The Blackboard

Per “denier’s hypothesis” of 0C/century.” Please avoid using “denier” in respect for those who suffered and died in the Holocaust. Recommend describing this as “agnostic” or “realistic” or “neutral”. (PS Recommend running a spelling/grammar check.)

Just found this. Might be of interest.
http://downloads.climatescienc.....al-all.pdf

edit: I tried posting a hyperlink but it’s reduced to . strange.

For some reason, the link didn’t show up. It reduced the to . Strange. Here’s the link:
http://downloads.climatescienc.....al-all.pdf

Thanks Chad. I’d read some papers by a number of the authors of that document.

Are these rejections gaining any credible attention whatsoever?

Or are they being ‘denied’ as well?

Sounds like a holy climate war to me.

;o)

Excellent work, Lucia. You have trudged through the muck that no one else seems to want to rake.

KW–
Like it or not, in the long run, the only thing that will matter is what happens in the next several years. I’m fine with that.

I am looking at model data to see if there is anything in the AR(4) models that makes me think the results I am posting are incorrect.

One way to ressurrect the 2C hypothesis is to accept a 1/f noise model. This would allow the 2001-2008 flatline to be written off as some unexplained “internal climate variability” process. The flipside is that it would do the same for the 1987-1997 uptick.

Hi Lucia.

Found your blog via a Rabbett link. Interesting.

How do you know that 7 years is enough data to actually do a meaningful statistical test of whether the expected 2C/century trend is contained within the global temperature data? I think the timescale for ocean-current circulation is on the order of a decade or more so 7 years seems as though it would be insufficient.

As an admittedly imperfect analogy, consider the following experiment: Water is flowing turbulently through a pipe. You want to determine whether the flow rate is increasing, remaining constant, or decreasing (within some uncertainty) and you have a very small hot-film anemometer positioned within the flow. The question is: How long must you collect data before you can apply a meaningful statistical test to your fluid velocity measurements and answer the question of whether the flow rate is changing? I think you’d agree that you’re wasting your time unless you use data spanning a period at least several times the characteristic timescale of the turbulent eddies, right?

So the relevant question here is whether 7 years is enough time to extract a meaningful signal from the global temperature data. I would argue that it isn’t based on the apparent timescale of the fluctuations in global temperature predicted in global climate model runs, which appears to my eye to be something around 5-10 years (see, for example, Figure 10.5 in the AR4 WG1 report).

Great work, Lucia. But do you think the Team can comprehend these difficult “standard” statistics?

BTW, I think you were right on the “rain shadow” phenomenon (hope you remember…). “Shoot and ask questions later” seems to be my MO, LOL.

Lucia, I’m lookuing forward to your response to Mike Powell. What is the length of the el Nino/la Nina oscillation? Shouldn’t a testing of the IPCC AR4’s projected central tendency of 2C/century extend at least over a few cycles? What happens if you start in 1950? 1960? 1970? 1980? It would be interesting to see a comparison of your shorter results against longer periods.

Mike Powell,
lucia already indicated that she thought the trend could easily change given additional data, so why ask if she thinks 7 years is “enough”? It’s obvious to me she doesn’t think it’s “enough” data to decide global policy. But it’s also obvious that it is “enough” data to say there is a mismatch between IPCC projections and real world observations. “Why” is the question. She has already indicated she is planning on working with more complex long-term-persistence error structures, so she is fully aware of your concerns. She has also said before that she doesn’t really like analogies. She’s pretty smart. She doesn’t need them.

As for the start date of 2001, TokyoTom, she has explained the logic of that choice already. She’s not trying to reinvent the wheel on trend analysis. She’s trying to get a truly independent out-of-sample test of the IPCC consensus hypothesis. Something that, for some reason, the alarmists aren’t interested in.

How about repeating the same analysis on an even longer and stronger down-trend in temps?

… Sept 1979 to May 1987 …

Then you can prove that so-called “Global Warming” stopped the year BEFORE Hansen’s infamous ego-trip to Congress!!!

How far back into the historic data do the CO and OLS cooling trends extend?

Might I suggest that newbies to this blog go back and read most if not all of the old posts? Many of these comments have been hashed over multiple times, and I think you would find them informative.

Mike Powell,
Good question. And I suspect you are the Mike I think you are. If you are, then I’ll use language more familiar to mechanical and chemical engineers when discussing the hot film aspect.

The issue you discuss requires a multifaceted answer. There are three things embedded in there:

1) The issue of the rate of change of the underlying trend. (This would be the what one might consider the “mean” or “ensemble average” behavior or “signal” — depending on terminology we read at blogs.)

2) The issue of the time constant associated with deviations from the mean. (This is related to the “turbulence”, or what is referred to as the “weather noise” at some blogs. It also relates to your hot film anemometer question.) and

3) The question of what the models say.

I’ll address each below:

1 It’s easy to address the question of the rate of change of the underlying trend. The central tendency of the IPCC projections show a linear underlying trend of 2 C/century during the first 30 years of this century. This is both shown on the various charts — as the dark line showing the “ensemble average” of model results, and can be reproduced by downloading the individual model runs and averaging.

So, the hypothesis test starts by assuming that is correct — for the mean trend.

Of course, no-one knows if the IPCC is correct. They could be incorrect in a number of ways — the real trend could be non-linear with time. The real trend could be other than 2C/century etc. However, for the purpose of testing, I make my null hypothesis be ” Assume the IPCC AR4 is correct and the current underlying trend is 2C/century, and linear.”

I take no other assumptions from the AR4. My hypothesis tests are based on features we see in the data. (The fact that I don’t will be important when we get to point # 3 with the models, as you are essentially suggesting I should use their weather noise too. )

Moving on the point (2): remember, I do my hypothesis tests assuming the “turbulence” or “weather noise” has features similar to that which appears in the data, not the models. (The discussion of the models will come later.)

2. In your hot film anemometer example, what one does to be certain that they can detected whether the signal is going down or up is to make sure we measured over a sufficient number of “integral time constants”, given the variability of the flow. (For those not familiar with the integral time constant, it is the integral of the area under the autocorrelation function for whatever random variable we are examining.)

Ordinarily, if we wish to detect a mean trend, then if at all possible, (and we are sane, had lab data, and enough flexibility to do the experiment in a way that avoids fancy statistics), we sample at intervals separated by more than three (or better yet 5) integral time constants. After doing so, we can assume the velocity variations due to the turbulence in adjacent measurements are independent. So, we could fit with Ordinary Least Squares and determine the trend using standard undergraduate methods taught in labs. (Of course, we would still check to see if things look ok — but the goal of estimating the time constant and sampling is to get nice clean data.)

Unfortunately, when using available meteorology data, it’s quite clear the monthly data are not separated by several integral time constants. If we check, we find the adjacent data points exhibit serial autocorrelation.

In particularly, after de-trending, the lag1 one autocorrelation for the data I’ve been looking at i themselves is about 0.4 to 0.6 (depending on which measurement set we are examining.) If the “noise” in the data AR(1), (i.e. “red”), this corresponds to an integral time constant that is less than 4 months — not years. However, we clearly would not have 12 independent samples in 12 months. We must account for the fact that the adjacent data are correlated.

The purpose of the “red noise” (or AR(1)) corrections is, in principle, to account for the serial autocorrelation feature in the data.

The approach to correcting for “red noise” is to adjust the number of degrees of freedom used in t-tests (or to determine confidence intervals) by substituting an effective number of degrees of freedom “Neff” for, N, the number you thought you had when doing Ordinarily Least Squares (OLS), by scaling with the lag 1 autocorrelation as follows: Neff =(1- rho1)/(1+rho1) N.

This scaling to reduce the number of degrees of freedom in the computation actually works — provided the autocorrelation for the data themselves is AR(1)– (i.e. red noise.) FWIW: In terms of continuous functions, the autocorrelation of an AR(1) process is an exponential decay. In terms of discretely measured data, it is rho(n) = rho1^-n. )

Anyway, I did this particular adjustment, and those are the results you see in table 1. In my tests, I estimate the lag 1 autocorrelation (and hence the integral time constant) based on the data. (It’s easy to calculate.)

If you examine the table, you’ll see I get rather large uncertainty intervals.

Although the best fit trend for this set of data is negative, that negative value is not statistically significant when compared to “0C/century”. This means that based on this test, we can’t rule out continued warming. The underlying trend could very well be positive. ( I would suggest that the underlying trend is positive. This is based both on the fact that the increase in GHG’s should result in a positive trend, the historic positive trend and the fact that “false negatives” are the norm with small data sets. )

However, using that test, I find the 2C/century for the IPCC data falls outside the uncertainty intervals using data from most of the reporting agencies. That is: the test says that, with p=95%, the underlying warming is less than 2C/century.

Is that enough though?
But there is a difficulty: If we examine the data, we notice it doesn’t look AR(1). So, the adjustment above might not be appropriate.

In fact, if we look at data from the previous period with little volcanic activity, it also doesn’t look AR(1). What it does resemble is data where the temperature is AR(1), but due to imprecise measurement we’ve overlaid white noise. (That’s what it looks like. It doesn’t mean that’s what it is…. )

Also, we know there is measurement imprecision because a) there is always measurement imprecision in all measured data and b) if we compare NOAA to Hadley to GISS etc. we see they don’t track each other perfectly.

However, it’s possible to come up with a correction for that form of “turbulence + measurement” noise in the data and correct.

I did that, and you’ll see the results in table 2.

Note: I still only get integral time constants on the order of several months.

Note that I’m still only using features of “turbulence” or “weather noise” based on the data. I have only a short span of data, so, this introduces the possibility of error. However, I’ve looked at data from a previous longer span of time when we had no stratospheric volcanic eruptions — which was roughly the 20s &30s. It happens that the “weather noise” for that period resembles the same “white noise” + “red weather noise” shape, with similar values of lag1 correlations etc.

So, it appears the spectral properties of the “weather noise” (or turbulence) we are currently seeing may be typical of periods with no volcanic eruptions.

