself.
Posted by: Tilo Reber | May 16, 2008 4:06 PM
"Want to take the bet, Tilo?"
June 1978 to May 2008. .2C per decade.
Don't know what HadCrut3v is.
Make it HadCrut3 and I'll take the bet.
Posted by: Tilo Reber | May 16, 2008 4:21 PM
Yes, it is not a perfect analogy, it wasn't meant to be, and it doesn't need to be.
Though asking you to bet 100% on past history makes the stupidity clearer.
Posted by: dhogaza | May 16, 2008 4:30 PM
Tilo, your bet offer to Annan would result in him losing 50% of the time even if he's completely right. IOW, you're an idiot and a troll.
Posted by: Steve Bloom | May 16, 2008 4:41 PM
Just to note that IMHO RP Jr.'s recent behavior can be explained as promoting his long-standing position of favoring adaptation over mitigation. He's constantly searching for new angles on this. I wouldn't be surprised to see him taking a role in the forthcoming second round of Lomborg's "Copenhagen Consensus."
Not inconsistent with the above, being seen to influence climate policy is what gets him funded.
Posted by: Steve Bloom | May 16, 2008 4:50 PM
"Tilo, your bet offer to Annan would result in him losing 50% of the time even if he's completely right."
It would also result in me loosing 50% of the time if he is completely right. The objective here is not to make money or to loose money, he should have an equal chance of doing both. The objective is for him to show that he believes the numbers that he is proposing without relying on error bands and probabilities to save his ass.
Posted by: Tilo Reber | May 16, 2008 4:53 PM
The problem is that you only lose 50% of the time if he is completely right. If it's supposed to be a good bet, you need to offer him more if he wins, to compensate for the fact that he could lose. It's basic economics - rational people require a risk premium.
Posted by: FhnuZoag | May 16, 2008 5:06 PM
"It's basic economics - rational people require a risk premium."
The risk premium is showing that he believes in his work instead of hiding behind error bands and probabilities in order to deny that the models have been falsified. I mean what is the point of such a claim if it has no real world significance. Yeah, the stinking models haven't been falsified. But they use up so much of their huge error band that they don't tell you anything either. If you think you know what the temperature trend is, then show that you are willing to put money on it. If you think that the trend can be anything from flat to .4C per decade, then shut up and don't ask the world to spend trillions on nothing.
Posted by: Tilo Reber | May 16, 2008 5:18 PM
The coin toss wasn't a perfect analogy, but it was pretty good. Analogies are difficult to get exactly right. "Greenhouse" is an example of that. Getting away from analogies, we can focus on better questions. For example, What about positive feedbacks? Water vapour responds to increased temperatures.... That's a fairly explicit reason to expect less warmth when including cold years for the first quarter of the test period. What is the precedent? It would be interesting to look at Hansen's 1988? model output with and without the prediction of a large volcano (keeping other factors equal).
Posted by: Steve L | May 16, 2008 5:22 PM
"For example, What about positive feedbacks?"
Natural variation swings to both sides of the trend line. When you swing to the plus side of the trend line you get more feedback than the trend itself. If you really have a trend and if you want it to show up as steeply as possible, then your ideal situation is to have the natural variability or noise be negative at the beginning and positive at the end. That will give you a steeper trend.
Posted by: Tilo Reber | May 16, 2008 5:32 PM
Tilo Reber
It is odd that you are now trying to dictate to Annan what is the model that he should be using. I am not sure whether my explanation for the modeling assumptions is correct, but, correct or not, this is how the model is set up. If you expect Annan to take your bet, you have to set it up so that according to his model, he stands to make a profit. Insisting that he take a bet because he breaks even on that bet according to some model of your own choosing is just silly.
Posted by: Sortition | May 16, 2008 5:39 PM
"The coin toss wasn't a perfect analogy, but it was pretty good."
No it is a terrible analogy. Any coin toss is completely independent of the previous one. Temperature increase, if the theory is correct, depends exactly on the amount of CO2 that was previously added. Let's try this. Let's say that I'm putting a teaspoon of gun powder into a can every second. And let's say that I ignite it every second. My explosion would always be of a given size - kind of like a constant trend. Now, let's say that I skip the ignition for one second. Then the next time that I ignite, I will have an explosion that is twice as large. Same goes for CO2. As long as you are putting it into the system you are going to get the effect (if the theory is correct). If you don't get the trend for a decade, then you should get twice the trend in a later decade (very roughtly) - otherwise it is not a trend.