This observation is purely empirical, and not based on the form of weather noise in models.

So, am I sure my estimate of the integral time constant is correct? Well, since I tend to prefer empirical data to model predictions, I think the time constant is close to correct.

However, there are those who suggest I should use the integral time constants from models.

Naturally, I’ve been looking at that. So, now onto issue (3), the models.

3) On the question of what the models say: I’m going to be uncharacteristically brief, mostly because I’m in the process of looking at this in great details, and I may revise things in a few weeks, after I double check my calculations, and compare to other empirical data sets etc. However, I can say a few things that aren’t going to change:

You point to figure 10.5 in the AR4. It’s a bit difficult to eyeball those to estimate integral time constants.

However, Gavin previously discussed the integral time constants and variability of 8 year trends based on that data. So, I’ll infer your thoughts fall along the lines of what he said. Gavin indicated that the standard deviation of 8 year trends in for the collection of models is σ=2.1C/century. (I get the same thing by downloading the model data)

In contrast, the red noise corrections estimate σ=1.1 C/century or less. (This depends on the data set — for a rough estimate, divide the 95% confidence intervals in the tables by 2. Notice that the surface based sets have lower uncertainty intervals than the satellite sets. The number also changes a bit each month — as expected for these sorts of things. ) If I adjust for measurement uncertainty, the analyses based on real earth data suggest σ=1.2 to 1.3 or so.)

So, the question is: Why are these so different? And which is wrong.

Well, if we turn to empiricism, I can point to a strong point suggesting the models are somehow wrong. It’s this:

If we compute the variability of 8 year trend for the full thermometer record it’s less than 2 C/century. Bear in mind the thermometer record includes a) ordinary measurement noise b) volcanic eruptions c) dramatic noise introduced by the jet-inlet to bucket transition and d) non-linear variations due to non-linear variations in GHG’s and anthropogenic aerosols. So, this strongly suggests the models variability in 8 year trends is too high.

If we compute the variability of 8 year trends to those during the period with no volcanos eruptions, the models over-predict the variability even more.

You can read more

That post includes this handy illustration:

So, the fact that the current “earth weather noise” seems to match the previous one in the data record, and the model “weather noise” doesn’t match either case for the earth weather noise, suggests (to me at least) that it is better to use the characteristics of earth weather when doing a hypothesis tests.

I think that comparison itself is fairly strong evidence the models are over-estimating the 8 year variability. However, it hasn’t convinced everyone.

Because it hasn’t convinced everyone, I’m currently looking at the “model weather” in more details right now. Anyway, maybe I’ll change my mind after I look at the “model weather”.

My first step was just to eyeball the “weather noise” in the models. Some models in the AR1 batch have very odd and unearthlike “weather noise”, (see FGOALs, EchoG.

I’m going to avoid saying too much about that now, but I will say that, based on what I’ve looked at so far, the model weather does not look like earth weather. Some in obvious ways. Others in more subtle ways. I’m focusing on periods with no-major volcanic eruptions, and looking at metrics related to performing a hypothesis testing with 90 month batches of data. (I will aslo admit that since I haven’t blogged or shown the results of these examinations of the model data, that people are allowed to point out that, as they haven’t seen these comparisons, they don’t have to believe mere hints.)

Still, for now, it appears to me the model over estimate the variability in 8 year trends.

Brad–
1) I have never, ever suggested global warming stopped. I don’t believe it has. These analyses not only don’t show that, but specifically state the opposite. The best fit trends from 2001 are negative, but tests for statistical significance indicate we can’t rule out positive “mean” trends.

It’s 2 C/century that’s ruled out.

2) Why would showing that the temperature decrease after El Chicon erupted in Chiapas prove global warming ended? You can see the down trend in a box below:

The current down turn is remarkable precisely because its not associated with a stratospheric volcanic eruption.

We understand the downturn associated with the the eruption of El Chicon. Similar downturns abound in the empirical record. We also have phenomenological explanations.

Reference. I don’t know precisely how far back the negative trends go. I picked 2001 based on publication dates of documents. Hunting for the longest possible period of negative is a) time consuming and b) cherry picking. So, I’m not going to do it. However, my guess is that the trend since 2000 is positive.

BarryW & Bender–
No need to lecture Mike.

With regard to hunting for past posts, the fact is, it’s difficult for newbies to do that. I haven’t expected my traffic to increase so quickly, or to blog so much. So,I didn’t organize the categories to make it easy.

Also…. I’m pretty sure I know who Mike is, and bender… he’s pretty smart too.

bender, thanks for your explanation of what Lucia`s about.

FWIW, count me as an “alarmist” who IS interested in testing how well the models perform. It looks like they`ve underestimated change in the Arctic:

http://climaticidechronicles.o.....after-all/
http://nsidc.org/news/press/20080610_Slater.html

Tom–
It is fairly widely admitted the models are unable to predict climate change on local or regional scales. The over predict in some places and under predict in some places.

So, the behavior you link to is just one more scrap in the mountain of evidence the models can’t predict local climate, possibly even on continental scales.

However, evidently, for some reason, one is supposed to believe the models are can accurately predict global climate change– which is the spatial integral of local climate change.

I focus on testing the claim about the global change. After all, if modelers do indeed admit that models can’t predict regional climate change why should I test that?

Lucia,

Thanks for the comprehensive answer about the volcanoes. That actually made good sense to me.

Can I ask two more questions about your trends?

Question 1.

You are compensating for Auto-Correlation – I gather that’s the tendency for a warm year to be followed by another warm year. (Temp at t correlates to temp at t+1)

If so, did you test for NEGATIVE Auto-Correlation? ie, the tendency for a warm year, or a warm series of years, to be followed by a COLD stretch of years at some later time? (Temp at t inversely correlated to temp at t+2, or t+3, etc.)

In laymans terms (that’s all I can undertand anyway), if the century long trend really is +2deg/century, then a warm stretch like 1993 to 2002 (with an above average trend of +3.3 deg/century) will tend to be followed by a below average stretch. Maybe that’s called “mean reversion”? I’m not sure if that’s the correct technical term.

At shorter times scales, this effect certainly stands out. Warm year 1998 was followed by cold year 1999. Warming trend 1986-1987 was followed by cooling trend of 1988-1989. The mechanism of El Nino/La Nina seems to be the global climate cop that keeps the trend in line.

Looking at the 30+ year trend, we seem to have just finished 7 year above the 2 deg/century trend line (2001 to 2007). Maybe its time for a few years below the trend line. A cooling-off period because the temps got ahead of themselves.

Question B

You are analysing surface temps, which are mostly influenced by lower troposphere, soil, and upper ocean temps. But that is ony a tiny fraction of the climate system that is capable of absorbing the hypothesized heat budget imbalance due to GHGs. Would your analysis look different if you used joules-into-the-climate-system instead?

The best proxy for this I can think of is global sea-level change. If there is a good estimate of ice melt we could subtract out, most of the remainder would be thermal expansion due to absorbing heat from the climate system. Since the oceans contain ~95% of the thermal capacity of the climate system, this should be pretty accurate, accounting for most of the supposed thermal budget imbalance. It also has the advantage that it is a much less noisy data set.

Another way to look at it – if you are analyzing a system that has significant unaccounted-for-fluxes into or out of it, then you might get confused by a trend that is explained by that missing flux. The atmosphere could be such a system. You are measuring the temperature change in the atmosphere, but it could be gaining or losing heat to the oceans, which you aren’t directly measuring (beyond the upper few meters).

Thanks for your good work,

Brad

Brad:
With regard to question 1: When I compute the autocorrelation, I get the number I get. If the lag 1 autocorrelation were negative, that would pop out. Neither the models nor the data show this.

However, there may be negative autocorrelation at large lags; the data suggests some. I’m looking into that further. With the short time series since 2001, it’s difficult to learn anything statistically signifcant at longer time lags. But, I’ll be looking at the longer string of data from the 20s and 30s.

So models show some very very very strong negative autocorrelations associated with massive El Ninos, but the model groups themselves indicate they think those indicate a problem with their models. (El Nino exists– but it’s not as strong as, say, the FGOALS model data indicate.)

Very large time period oscilations may exist. (For example, the PDO is thought to exist.) Pulling the features out of the historical data would be difficult partly due to the darn volcanos which have a very heavy signature.

Identifying the spectral properties at long time scales from “model data” would require a) looking at the control runs and b) assuming they are right.
I haven’t done the first.

On question 2: Yes. Things would look different if I examined a different metric. However, figuring out the ocean heat content in Joules is a project in and of itself. One can test the trend only after figuring out the Joules etc. So, I’m not doing that for now. For now, I’m looking at GMST.

Lucia,

You talk as if there is no difference between GCM and RCM’s and you confuse them as well. The question of 2C/century is for world average surface temperature and has nothing to do with RCM’s. Are you saying that both kinds of models are worthless? That 2C/century is wrong –see:

http://climateprogress.org/200.....d-in-1998/

–and that RCM’s tell us nothing useful about future regional climate?

One of the ironies of model critics is that they often have more unrealistic expectations of what the models can do than do the modelers themselves.

Blogging for the future at Climaticide Chronicles

GISS, HadCrut and NOAA/NCDC data may or may not really represent any physical reality. How well do near Earth air samples and top of water samples reflect rises in energy levels in the entire system? Got me. UAH/MSU and RSS on the other hand probably pretty well show the average of the atmosphere they look at, but that doesn’t tell us much about the ocean.

Brad:

While I agree that energy levels would be a better way, what’s being tested is GISS, HadCrut, NOAA/NCDC, UAH/MSU and RSS and if they show with a certain level of confidence a certain trend. So looking at some other measure isn’t really applicable.

Johnny Rook–
I don’t know why you think I’m confusing Regional climate models with GCM’s. I was responding to TokyoTom who I understood to be saying GCM’s incorrectly predicted warming at the poles, and I was saying it’s well known GCM’s regional predictions are poor.

No one has been discussing regional models, so I think most would infer the word “model” used in both Tom and my comments meant GCM’s not RCM’s.

With regard to RCMs: I’m not testing those nor am I delving into the literature about their individual projections.

Phil,

I’m trying to suggest that surface temps may not capture the useful information about the climate system that we are interested in. The original question was whether the long-term trend of +2deg/century is falsified. At long time-scales (a century, for example), there is no danger that an imbalance in the global heat budget will be mostly “hidden” in the vast storage capacity of the oceans, but ignoring this possibility for short or medium time-scales seems rash.

If the atmosphere is loosing heat to the ocean, and the atmospheric temperature is going down, can we really say that global climate is cooling?

If it’s global [i]climate[/i] we are interested – not heat sloshing back and forth between reservoirs, measuring only one of the reservoirs (and the especially the smallest one) won’t give a robust answer with any predicitive power.

(sorry about the tags, is there a posting guide for local html rules?)

Lucia,

Thanks for looking into the negative autocorrelation at larger lags. That is, indeed, exactly what I was driving at. I understand you have found the t+1 lag to be positive, and now you have found some of the t+2,3,etc lags to be negative? How will you incorporate that into your falsification model?

It almost seems as if my two questions have converged into a single point. If heat sloshes between reservoirs, but the total is not changing much, then we should see oscillations when we examine only one of the reservoirs. But the oscillations could be negative correlated over the medium term as the sloshing reverses direction.

Thanks,

Brad

Brad–
The IPCC models are AOGCM’s. So, in principle, heat transfer from the atmosphere to the oceans is considered in the models. The AR4 projections describe surface temperatures, so I compare them to surface temperatures.

If the prediction of surface temperatures are inaccurate due to flaws in the heat transfer rates to the ocean, then that would be an explanations for the falsificaiton, but it would not overturn it. It would be one has identified precisely what went wrong when making predictions.

On the negative autocorrelations’ effects on models: The exact way to incorporate negative autocorrelations varies depedning on the precise shape of the autocorrelation function. However, generally speaking negative loops in the autocorrelatin function reduces the integral time scale. If I neglect these, the falsification appears stronger.

There is a difficulty if we know precisely where we are starting in the “loop” though. That’s why several early posts discuss the ENSO cycle. If we ‘correct’ for ENSO, the falsificaiton becomes stronger not weaker. This happens because while correcting for ENSO increases the trend in temperature, it also reduces the uncertainty intervals. We end up concluding 2C/century is further out.

The big difficulty comes with the PDO. But I haven’t been able to get anyone to suggest how much the PDO could affect GMST. So… on that one, I don’t know.

Lucia
Insightful reading your discussions.
Not being a statistician, here is some brainstorming to explore/better understand what you are doing.

1) Question on the correlations between climate and solar variability described in:
Is climate sensitive to solar variability? by Scafetta and West Physics Today March 2008

Scafetta & West show the 11 year solar cycle influencing the temperature. Is that significant relative to your “time constants” and the 7 year data?

2) Do their statistical methods require more elaborate statistical analysis of the data to evaluate 95% limits on underlying trends?
3) Or can Scafetta and West’s models be distinguished from the IPCC’s 2C/century from anthropogenic CO2?

4) Similarly, Roy Spencer showed combinations of weather oscillations could account for much of the global temperature trends. e.g. See:
Testimony of Roy W. Spencer before the Senate Environment and Public Works Committee on 22 July 2008
Can the fit for Spencer’s models be compared to IPCC’s?

5) So a long term question:
Can combinations of methods of Scafetta & West with Spencer provide higher correlation with experimental data than IPCC’s 2C/century mean of GCW models?
(Possibly adjusted for stochastic volcanic eruptions.)

6) How do such models influence your present quest to test the IPCC’s 2C/century?

Feel free to mull over/answer at your convenience and time or not.

Lucia,

You said:
“The IPCC models are AOGCM’s. So, in principle, heat transfer from the atmosphere to the oceans is considered in the models. The AR4 projections describe surface temperatures, so I compare them to surface temperatures.”

So please help me out here, one step at a time:

Step 1. The AOGCMs incorporate heat transfer from atmosphere to ocean (and vice versa), just like the real world.

Step 2. Therefore in the long run, none of the reservoirs can be out of balance with the others, in the model. But do the models have short-term imbalances between reservoirs?

Step 3. The AR4 projections describe surface temp, which we already know must be mostly in balance with the ocean in the long run (step1 and 2).

Step 4. You have analyzed a medium-to-short term temperature series, but it only measures the heat content of one of the reservoirs. If you time scale were long enough to be sure that there were no transient fluxes from that reservoir to any other (say, the ocean),then that would seem to match the IPCC methodology, and could be compared to the AR4 surface temp projections. But how do we know there were no such short to medium term fluxes?

If the atmosphere actually did lose heat to the ocean in the 7 year time span you analyzed, is your analysis valid?

Thanks for your patient dialogue,

Brad

So please help me out here, one step at a time:

Step 1. The AOGCMs incorporate heat transfer from atmosphere to ocean (and vice versa), just like the real world.

The models include parameterizations that are supposed to mimic the real world process.

Step 2. Therefore in the long run, none of the reservoirs can be out of balance with the others, in the model. But do the models have short-term imbalances between reservoirs?

At any particular time, one reservoir can be out of balance with another. This is true for the real world also. For example, if the ocean and atmosphere managed to be perfectly in balance, and then a volcano erupted, both would cool. Then, after the dust veil cleared, they would start to heat.

But, because of the way things worked, the air might cool more initially, then “cooth” from the air would travel into the warmer ocean. Then, when the dust veil cleared, the opposite could happen. This happens all the time in all sorts of systems.

The models are supposed to be able to describe these short term transfer. If they are realistic, they will do it in a realistic way.

Step 3. The AR4 projections describe surface temp, which we already know must be mostly in balance with the ocean in the long run (step1 and 2).

Things only get in balance “in the long run” if the forcing becomes steady “in the long run”. But, assuming forcing becomes steady, then what you say is supposed to be more or less true.

Step 4. You have analyzed a medium-to-short term temperature series, but it only measures the heat content of one of the reservoirs. If you time scale were long enough to be sure that there were no transient fluxes from that reservoir to any other (say, the ocean),then that would seem to match the IPCC methodology, and could be compared to the AR4 surface temp projections. But how do we know there were no such short to medium term fluxes?

Short or medium term fluxes in what? Between the ocean and the atmosphere? The IPCC methodology for predicting the temperature rise in the next 30 years does not assume the ocean and atmosphere are in balance.

In fact, the IPCC graphs suggest the contrary. According to the IPCC, there is heat “in the pipeline” after 30 years. That is to say– the ocean supposedly lags the atmosphere, and some heat would transfer from the atmosphere to the ocean even if we stabilized. In other word: The IPCC projections models themselves assume that, on average, heat will be transferred from the atmosphere to the ocean during the first 30 years of this century.

The IPCC methodology claims to have modeled the history from at least 1900- now, and to apply all the correct fluxes , including those between the different reservoirs — and to model the full climate system. They make a projection about GMST.

The question is then: Having done all that, is the projection for GMST correct?

If the atmosphere actually did lose heat to the ocean in the 7 year time span you analyzed, is your analysis valid?

Yes, my method of testing the models would be valid.

Moreover, according to the IPCC, there is “heat in the pipeline”, which is a short hand term to express the idea that the atmosphere and ocean are out of balance. They are supposed to be out of balance in a way where heat transfers from the atmosphere to the ocean. At the same time, the atmosphere continues to gain energy from the sun.

The models say the net effect is for the atmosphere to heat at a rate of 2C/century (on average.)

The AOGCM’s are supposed to correctly predict the rate of heat exchange between the ocean and the atmosphere. This isn’t an “external factor”; it’s part of the physics supposedly contained in the AOGCM’s.

My intent was not to lecture, only to inform others that you’ve spent a lot of time building up to this point so they would be able to get up to speed, although your last posts should have done a good job of that. I apologize if you or anyone else took offense.

The way I see it, the models nominally claim to account for every important factor. If they do, then heat exchange to/from the oceans would be factored into their surface temperature predictions. If they do, then variations on a decade scale should show up in their predictions. Etcetera.

When I look at published graphs of model predicted temperatures they do show decade scale oscillations. So, they are trying in that sense. Thus, it is not unreasonable to investigate their ability to do so accurately.

The thing is, from what I have seen, the actual global temperature change usually seems to be significantly less than the predictions — once one gets about a decade out from a starting point (where temperature data fitting ends and temperature prediction begins). Some models are better than others.

Anyhow, I think what Lucia is doing is valid. The onus is on the models (and modelers) to fit with reality during their predictive period. She is showing that they likely do not.

Beyond that, explaining why they don’t isn’t her job — rather, its the modelers’ jobs. However, I want some “proven” scientific facts (not “learned opinions”) to precisely explain why the models are too high and how future models will be corrected for the relevant phenomena. I do not accept the “hand waving” opinions (in lieu of fact based explanation) that getting it wrong on the decade scale during the predictive period is irrelevant.

Just an opinion.

The trend per decade/century calculation depends on the time that one starts measuring it from.

Awhile ago, I calculated the trend per decade using the monthly Hadley dataset starting in Jan. 1850, then Feb. 1850 … all the way up to the current month.

There is definitive trend in that trend.

Starting in 1850, the trend is 0.04C per decade (one-fifth of the IPCC predictions.)

The trend grows very slowly so that it reaches 0.08C per decade by about 1940.

The trend continues growing slowly (almost linearly) so that it reaches 0.15C per decade by 1962 …

and the magic 0.2C per decade is reached if you start measuring from 1991.

The trend per decade then starts declining again so that if you start measuring in March 1996, it is down to 0.1C per decade again.

The trend from May 1997 is virtually 0.0C. A few small ups and downs …

And then the trend per decade starts declining precipituously.

Five years ago today, the trend is a scary -0.35C per decade.

The average average over the entire period is falsifiable 0.07C per decade.

Lucia, I think there is a lesson in this if you replicate this work.