Posted by: Tilo Reber | May 16, 2008 5:43 PM
"It is odd that you are now trying to dictate to Annan what is the model that he should be using. "
No, I'm just questioning your assertions about how his model works, since your explanation makes no sense with regard to CO2 forcing. Is there a layman's writeup somewhere that explains his model. I am suggesting the bet based simply on the model that he seemed to be defending when he made those two statements.
Annan: "over the 30 year time frame there will be strong warming"
and
"Also, they don't just predict "upward march" but they also quantify it (~0.2C/decade for the foreseeable future)."
Posted by: Tilo Reber | May 16, 2008 5:50 PM
Hm, no answer from bi about his suggested bet. Either he is thinking it over or he has something better to do for a while.
In any case, I have something better to do for a while, so I will drop of.
bi - here is the 30 year HadCrut3 minus May. I'll update the chart when May comes in.
http://bp1.blogger.com/__VkzVMn3cHA/SC32i9FfAPI/AAAAAAAAABU/gNN5A32MlqM/s1600-h/30+Year+HadCrut3.bmp
Posted by: Tilo Reber | May 16, 2008 5:57 PM
Tilo shows no more logic here than he ever has, anywhere.
May I suggest the troll be quieted?
It's hilarious, though, that he proves the point that basing a bet that includes a historical component is stupid.
Gun. Dick. Take aim. Fire.
Posted by: dhogaza | May 16, 2008 6:56 PM
Both of the statements you refer to are consistent with the random walk with drift model as I presented it. Your interpretation of those statements as indicating some kind of unobserved deterministic trend is your own. One simple way to clear these things up is to ask Annan.
I am not aware of any writeup, but if you do find one, I am interested.
Posted by: Sortition | May 16, 2008 6:59 PM
Tilo, you are so full of shit.
If you don't believe global warming is happening and the models are junk then make a bet that actually reflects your beliefs. Be a man or a close approximation. You will get takers and you will get taken.
That is, until you claim one of you conspiracies. Please.
Posted by: Boris | May 16, 2008 7:05 PM
No, actually the coin analogy was just fine. You seem to think that climate scientists believe C02 is the lone driver of climate change. You would be mistaken. There are any number of factors that drive global mean temperatures, to say nothing of the chaos of their confluence and feedbacks, so you are indeed introducing bias with your biased endpoint and no differently than in dhogaza's coin toss example. On that basis alone, the bet is a non-starter.
This gets to another point about your rant, namely error bounds. These exist because models do not account for everything. That there are no frictionless surfaces. There is this random idiosyncratic innovation component of real world outcomes that is not confined to the particulars of the climate system. Indeed natural phenomenon are almost exclusively defined in such ways. So your irritation with these bands only shows just how ignorant you are of modeling or forecasting.
Surely you should be satisfied that forecasters will be proved outrageously wrong if the temperature has not moved in 30 years, which will be way outside of the appropriate error bounds. Why shouldn't you be satisfied with that bet without trying to impose an economically idiotic version to prove something to a bunch of credulous ninnies by proving absolutely sweet FA?
Posted by: Majorajam | May 16, 2008 7:21 PM
ATTN: TIlo.
RE : Simple method for analysis of temperature records.
I'm just going to make your day! Big Time! Here are the results of a multi-decadal analysis for the Quatsino temperature record from 1895-2007 for Sept 21. I'm attempting to obtained an estimate of the effect of weather on Tmax and Tmin at this site. By choosing the equinox the amount of sunlight is constant and the photoperiod is 12L:12D. Also any effects of ENSO are minimal at this time of the year. The ENSO is nothing but a trouble maker and imparts great variations on temperature at other times of the year.
Interval Tmax Tmin.
1895-99 14.8 09.0.
1900-99 14.5 07.2.
1910-09 17.4 08.0.
1920-29 16.0 07.1.
1930-09 15.3 08.7.
1940-49 17.9 08.6.
1950-59 18.8 09.2.
1960-69 16.7 09.7.
1970-79 17.6 08.4.
1980-89 16.8 09.2.
1990-01 19.8 09.2.
2001-05 15.5 10.2.
2006-06 14.0 10.0.
2007-07 14.0 09.0.
Mean Tmax 17.0 +/- 3.0, Range 14-20.
Mean Tmin 08.5 +/- 1.5, Range 07-10.
Tmean 13.0 +/- 2.0.
Note: The error is the classical average deviation, AD.
It is just 100% pure luck that Tmax and its AD are exactly twice that of Tmin and its AD.
Presumably, the most important factors which cause variations of temperature at this site are (1) clouds with rain, (2) clouds with no rain, and (3) the wind.