Very interesting post and conversation. Lucia, selecting periods of no or minimal volcanic activity is a good start. Detrending ENSO variation in addition I think would be an interesting thought. There is still more known and probably some unknown natural oscillations/variations that would muddy the waters but every step in the right direction is useful.

Lucia, my simple point is that while the GCMs may have the “climate sensitivity” (average global warming per CO2e doubling) wrong, people shouldn’t ignore the obvious signs of REAL sensitivity in the climate and of rapid ongoing climate and ecosystem change.

The rapidity of these changes (not confimed solely to the Arctic), even while it is difficult to attribute them to GHGs, should make us more and not less concerned about our ongoing tinkering with the planet.

Lucia,

Thanks for that very thorough explanation — I understand much better now what you’re doing and why. I’ll give it a bit of thought and read through some of your archives before commenting much further.

I should’ve read further through your previous posts before adding my previous comment — I apologize to you and your readers for that. Mostly I just wanted to say “hi” but after reading through your post I was struck by the chaotic-noise timescale problem and wondered how you addressed it. I *knew* this was something you would’ve considered — sorry if anyone intrepreted my comment otherwise.

I stumbled across your CV on the Internet about a year ago and was glad to see you’re working on the climate change issue. Now your talents can be put to much more valuable use than when they were applied to the “stirring up stuff in tanks” problem. (Yes, I *am* the Mike Powell you think I am…) I’m also glad you’re blogging about your work — I’ve added your site to my weekly reading list.

Back to the topic at hand, though, I’m surprised the data show an integral time constant of around 4 months. That just *seems* too short based on what little I know about ENSO and so forth. I’m *not* saying your analysis is wrong — just that it’s surprising to me. Something more for me to think about…

I am going to start calling you st. lucia. You have too much patience!

Lucia, thanks yet again for this good natured, illuminating and clear discussion. Mind like a steel trap! Keep it up.

Lucia
The results you describe above, rejecting the model trends over a short time period raise 2 questions:
1) [Further to Bill Illis's comment above] – If progressively earlier start dates are selected, are the model trend outputs still rejected? My suspicion is that this will be the case back to at least back a few years beyond 1998, but that if you were to take the start dates back into the late 80s or thereabouts, the trend may start to ‘fail to reject’ because of increased uncertainties and because of pulling down the left side of the graph. Presumably this exercise could reasonably be continued back to 1979 and the start of the satellite records.

2) Alternatively, keeping the same start date as used in this analysis, how much temperature increase would be necessary in the next X years for the 2C/century trend to not be rejected? I.e. if you were to project the temperatures over say 5 or 10 years (some length of time that GISS would consider long enough to be representative of climate rather than weather), how much above the 2001-2008 average or trend would they need to be to allow the 2C/century trend to be statistically in agreement with the temperature?

Mike-
I figured it had to be you. Mike is a common name. Powell is a common name. But how many Mike Powells would automatically know I would know what a hot film anemometer was?

The blog is a hobby. I work at Argonne part time. Today, I’ve got to fiddle around seeing how changing some thresholds affects false positives on detectors for sarin gas. . . (Homeland Security. )

I work very part time, which is what I want. That leaves time for downloading climate data and blogging. If this were actual ANL work, I’m not sure I could blog about it! (Their publication guidelines for work related stuff are similar to PNNLs. So… I probably would have difficulties.)

Yes. It is surprising the data show such a short intergral time scale. However, it turns out that when you look at “ENSO” corrections, the total amount of energy in ENSO isn’t all that large compared to the amount at other scales. It’s large enough to be noticable, but weather noise contains a lot of variabiilty that isn’t ENSO. (Think about it– if it were huge it would be easy to predict next month’s weather. We just check ENSO! But, even when we know if it’s a La Nina or El Nino, it’s still difficult to guess.)

Also, ENSO itself isn’t that predictable. It’s an “oscillation” not a cycle. The oscillation has time scale between 2 years and… erhmm.. 4? It’s not sinusoidal. So, even that energy is scattered across a fairly broad range.

The other issue with ENSO is that I’ve looked at it several times.

Here’s long ago:
http://rankexploits.com/musing.....ne-orcutt/

Here’s more recent:
http://rankexploits.com/musing.....l-falsify/

There are statistical “issues” when correcting for ENSO, and they aren’t resolved in the analyses I linked. But as you can see, ENSO doesn’t explain the problem. (And based on looking at the correlograms for models, ENSO isn’t the reason for the large 8 years scatter in the data. It looks like a few models have very, very long integral time scales, that really looks like “AR(1)+Red”. The measurement noise issue was not the only reason I’m looking at that type of shape of correlogram! But, I’ll be blogging that more after I’ve looked more and made sure I don’t have “boo-boos” in the calculations. )

Bender&Fred–
I’m not that patient. But I happen to know Mike, and he knows me. We worked on lots of projects together in the past, so we are both used to asking each other questions like that. So, that puts his question and my apparent “patience” in context.

(Example of things we worked on: http://www.google.com/search?h.....tnG=Search )

Ian: We’ll get “fail to reject” by only pushing the start date back a little. We get alternating “reject/fail to rejects” if we push it back a bit farther. One of the reasons we start to “fail to reject” is that once you include a Pinatubo in the time span, the error bars explode. There is lots of variability associated with volcanid eruptions.

If we push it far enough back, we will reject 2C/century for certain because it didn’t warm 2C last century.

But rejecting 2C/century since 1800 is pointless because no one “predicted” warming at a rate of 2C/century during that period of time. The 2C/century is predicted for the first few decades of this century.

When testing a hypothesis, we need to limit the hypothesis to one that was actually proposed.

TokyoTom is making an interesting argument: We should overlook the fact that global average temperatures are significantly cooler than expected for a growing period of time and should instead focus on other short term regional outcomes instead, such as the Arctic. Usually alarmists make this argument in reverse–for example, if one can’t eliminate evidence of a Medieval Warming Period somewhere, one declares it to be a mere regional phenomenon that has no global significance.

It is my understanding that there is such a thing as an Arctic amplification phenomenon such that when the northern hemisphere warms (for any reason) the Arctic is supposed to warm faster. During a multi-decadal period when the global temps (more in the north than the south) went up 0.6 of a degree, the Arctic went up 1 degree. That would be largely unremarkable unless AGW modelling included amplification and predicted more warming than that.

Global temps have not kept up with IPCC predictions (however vaguely and untestably those predication are defined) and I suspect that somewhere in the model collections, an amplified Arctic warming has not kept pace with those predictions either.

It seems a bit desperate to trot out breathless anecdotal accounts from folks presumably not old enough to remember the big melt of 1922 in order to downplay the heretical lukewarmist implications in lucia’s remarkable work.

Cheers.

“TokyoTom is making an interesting argument: We should overlook the fact that global average temperatures are significantly cooler than expected for a growing period of time and should instead focus on other short term regional outcomes instead”

George. I`ll thank you not to put words in my mouth; one might say “it seems a bit desperate”, but those are your words, not mine. My point was clear; just as it`s fair to compare the models with reality, one should not forget that the climate is still changing rather remarkably.

Yes, Tom, it’s changing; but not from CO2 forcing.
==============================================

Climate changes, so what else is new? The remarkable thing is that it is not remarkable. Sat photos have shown that there were rivers where the Sahara is now. Earth has been warming since the last ice age, and sea levels have been rising since then. CO2 has been up and down over millennia.

A hypothesis has been made that the models represent reality at a level that can estimate the future temperature. The predictions have not panned out so far, but they are being used to support apocalyptic scenarios. That’s important because governmental policy is being set based not on the science but on the extreme projections. Bad science and bad policy.

Tom,
Just as the climate changed rather remarkably at the end of the little ice age a century and a half ago. The fact that Galveston Bay froze over in 1822, 1886, and 1899, and has not frozen over since is evidence of quite a remarkable change in climate, but it is unrelated to CO2.

Kim: “it’s changing; but not from CO2 forcing.” I challenge you to support that conclusion. Certainly it`s not in evidence in any of Lucia`s work. Pielke Sr has said he thinks the CO2 contribution to temp rise has been about 25%.

Barry: “The remarkable thing is that it is not remarkable. Sat photos have shown that there were rivers where the Sahara is now. Earth has been warming since the last ice age, and sea levels have been rising since then. CO2 has been up and down over millennia.”

Just to clarify – are you denying that man`s activities – agriculture, carbon black, aerosols, GHG releases – have or can ANY effect on climate? That there is simply no cause for concern, and we should abandon any fantasy of terraforming other planets (and any worry that we may be terraforming Earth for dinosaurs) and that “geoengineering” can`t work?

Sea level rise post-ice age largely ended centuries ago, but have been accelerating over the past century.

Our limited ability to affect change in the past is not evidence that we have no effect on climate now. (Further, there are some arguments that carbon and methane releases resulting from agricultural and deforestation practices helped contribute to warming over the past several millenia.)

Kazinski: “The fact that Galveston Bay froze over in 1822, 1886, and 1899, and has not frozen over since is evidence of quite a remarkable change in climate, but it is unrelated to CO2.”

It may be hard to spot any signal of human influence over natural noise, but again, our limited ability to affect change in the past is not evidence that we have no effect on climate now.

ToykoTom

Sea level rise did not end centuries ago and was occurring before the significant rise in CO2 or any other possible human influence. Plus it’s gone down over the last few years along with the temps. How does that work with CO2 going up?

No I don’t deny that man has an effect on climate, so does everything else in the biosphere. Focusing on one GHG, that may raise the temperature a degree or so, to a level of hysteria is what I object to. Denial of the MWP and LIA, because it doesn’t fit dogma I object to. Man, at this stage of the game, is going to have regional effects that are much more of a concern than that. For example, the damage to the fisheries from overfishing, and runoff should be where we are focusing our efforts.

And if the human effect is less than natural noise assuming there is a signal when you can’t extract it from the noise just because you think there must be isn’t science.

As for teraforming planets, you’ve been reading too much sci-fi. We’re not even in the bush leagues when it comes to that subject.

Tom, I’ve oversimplified. The climate is cooling, so any warming effect of CO2 can be tolerated until we understand it well enough to do something about it, if necessary. But why did you bring up the strawman of suggesting that lucia’s work here led to my assertion?
===============================================

Tom–
I think Kim and I both agree that many of his beliefs spring from sources outside what I do. As far as I can tell, the results of any analyses I have done neither prove nor disprove his assertions.

I think the balance of the evidence goes contrary to kim’s position. So, for example, I think CO2 is contributing to the warming and melting. But that’s also not a direct result of my tests of AR4 GCM’s ability to forecast.

What I am doing is rather narrow in scope.

I think it worth remembering the essential message of the IPCC which is:

1) The warming since 1960 can only be explained by CO2
2) The planet will continue to warming at an accelarating rate as long as we emit CO2
3) The consequences of the warming are very bad
4) Rapid reductions in CO2 emissions are the only way to deal with these negative consequences.

One can completely reject some or all of those claims without rejecting the basic premise that human emitted CO2 is causing the planet to warm. It is also important remember that the IPCC claims are based primarily on an anlysis of GCMs outputs which may or may not have a connection with reality. If these GCMs are not reliable then the IPCC claims must be dismissed as ‘not supported by the evidence’ (note: ‘not supported by the evidence’ does not necessarily mean ‘wrong’)

That is why we really need to validate the GCMs against the real data using robust techniques that are accepted in other fields that use computer models to make economic/safety decisions. Lucia’s analysis is simply one many that should be conducted by people with no vested interest in the outcome of the analysis.

More importantly, model validation is a on going process and we must be prepared to change our minds as new data comes available.

Lucia
See:
SAP 1.2 DRAFT 3 PUBLIC COMMENT
Chapter 1 Executive Summary 1
CCSP Synthesis and Assessment 1 Product 1.2
2 Past Climate Variability and Change in the Arctic and at High Latitudes

Especially Page 45, 46
—————————
965
966
967
Figure 6.1. A “Weather” versus “climate,” in annual 968 temperatures for the
969 continental United States, 1960–2007. Red lines, trends for 4-year
970 segments that show how the time period affects whether the trend appears
971 to depict warming, cooling, or no change. Various lines show averages of
972 different number of years, all centered on 1990: Dark blue dash, 3 years;
973 dark blue, 7 years; light blue dash, 11 years; light blue, 15 years; and
974 green, 19 years. The perceived trend can be warming, cooling, or no
975 change depending on the length of time considered. Climate is normally
976 taken as a 30-year average; all 30-year-long intervals (1960–1989 through
SAP 1.2 DRAFT 3 PUBLIC COMMENT
Chapter 6 Past Rates of Change 46
1978–2007) warmed significantly (greater than 977 95% confidence), whereas
978 only 1 of the 45 possible trend-lines (17 are shown) has a slope that is
979 markedly different from zero with more than 95% confidence. Thus, a
980 climate-scale interpretation of these data indicates warming, whereas
981 shorter-term (“weather”) interpretations lead to variable but insignificant
982 trends. Data from United States Historical Climatology Network,
983 http://www.ncdc.noaa.gov/oa/cl...../cag3.html (Easterling
984 et al., 1996).
985
———————————
You may wish to weigh in with your expertise and analysis shown here. Deadline in Monday August 25th.

lucia and mike.

this is too funny. I think I told lucia before that my father in law did static mixer design. mixing in a pipe, a very tricky problem. he would sit there for hours showing me the math.boggled me, who could think flow in a pipe could be so complicated. Anyway, One day I ask him, how did this flow in a pipe problem come to interest you. “oh that was easy, I built klystrons for varian” waves in a tube .fluid in a pipe. haha.

One of my favorites

http://www.komax.com/products/sludge_mixer.html

Mosh

Plumbing is plumbing!

A friend of mine thought that a course he saw that was named something like “turbulent flow in lumpy fluids” was really funny. It was on waste water management.

Kim: “The climate is cooling, so any warming effect of CO2 can be tolerated until we understand it well enough to do something about it, if necessary.”

I agree with Lucia`s disagreement with you (such an appeal to authority is unfair, I know, but I couldn`t resist; that`s how us enviro-fascist alamists work .

Further, presumably you do realize the long-term nature (hundred of years) of the forcing and climate response AND the continued growth of forcing activities, so that delay runs serious risks of finding that we are UNABLE to respond, other than to prepare for and deal with what may be largely unavoidable??

“But why did you bring up the strawman of suggesting that lucia’s work here led to my assertion?”

Sorry, I didn`t suggest it did, but rather simply remarked that her work doesn`t support you. Can you confirm that you do NOT take the view that Lucia`s work supports you?
**********************************************
TT

Raven, I have to disagree with both your summary of the IPCC`s “essential message” and your conclusions about the centrality of the GCMs

“1) The warming since 1960 can only be explained by CO2″
Isn`t the IPCC position a more modest one that human activities (GHGs, soot, etc.) are chiefly responsible?

“2) The planet will continue to warming at an accelarating rate as long as we emit CO2″
Doesn`t the IPCC recognize the logarithmic nature of the forcing?

“3) The consequences of the warming are very bad”
The IPCC acknowledges that there are benefits as well as costs (heck, it costs money to adapt to beneficial climate changes, too). Isn`t there position that, in net, the costs outweigh benefits and that there are very significant risks?

“4) Rapid reductions in CO2 emissions are the only way to deal with these negative consequences.”
The IPCC notes that even rapid reductions won`t have effects for decades, so that adaptation is needed, is not focussed solely on CO2, and notes that sequestration, CCS and geoengineering may be various ways to mitigate.

“the IPCC claims are based primarily on an anlysis of GCMs outputs which may or may not have a connection with reality. If these GCMs are not reliable then the IPCC claims must be dismissed.”

The IPCC claims are based on physics; the GCMs are simply tools – and obviously ones that will forever remain imperfect – to understand the consequences of our planetary tinkering – as we have even greater limitations on our ability to create actual duplicate Earths on which to run our experiments. Does the fact that weather foercasting is imperfect mean that the models on which they are based have no value?
*******************************************
TT

Tom, I don’t believe the hundreds of years effect hysteria. Increased CO2 in the atmosphere and the oceans will increase the feedback mechanisms which sequester the carbon back out of the biosphere. Insofar as lucia’s work supports my belief, it is in the disconfirming of the IPCC’s model of CO2′s greenhouse effect. It has been exaggerated.
================================

Kim–
Yep. You understand the relationship precisely correctly. I’m looking at the accuracy and precision of projections. People use models to estimate sensitivity, to assist with attribution studies and projections. In so far as the models are either inaccurate or imprecise, it’s possible that the effect of CO2 could be exaggerated or underestimated.

On a thread that Roger Pielke, Tom seemed worried that “some” were confused about the meaning of my hypothesis tests or Rogers “Consistent with Chronicals”. I don’t know who those “some” may be, but clearly, you know how the two connect.

My impression is most my readers understand that what I do is rather narrow, and how it falls into the larger question.

Tom,
“Isn`t the IPCC position a more modest one that human activities (GHGs, soot, etc.) are chiefly responsible?”
If you look at the various forcings you will see that GHGs (primarily C02) are the dominate human induced positive forcing. Aerosols cause a net cooling which offsets the GHG warming. IOW – according to the IPCC the warming of the planet since 1960 cannot be explained unless they include GHGs. (See AR4 Chpt 2 Fig 2.22)

“Doesn`t the IPCC recognize the logarithmic nature of the forcing?”
All of the business as usual projections show temperatures following an expontential curve do to increased human emissions and the effect of any tipping points. Even if we eliminate CO2 emissions by 2050 the IPCC predicts that the warming from 2000-2100 will be at least double that from 1900-2000.

“Isn`t there position that, in net, the costs outweigh benefits and that there are very significant risks?”
They may acknowledge that there are some positive effects but the IPCC message is the bad effects are so overwhelming that no cost should be spared in an effort to reduce emissions. This claim is made despite the fact that there is zero evidence of any net negative effect as a result of the warming of 0.7 degC from 1900 to today.

“The IPCC claims are based on physics; the GCMs are simply tools”
As I noted in my original post the IPCC claims go way beyond the simple physics of radative forcing and postulate that *only* GHGs emissions can explain the warming and that warming will accelerate unless we do something. These are conclusions based on the output of GCMs and would not be credible without them.

In fact, I think GCMs actually reduce our understanding of climate because they can only take into account things that can be modelled. For example, there is ample evidence that long term “ocean weather” affects climate on the decadal scale yet these effects are automatically dismissed because they can’t be modelled. The long term effects of cloud variations are dismissed for the same reasons.

In short, GCMs, like a hammer, are useful tools but it is important to understand their limitations. Anyone who claims that a GCM output should be treated as a proven fact is someone who does not understand the limitations of the tool. For that reason, any economic decisions we make must take into account the possibility that the GCMs are very wrong.

Hi Raven,

The claim that models are based on physics raises my blood pressure every time it is made. It carries the implication that the models must describe reality because physics is highly regarded for its ability to make predictions about our world.

When it comes down to it, we have some universal laws, some measured physical constants and some empirically determined relationships. These are probably enough to do sums about radiation distributions if everything else stays the same. That is where the problem starts. With the climate, just about everything affects everything else and we cannot use our tested methods of running controlled experiments to systematically explore all the interactions.

We are reliant on nature to provide all the conditions for a non-repetitive experiment, leaving us to try to disentangle all the resulting patterns in the variables. Until we have done this we are in no position to do sums in a GCM and claim we have the physics right.

From the very start models have had the idea of energy balance built into them. Radiation induced extra energy will be fed into the mass of the earth and so it must warm in order to restore its equilibrium. When it comes down to it, every model from the very simplest to the most sophisticated contains this built in assumption and it is no surprise that they all show warming.

As we have not had radiation monitors across all parts of the globe for an extended period it is hard to be even sure that there is an increased energy input in practice. The second part of the assumption that something like Newton’s law of cooling applies is similarly untested.