Since this site is at the ocean-land interface, Tmax is a measure of the sea breeze temperature and Tmin is a measure of the land-breeze temperature. Thus it is not appropiate to use the Tmean metric since Tmax and Tmin are measures of two different types of air.
Note the decline in Tmax at 2001 and thereafter: the total drop is a huge 5.8 K as compared to the really hot 1990-99 interval. This is the more important metic since it related to the sea breeze coming out of the vast Pacific Ocean. I can't wait for Sept 2008. Actually the big drop in mean temperature occured in 2001.
Suppose this type of analysis was used at other sites using only the equinoxes and the soltices and the mid points between them or any minimal set of equally spaced intervals. We would probably get just as much useful info than trying to "extract a climate change signal from the background noise.", which is a phrase I read in one of the SPM reports.
All climate, like politics, is local. It make little sense to me to try to compute a useless metric like the global mean temperature.
I welcome hostile peer review!
Double the bet Tilo!
Posted by: Harold Pierce Jr | May 16, 2008 7:32 PM
Since this site is at the ocean-land interface, Tmax is a measure of the sea breeze temperature and Tmin is a measure of the land-breeze temperature.
?Que? I don't know about where you come from, but where I come from sea breezes tend to be cooler than land breezes.
If "All climate is local" why does ENSO appear (obviously falsely so, if the quoted phrase is true) to affect conditions across half the planet?
Posted by: Phoca | May 16, 2008 9:15 PM
"Double the bet Tilo!"
Thanks Harold, but the warmers don't have the courage to take the one already there.
Posted by: Tilo Reber | May 16, 2008 10:38 PM
"I am not aware of any writeup, but if you do find one, I am interested."
Oh! Then what do you base your assertions about how Annan's model works upon? Does he have a post on his blog? Anything?
Posted by: Tilo Reber | May 16, 2008 10:41 PM
"Tilo, you are so full of shit."
Ah, Boris, so nice to see that you are here contibuting to the planet's hot air - as usual. With Boris around, we will never have to fear for an ice age.
Posted by: Tilo Reber | May 16, 2008 10:45 PM
Yes - I am going on the explanation in Annan's posts linked above.
The only offer you make is a 1:1 ratio on a bet that according to Annan's model he has less than 50% of cashing on - it is small wonder you find no takers.
In the meantime, you have already accepted the possibility of a warming trend (which makes you one of the "warmers" yourself) and although you are vehement about it being lower than .2C/decade your not willing to say even that it is lower .15C/decade. It seems it is you who is lacking in courage.
Posted by: Sortition | May 16, 2008 10:59 PM
" if the temp increases by .6C or more over the 30 years from 2000 to 2030 "
you mean from "point" A in time to point B and all points between? and take a trend? not any kind of thirty year average centered on 2000 and 2030? (yes that requires waiting until 2045) do you not grasp the vast difference between the two paradigms?
in any event, a point in time is not defined by a year; you need to specify 12:00 midnight GMT July 1 2000 or something like that; then you will obviously need datapoints of comparable granularity between. going to be some big dataset. perhaps consideration of how that sounds will demonstrate the silliness of that idea.
or is what you are proposing taking the one year averages for 2000, 2001...2030? as i suspect you are thinking, at some implicit level? i believe it has been pointed out that a 30 year average is the norm, in order to provide a decent degree of damping of short term fluctuations. that doesn't work for you, eh?
Posted by: z | May 16, 2008 11:03 PM
"No, actually the coin analogy was just fine. You seem to think that climate scientists believe C02 is the lone driver of climate change. "
No, the coin toss isn't even remotely close to fine microbrain. I never said that CO2 is the lone driver of climate. I didn't say that scientists believed that. And I don't believe that. But the point that is being pushed by the warming cult is that CO2 is the reason for the upward trend in the climate. And the other point that is being pushed by the warming cult is that x amount of increase in CO2 causes x amount of increase in temperature. Depending on who you talk to, I believe that the number is about 3C per CO2 doubling among the warmers these days. This means that any climate change that is caused by natural effects will have that CO2 forcing effect overlaid upon it. Now, solar cycles are only about 10.7 years long. For PDOs the "D" stands for decade. The effects of volcanoes is even shorter. Meanwhile, CO2 keeps rising at a constant or slightly accelerating rate. And while the other natural effects may keep the built in CO2 forcing hidden for a time, they should not be able to keep it hidden for 30 years. So when the natural effects are removed, or reversed, then all of the effects of the built up CO2, for the entire time period, should become manifest. This is really a simple concept. It's beyond me how this forum could be so stupid as to not understand that. Do you understand the simple principle that a trend involves both a change in temperature and a change in time. If you don't catch up to the trend every time that you have a cooling period, then you have the wrong trend. If you say that the warming trend is .2C per decade, then that makes eveything outside that trend noise or natural variation. It means that if you are below the trend for a while, then you must go above the trend for a while. If you don't, then you have the wrong trend. Is there anyone on this forum with a brain?