GCMs are good for telling you what would happen if the physics embodied in the model were a true representation of the earth climate system. How much they can tell you about the real earth seems to be the key question.

That is why I think Lucia’s work is so important as it is directly aimed at answering this question. The thing to remember is that it is nature that is running the experiment and we have to be patient until all the results are in. If climate really does exhibit LTP we might have to wait a very long time!

Jorge, funny. I was going to say that the claim that the models are “based on physics” makes my blood boil

Saying the models are based on physics is like saying disney’s the “lion king” is based on biology. Both statements are true but that does not mean the end result is an accurate representation of reality.

Here is a sumamry of the US CCSP report on aerosols. They make it clear that aerosols are used as “tuning” factors that allow the GCMs makers to produce whatever results they want (i.e. CO2 sensitivity within the “consensus” range).

http://www.climatescience.gov/.....ec-sum.pdf

“The Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4) reported on the results of some 20 participating global climate models. These models can reproduce the observed trend in global mean temperature over the twentieth century due to changes in atmospheric concentrations of greenhouse gases and other forcing agents including aerosols. When anthropogenic aerosol forcings are not included, the models tend to generate too much warming. However the ability of climate models to reproduce the global mean temperature change over the past 100 years appears to be the result of using a “tuned” aerosol forcing. Although different models exhibit a wide range of climate sensitivity (i.e., the amount of temperature increase due to the increase of CO2), they yield global temperature change, which is similar to the observed change. Apparently this is because the forcing by aerosols differs between models. For example, the direct cooling effect of sulfate aerosol varies by a factor of 6 among the models, because of different extensive aerosol properties (e.g. sulfate amount) and different intensive properties (e.g. scattering efficiency) used in the models. Greater disparity is found in the model treatment of other aerosol types such as black carbon and organic carbon. Even the choice of which aerosol types and which aerosol forcings are treated in a particular model varies. Some models include only the direct aerosol effect, whereas others include an indirect effect in which the
aerosols modify cloud microphysics and hence cloud brightness. In addition, the aerosol indirect effect on cloud brightness varies by up to a factor of 9 among models. This situation is in part a consequence
of the large uncertainty in the mechanisms and magnitude of climate forcing by aerosols, and in part due to the differences in cloud amounts between models.”

Tom,

When you refer to the IPCC`s “essential message” are you referring to the summary for policy makers or the body of the report? The body of the report reads like a typical scientific exchange with lots of qualifications and uncertainties and a good deal of hopeful remarks about how they are getting better and better at various things (like incuding clouds, for example). But the summary reflects very little of the uncertainty (about 10% of it according to their pseudo-numerical estimate). Of course, it’s the summary that informs most of the press.

As I understand it, the summary was published months before the report it was meant to summarise, so it is in fact best understood as a summary BY policymakers rather than FOR policymakers.

Of course, the 2C/century prediction itself has uncertainty associated with it. To treat it as uncertainty-free is to conduct a one-sample test against a population mean. To include that uncertainty is to treat it as a two-sample test: a theorical sample vs. an observed sample. Very different. The two-sample test is harder to refute than the one-sample case because the two distributions share more overlap.

Recall that Gavin tells us that for any given 8-year period model E simulations produce a cooling trend 9 times out of 55. So the GCM sample mean is not 2C/100y but something like, say, 2+/-1C/100y.

Top scientists agree revolutionary climate control project can reduce hurricanes, droughts and wildfires.

Climatologists, biologists and physicists from all corners of the globe agree that U.S. based Gravitational Systems, L.L.C.’s revolutionary clean power climate control project INDRA will improve the lives of billions of people around the world.

Concerns have been raised about the projects impact on biodiversity as deserts are terraformed to rainforests.

Mr.Henderson, director of research and development for Gravitational Systems, L.L.C. ( a clean power developer), explains that the INDRA project, a proposed network of specialized evaporation channels moving sea water from the oceans toward the deserts, will convert world deserts into biodiverse rainforests. Deserts which cover 1/3 of all dry land will be terraformed into productive land.

The INDRA systems will give mankind control of the weather, ending dangerous storms such as hurricanes, typhoons, tornadoes, and dry heat waves within a decade.

Vast rivers can be turned on and off in hours, and reservoirs and salt marshes drained or replenished in days.

The increased bio-mass of the terraformed deserts will begin to reverse both global warming and thermal sea level rise. UNFCCC cap and trade certification of the INDRA project will allow individuals and business to fund the plan through carbon offsets. The initial projects will be targeted north American, and north African deserts.

For more information contact:

Inquiry@gravitationalsystems.org
Gravitational Systems, L.L.C.
P.o.Box 2066
Washington, D.C. 20013
website: http://www.gravitationalsystems.org/INDRA

Kim, so you don’t believe the “hysteria” that forcings act on scales of centuries and millenia? Pray tell, can you let us know what you do “believe” as to how soon we feel the full equilibrium effects of natural and anthropogenic forcings and feedbacks, how long CO2 and other GHGs remain in the atmosphere, and why the Keeling curve is still rising?

My own understanding – an hysterical one, admittedly – is that a third or more of CO2 emissions remain in the atmosphere for centuries (so that the full forcing of previous emissions and feedbacks will similarly be exerted for some time to come), and that the chief mechanism for long-term removal of carbon is weathering of minerals and sequestration over millenia via the formation of calcium carbonates by marine life (a process that ocean acidification slows). Take a look at the two most recent publications by Richard Zeebe in Science and Nature Geoscience here, for example: http://www.soest.hawaii.edu/oc.....tions.html.
************************************
TT

Raven:

1. I agree that the IPCC takes the position that the warming of the planet since 1960 cannot be explained without including the forcing from the GHGs (which the IPCC says is the predominant human forcing). However, this is not the same as saying the “The warming since 1960 can only be explained by CO2.”

2. Now you say that “All of the business as usual projections show temperatures following an expontential curve do to increased human emissions and the effect of any tipping points. Even if we eliminate CO2 emissions by 2050 the IPCC predicts that the warming from 2000-2100 will be at least double that from 1900-2000.”

The IPCC recognizes a logarithmic forcing, as I noted. AFAIK, only the A2 BAU projection looks anything like exponential, because it tracks escalating GHG emisssions – a track that the Keeling curve actually follows.

That we may continue to warm even if all GHG emissions were fully mitigated is a question of the long-term nature of the forcing.

These points are far from an argument that, as you put it, “The planet will continue to warming at an accelarating rate as long as we emit CO2.”

3. The IPCC position is not a simple one that “The consequences of the warming are very bad”, nor does the IPCC argue that “no cost should be spared in an effort to reduce emissions”.

Finally, can you point me to anything that establishes “the fact” that “there is zero evidence of any net negative effect as a result of the warming of 0.7 degC from 1900 to today”?

4. You first argued that the IPCC’s message was that “Rapid reductions in CO2 emissions are the only way to deal with these negative consequences;” while I noted that their position is more sophisticated, it think it is fair to conclude that they argue that aggressive mitigation is warranted.

However, as noted in 1,the IPCC does not “postulate that *only* GHGs emissions can explain the warming”. Yes, the IPCC A2 projection suggests a fairly straighline increase in warming (not acceleration) “unless we do something”. While the various projections are certainly based on the GCMs, one does not need a GCM to concluding that increasing climate forcing might warm the climate.

I’m happy you agree that the GCMs are “useful tools”. Of course it is important “to understand their limitations”, we should not treat GCM output as a proven fact, and we must take into account the possibility that the GCMs are likely to be wrong (this cuts more than one way, of course).

I fail to follow your argument that GCMs “actually reduce our understanding of climate because they can only take into account things that can be modelled”. That some aspects of the climate system are not well-modelled means we have to take the outoput with a more than a grain of salt – but again, the unmodelled aspects may cut in various directions and research is actively preceding on them.
*****************************************
TT

david, have you read the SPM as well? It is full of qualifications. Which of its statements do you think are unjustified/overstated?

http://www.ipcc.ch/pdf/assessm.....yr_spm.pdf

The SPM was approved by goverments – as the governments require. Many, like the US, Australia and the developing geconomies that were hardly champing at the bit last year for GHG restrictions. I think that if anything, the political checks on the IPCC toned down the result, as we know happened in the past.
*************************
TT

Tom,

You are parsing words too much and I don’t think I want to start arguing with a lawyer about the meaning of words so I will concede that I was not precise in my previous statements. However, the essense of my argument is that the IPCC claims that warming from 1960 cannot be explained without included human GHGs (the majority of which is CO2) and that there are no natural factors that could explain any of the warming (i.e. Chap 9 Fig 9.22 shows the earth cooling since 1960). This assumption is necessary to get the models to match historical temps given their assumptions w.r.t CO2 sensitivity. If the IPCC has missed significant natural factors then they would need to revise the CO2 sensitivity estimates downward. This, in turn, would reduce the amount of future warming. That is why I said rejecting the IPCC claims does not require one to reject the basic premise of CO2 induced warming.

When it comes to the future protections the graphs I was looking at clearly showed an exponetial curve for the BAU scenarios. It does not make a difference whether this comes from the assumptions about CO2 sensitivity or the growth rate of emissions – the IPCC is still projecting exponetial growth in temperatures.

I think the onus is on you when it comes to providing conclusive evidence of the harms caused by the 0.7 warming to date. I can’t think of any harms.

Lastly, computers are the dumbest things in the universe because they do exactly what you tell them to do. This means one cannot prove anything with a computer model because a computer model can only account for things that you already knew about. Unfortanately, many people in the climate science community now see models as way to “prove” whatever pet theory they may have and automatically reject any theories that cannot be demonstrated in a model. This can result in bad ideas getting acceptance and the pre-mature rejection of good ideas. Such problems do not exist in sciences where real life experiements can be used to seperate the good from the bad.

[...] Data and Falsification Filed under: Uncategorized — cshme @ 5:28 am Lucia recently posted another in a series of hypothesis tests comparing the central tendency of the IPCC [...]

Bender- Great point. I’ve completed some calculations doing an analysis along those lines. Nothing terribly out of line compared to what Lucia has found, but my results differ significantly.

“No natural factors can explain the temperature increase since 1960.”

First of all, there was “cooling” from 1944 to 1976-79.

Second, the PDO shift to more, longer lasting and more intense El Nino’s can explain almost all of the warming since 1979.

Third, the unnatural part of the increase which is easily explainable is the “adjustments” made by Hansen and Jones to the temperature record combined with the increase in the unnatural urban heat island effect.

I believe that explains most of it.

Now throw in some modest warming from increased GHGs (1.5C per doubling) and modest cooling from aerosols and three big volcanoes and we have explained just about everything.

Tokyo Tom. Comment#5178. So you fall back on the ever reliable “hard to spot any signal” proof? Nice one.

http://www.acrim.com/RESULTS/E....._fig26.pdf

See correlation of Solar to Temp since 2001, no volcanos etc.

There was cooling up to 1979. Then in 1981 Hansen releases a paper predicting that Co2 would cause between 0.2 and 0.4 degree of warming from 1980 to 2000.

And actual temperature trend measured both by ground records that could be influenced by UHI, but are mostly influenced by sea temp recordings (70% of earth surface), and only one of which was ‘adjusted’ by Hansen, and by satellite records, between 1980 to 2000 is quite close to the 0.4 degree point.

And PDO vs Co2? Both have enough correlation with the warming to be considered possible candidates to explain the warming. Co2 has a well known and modelled physical mechanism by which first 1/3rd of the observed warming is caused by direct radiative forcing, and the other 2/3rds of the warming happens due to feedbacks such as ice albedo and water vapour GH feedback.

Can anyone tell me how much forcing the PDO has on the climate? What feedbacks might be needed for PDO to explain the observed warming? Or some other equivelant details about how the PDO affects climate to put the PDO theory on anywhere near level footing with the CO2 theory? Have there been any successful predictions of future climate trends based on observations of the PDO?

Tom,

For example

“The understanding of anthropogenic warming and
cooling infl uences on climate has improved since
the TAR, leading to very high confi dence7 that the
global average net effect of human activities since
1750 has been one of warming, with a radiative
forcing of +1.6 [+0.6 to +2.4] W m–2 (see Figure
SPM.2). {2.3., 6.5, 2.9}”

with the footnote:

“7 In this Summary for Policymakers the following levels of confi dence have
been used to express expert judgements on the correctness of the underlying
science: very high confidence represents at least a 9 out of 10 chance
of being correct; high confidence represents about an 8 out of 10 chance of
being correct (see Box TS.1)”

So what does this footnote mean? Clearly the degree of confidence can’t be 10/10. That would mean that the job is done, much reduced funding, no new supercomputer. And in any case, even journalists know that the world is uncertain and that climate science would be an unlikely exception. So it seems to me that 9/10 is the absolute practical maximum (OK, you could do 99/100 but that would defintely look suspicious). I think this directly contradicts the uncertainties discussed in the main report.

The other point here is the IPCC explanation of how this number was arrived at. Not a quantitative model of some sort, as you might expect, but “expert judgements”. Who are the experts? The authors of the models? And is each group just judging their own model, all awarding themselves 9/10? Or are they all judging all the models (forgotten how many)? In that case the outcome is a little odd, in that the models differ so surely the authors of each model are less confident of the other models than they are of their own. I feel a theorem coming on, but fortunately it’s not necessary in the light of Lucia’s work.

But I do agree that the SPM contains more qualifications than my comments implied.

Chad, #5218, has shown that lucia’s conclusions do not apply for 4/15 of the models – the ones with the largest spread among individual runs. (I’m amazed that he even read my #5211, let alone answered it.)

Bender–
I’m just back from a long weekend.
I’m trying to figure out what chad’s post is saying he found.

It looks like he ran a two sample t-test to decide whether the trend in say “model A” was equal to the trend in the data. Right?

Yes. The model with huge weather noise failed to reject. The models with smaller amounts of weather noise (particularly close to that of the earth rejected.)

I’m actually trying to do things much more slowly than Chad to try to figure out other questions. As in really low level questions. So, next week is slated for posta that should fall under “Boring posts that must be written before getting to the ones we really care about!”

lucia, I am pretty confident your slower, more methodical approach is going to lead to deeper insights. His presentation also contains some superfluous junk (like ACFs of smoothed MA process) that distracts more than informs. I liked his method of taking ACFs of the unsmoothed output from individual GCM runs. The product is very clearly unambiguous white noise. An ACF of an LTP process will not behave well. I’m not even sure the autocorrelation coefficients converge over time. My hunch is that they don’t!

BarryW, thanks for your response (#5182).

- “Sea level rise did not end centuries ago and was occurring before the significant rise in CO2 or any other possible human influence. Plus it’s gone down over the last few years along with the temps. How does that work with CO2 going up?”

Good question at the end, but you’ve dodged the gist of my point. As Wikipedia notes, from 3,000 years ago to the start of the 19th century sea level was almost constant, rising at 0.1 to 0.2 mm/yr. Since 1900 the level has risen at 1 to 2 mm/yr; since satellite altimetry from TOPEX/Poseidon indicates a rate of rise of 3.1 ± 0.7 mm/yr (over 1993-2003). More recent satellite data shows that levels have rose by an average of 3.3 millimetres per year between 1993 and 2006 (while TAR projected a best-estimate rise of less than 2 mm per year).

- I object to both “denial” and “hysteria”, which are two sides of the coin of our problems with our cognitive predilections.

- I agree that we also have other problems to worry about, including ocean fisheries, dead zones, tropical deforestation etc. Such problems stem from the lack of effective ownership. While we certainly ought to pay attention to these, that’s no argument that we should ignore our own influences on climate.

- You may not have noticed, but your summary of what “isn’t science” isn’t science. Science has established that human activities are creating a net forcing; trying to identify the effect of that forcing on climate is also science. Presumably you don’t assume that a forcing has no effect.

- Who said that we’re in the majors when it comes to terraforming? I do appreciate your implication that deliberately altering climate may be possible – that, of course is what underlies the discussions about possible geo-engineering.
*******************************
TT

Raven, sorry for holding your feet to the fire on your ambiguous and over-broad statements. (That’s just something unfair hysterics like me do!)

I do agree that the SPM says that man has been exerting a forcing over the past few decades, during which it concludes natural forcings have been negative, with the result that human activities must be responsible for some of the warming.

Do you really think that there have been no costs to the warming so far (as opposed to no NET costs)? If so, you have missed an awful lot of information. I discuss a small example of a cost of warming here: http://mises.org/Community/blo.....sheds.aspx
*********************************
Tom

A bit of tract but seems plausible… Is this serious? Any comments please
http://www.globalweathercycles.com/
The author, a meteorologist claims to have found a definite link between changes in the moon gravitational pull on earth over 500 milion years is linked to climate change .. by shifting sea/air pressure areas north and south

vhguerrini

Vincent– Well, I’m not going to spend $10 on the ebook to learn more about his theory.

I also have to seriously doubt anything that makes claims like this:

GWO also documented a near 100 percent correlation between PFM cycles to regional droughts, regional floods, regional hurricane landfalls and regional seasonal precipitation.

A 100% correlation between one random looking thing and another? Well, if this guy did find this, he ought to start trading commodities. If you know all the regional droughts and floods a head of time, you can do a good job predicting crop failures. (Better than Joseph with his technicolor coat and dream interpretation for Pharaoh!)

#5232
o ye of little faith

#5231

This brings new meaning to the term “mooning” someone.

#5232

I’m sure he’s smart enough to have done that, or he feels that he must work for the betterment of mankind rather than crass commercialism.

Maybe the author should do an infomercial.

#5231,#5232,#5333
Yet,wouldn’t it be good if someone could determine whether gravitational forces within our solar system might contribute to the baroclinic instability that affects so much of our weather and to the planetary oscillations — including PDO, AMO, El Nino and La Nina — that impact climate? Or do we have all the answers we need to those? When the moon hits your eye like a big pizza pie, don’t ignore it.

I don’t know if you have seen it, but Robert Grumbine has a response to this post:

http://moregrumbinescience.blo.....eas-2.html

Vincent, (#5231) I don’t know anything about the book, but slightly related (although differing proposed mechanism), there have been various hypotheses offered and kicked around wrt to varying tidal effects increasing and decreasing verticle mixing in the oceans. Here’s an online, freely accessible one from (the late) Charles Keeling:
http://www.pnas.org/content/97/8/3814.full
One of the references, also freely available online, discusses more than I’ll ever want to know about oceanic mixing:
http://ocean.mit.edu/~cwunsch/.....nsch98.pdf

Darwin–
It’s one thing to speculate that the moon has some effect of some sort, explain your theory and see if it holds up when compared to data. The explanation might turn out to be convincing or it might not.

It’s another thing to launch a web site promoting a $10 ebook proclaiming they’ve found a perfect statistical correlation between “something” and “global warming”, and I can learn more by sending them money. I’d bet a plate of brownies that the perfect statistical correlation described in the $10 ebook will be the result of massive data mining and over fitting. Call me a cynic.

I think I’d rather spend my money on Tom Chalco’s bioreasonant t-shirts.

#5239 Or one of those mugs where the polar ice caps disappear when you pour your coffee. Hey lucia, did you know wind power is killing bats? Guess the mechanism.

Are the bats running into the blades? Or is it screwing up their ability to find prey by interfering with the sonar or whatever it is they bats use to find things?

Nope. There’s a significant low pressure area near a working airfoil.
They’re mammals. The alveoli in their lungs explode and they drown in their own blood. Though I wonder if they fly into danger because they see bugs also caught up in the problematic air and go after them.

The migrations show up on radar, are quite constrained in airspace and timing, and bats migrate mostly during less windy nighttime conditions, not during peak load times. It’s manageable.

Supersonic ‘screamer’ whistles like the ones used to warn deer off the highway might work.

Migration happens only occasionally, but feeding happens nightly, windy or calm. Take out the bats and up go the bugs.