Posted by: Tilo Reber | May 16, 2008 11:07 PM
I guess that according to Tilo, we're all cowards for not betting on what amounts to a coin toss.
I mean, if it's a 50% chance either way, even if the models are totally correct, then it's just pure chance and it doesn't prove anything more than a coin toss does.
So okay, Tilo, you want to bet $10,000 on a coin toss? The advantage is that it can be done now, nobody has to wait, and it's just as meaningful.
I call heads.
Posted by: anonymous coward | May 16, 2008 11:08 PM
Z "that doesn't work for you, eh?"
What in the hell are you babbling about? I'm saying take the monthly global temperature data from HadCrut3 beginning at January of 2000 and ending in December of 2029. Put the damn thing in an excel spread sheet and have excel run a linear regression trend line through it. Then look at the amount of temperature rise for that line over that time period.
Posted by: Tilo Reber | May 16, 2008 11:18 PM
"I call heads."
Right. When you have confidence that you have a 50% chance of winning you will make the bet. If Annan was confident in his .2C per decade, he would make the bet. My reason for making a 50% bet is that I don't believe that Annan will stand behind his .2C prediction. And I don't believe that any of the other climate science AGW luminaries would stand behind that figure either. They are ranting, pontificating cowards that are going to cost the world trillions in carbon taxes, trillions in blocked development, and a significant chunck of human freedom for a problem that isn't a problem. But while they dump their crap on humanity, they are unwilling to put any of their own money behind the predictions that they distress us with.
Posted by: Tilo Reber | May 16, 2008 11:29 PM
Why? I would never bet $10,000 on a single fair coin toss? Why would I? It's not penny change for me, 50% odds are far too low.
And of course you've rigged it so the odds are much worse than 50% by including historical data.
Please keep posting, Tilo. You are adding to the online literature showing that denialists can't tell their dick from their assholes (because, after all, you are a dick AND an asshole), and therefore do nothing but weaken the cause of more sophisticated denialists.
Posted by: dhogaza | May 16, 2008 11:43 PM
"Yes - I am going on the explanation in Annan's posts linked above."
Are you talking about his Pielke posts? I don't remember a description of a random walk model that doesn't catch up when it drops below the expected trend. Could you point a little more precisely.
"your not willing to say even that it is lower .15C/decade. It seems it is you who is lacking in courage."
If I were a climate scientist I would give you a number that I would be willing to stand behind and I would let you pick the side of that number that you wanted to take on the bet. I wouldn't hide behind huge error bands and huge uncertainties while at the same time expecting legislators to make policy on the basis of my guesses. I wouldn't, like that moron Hansen, run around testifing to any legislator who will listen about death trains and 25 meters of sea level rise.
What I will do is state that global warming will not cause mankind more harm by doing nothing about it than it would by paying to try to make it go away. Now if you can quantify that statement in a way that we can both agree to, then we can make a bet on it.
Posted by: Tilo Reber | May 16, 2008 11:50 PM
So if you expect to make a dollar at the expense of a dollar you will make the bet.
Reber must think Annan is as stupid as he is.
Posted by: Chris O'Neill | May 16, 2008 11:55 PM
"It's hilarious, though, that he proves the point that basing a bet that includes a historical component is stupid."
Ah, I knew that someone would jump to that conclusion. I should have known that it would be you dhogaza.
If you have a bet where the unmanifested portion of a temperature rise will have a chance to express itself, then having some historical component is not a problem. Now, most of the elements of natural variation will have time to reverse themselves in a thirty year period. So the CO2 that has piled up will have a chance to raise that temperature trend to where it should be. The only element that will not have time to go through a switch, that I can think of, is the Milankovitch cycle. But it's effect should be fairly small, and it should not change significantly over a 30 year period. In any case, I don't know what the direction of the effect of the Milankovitch cycle is at this point, so I cannot tell if it will help me or hurt me.
In any case, having a month for any unmanifested CO2 induced temperature increase to manifest itself is completely different from it having 22 years to express itself because of the length of the cycles involved.
Now, let's go on to another problem. If the past 30 years, beginning in 1978 have not given us .2C per decade of temp increase, an