Hank–
Wow! That’s amazing. I googled. The blades don’t move all that fast, and the wind velocities are clearly going to be low mach number. The pressure drop can’t be all that huge. (Of course, it’s “huge” compared to just flying around–but it’s not something I would have dreamed would damage anything.)

So…. bat lungs must be really delicate.

I posted your link to a new climate blog and the author responded with the following critique:

http://moregrumbinescience.blo.....eas-2.html

I would appreciate if you respond either here or on his site

Jonathan–
Gosh! Which thing do you want me to address? That’s quite a catch all.

The blogger, “penguindreams” is not the first to defend the IPCC projections against falsification by the odd claim that the IPCC projections do not have a central tendency and/or aren’t projections or whatnot. Evidently, he believes no central tendency exists so because the authors of the AR4 didn’t use that specific term.

The IPCC did make projections. They are discussed in words and figures. This particular figure appears in several places in the document:

This graph appears in chapter 10, and again on page 61 of the technical summary.

Central tendency is a term of art in statistics; Penguindreams,the author of that blog, might do well to look it up in the dictionary. Penguindreams can look here.

You’ll note that the term is most typically used to mean “average”, which is how I use it. The “average” value is illustrated on the IPCC figures by the bright lines inside the fuzz. So, whether or not they IPCC used the word “central tendency”, the central tendency is there for all to behold. It’s the dark solid lines indicating the “average” for all models.

Evidently Penguindreams also thinks the IPCC projections do not have a central tendency of 2 C/century.

The AR4 does provide numbers very, very close to that in tables in the text– more over, if you examine the graph, you will see 2C/century goes right through the “average” curves– at least during the first two or three decades. (And I say first two or three because the authors of the AR4 switch back and forth in the AR4 itself.)

If Penguindreams doesn’t believe 2 C/century is correct, he might want to visit real climate, which adressed the falsifications issue. Gavin says”

This figure shows the results for the period 2000 to 2007 and for 1995 to 2014 (inclusive) along with a Gaussian fit to the distributions. These two periods were chosen since they correspond with some previous analyses. The mean trend (and mode) in both cases is around 0.2ºC/decade (as has been widely discussed) and there is no significant difference between the trends over the two periods.

Though Gavin doesn’t agree with my method of testing a hypothesis, you will see that he does agree the central tendency of the IPCC projection is 2C/century for right now.

Of course, if Penguindreams wants me to address a different number– possibly 2.1 or 1.9 C/century, I would be glad to address it.

He seems to be worried that the IPCC GCMs did not account for the well known 11 year solar cycle. First, some of the GCM’s do account for the well known 11 year solar cycle. So, the projections were not run with the sun on “full bright” associated with the top of the solar cycle near 2000.

Second, those that don’t account for the variability solar cycle set the level to “medium”. The don’t set the sun on “high” in 2000 and then let the models roll!

Third: the reason they don’t include the 11 year solar cycle is the modelers believe this doesn’t matter much — either in the forecast of the hindcast back further than 1900.

Fourth: Had the IPCC “method” believed the solar cycle mattered, they could have based their projections on the subset of models that include the 11 year solar cycle in their computation. It’s a well known cycle– in fact, it’s more predictable than the GHG loadings! But, the IPCC specifically decided to project using a combination of models, some of which include the 11 year solar cycle and some of which do not.

Permitting modelers to neglect the 11 year solar cycle is the IPCC “method”. Their projections are what I test. I don’t need to “correct” their projections for this. I test the projections they made based on their judgment. If others wish to test the projections they think the IPCC should have made but which they IPCC chose not to make, they can do so.

“Correcting” for the well known 11 year solar cycle would not, be a test of the IPCC projections!

So anyway on Penguin dreams bullets:

* We have to compare 20 year average vs. 20 year average (the variable being suggested as meaningful in the report, and 30 would be better).

There is a general principle that we must compare like to like.

However, there is no general principle restricting us to 20 years. The IPCC communicated their projections in multiple ways. One method was to provide the graph above which shows the smoothly varying projection. The other way was to provide a table with readings taken from the graph at the 20 year point.

The IPCC communicated their projection in a prominent, and widely circulated figure, shows a linear trend over the first few decades of this century. That linear trend applies now. I compare the central tendency for the trend they predict for the period between 2001-now to the range of trends consistent for data during the same period.

That’s comparing like to like. That the IPCC also gave specific numbers at 20 years doesn’t preclude my testing the projection they communicate in their figure.

(Oddly, if we really did have to wait for 20 years to pass, the IPCC would be violating their own rule by having compared their TAR projections to data already. That document also contains a table showing their projection 20 years into this future– and yet, somehow they compared data to models in 2007. Go figure!)

* We have to look only at global mean surface air temperature.

I am comparing to global mean surface temperature as measured by HadCrut, GISS Land/Ocean or NCDC/NOAA measure. The 2C/century is inconsistent with that.

I also compare to the satellite measurements, using the readings near the earth’s surface. Some people don’t like the satelitte comparison, but the mere fact that I list it doesn’t somehow transform HadCrut into something other than global mean surface air temperatures.

* We have to adjust for the fact that the sun has been giving less energy than assumed in the projections.

No we don’t need to adjust for this.

(That said, this is an ongoing issue with JohnV, who is trying to look into what the models actually say about the sun.)

* In making the test, we have to allow for the fact that the 0.2 C/decade itself has error bars (due to interannual variability and between-model variability)

There are many possible approaches testing and many things that can be tested.. I account for interannual variabilty of the true earth with my uncertainty intervals. See the ±x.xC in my tables above? As for the between-model variability, whether or not one accounts for that depends on the question one wishes to ask.

I wish to ask the question: For now, I ask if the central tendency of the projection falls inside the range consistent with the recent earth GMST?

There is no reason one may not ask this question about the central tendency. If one can ask it, one can test it. To answer this question, on ignores model-to-model variability. Period. The reason for this, is one is asking about the central tendency only.

The 2C/century central tendency for the projection falls outside the range consistent with earth’s weather since 2001.

One could later ask other questions — many of which are interesting. Determining the answer to those other questions might require us to consider the inter-model variability. But answering the one I ask does not require me to consider that.

If Penguindreams wishes to do hypothesis tests to test other questions, he’s welcome to do so.

Altogether too rational. I will be impressed if penguindreams learns anything from this.

Bender–
Did you read it? It goes on and on. Evidently, I should use four satellite data sets rather than two. However, I’m actually supposed to use none at all. And RSS is bad because it doesn’t cover the poles. (Never mind that the IPCC figures all compare observations to HadCrut, which also doesn’t cover the poles. )

Also, evidently, one of his commenters says: “I also find it odd that Lucia has not published a similar analysis on older projections for climate change, such as IPCC reports 1-3, or Hansens 1981 paper.”

Erhmm…. I’ve posted something on all of those. Is Hauber under the impression that ever single possible question about every single projection ever done will be discussed in every single blog post? Just imagine how long they would all be?!

I focus on the most recent IPCC projections for the obvious reason: They are the most recent ones. If the IPCC still adhered to the TAR projections, I’d focus on those.

Benderdreams, (#5270) Grumbine is a knowledgable fellow, which is why I was pretty puzzled when I read his response.

When people are losing their heads like this it is a great time to buy some stocks and sell others. Just make sure you have a hedging strategy for when the mania swings the other way. Know the early warning signs and consequences.

Lucia,

Here is what Grumbine says about 2 deg/century:

“Two points about 0.2/decade for 2 decades vs. 2 C/century. First, it’s a matter of honest reporting. If the source says one or the other, you should quote the one they say. Second, how much money would you owe me if it were $1/months for 2 months rather than $12/year? The latter leads you to a different,and erroneous, conclusion. Different side is that the original said ‘about 0.2/decade’. Anything from 0.15 to 0.25 can be ‘about’ 0.2.”

Since his blog is read by students in the Boulder area, I would appreciate it if you post a response on his Blog.

Thank you

#5282

That is one of the most specious arguments I’ve seen. If the .2/deg per decade is projected to go on for 10 decades it’s the same bloody thing (2deg/C).

From Gumbine:

It’s a little work, I know, but in order to have intelligent disagreement, or agreement, we have to know what the other is saying

Obviously he doesn’t believe in following his own advice, since he has neither checked your other posts for background nor asked reasonable questions. Says you didn’t use four sat data sets but provides no links, which he complains you didn’t do. And insulting since he implies cherry-picking (lying) in your choice of data. If students are reading his blog this is sad, even if he’s getting some things right. If he was a little snarkier he’d fit right in at Open Mind or RC, he just has to try a little harder that’s all.

#5239, you can make that point without denigrating the idea that gravity matters, which you now have. (… made the point without denigrating …):<} Of course, what the guy is doing isn’t much different from following the lead of Nature or Science where you either having to subscribe or pay out even more money for a single article — he’s just doing it on a smaller scale. I will save my money for trips to BA and Villa La Angostura, and my time for reading your blog.

Darwin–
I’m not denigrating the possiblity that gravity matters. I am suggesting that the specific site suggested above seems unworthy of my $10. I can get journal articles through work or from the Lisle public library. I would have to buy that self published ebook to learn the contents.

It may be unfair prejudice toward self published ebooks, but generally speaking, I don’t trust self published ebooks promoted with slick looking web sites.

I agree it would be nice if Nature and Science articles could be downloaded for free by anyone with an internet connection. But it’s still a bit different from the ebook!

Lucia, (#5290) You’re asserting that the book costs $10, yet we see on the website that it’s going for a paltry $9.95. Clearly this is not reliable reporting. But if you really wanted to read the book, if I gave you a dollar per month, ermm, the sun’s been awfully quiet, ummm, hence, therefore, the book’s predictions must be correct!

Mike N.

Sorry for my unforgiveable rounding up! I will fork over the $9.95, print out the book and eat the paper. Do you think Maple syrup will improve the flavor?

This really belongs over at
the post in which Lucia takes me to task, but comments are disabled over there.

Let me begin by illustrating the level of comment included in Penguindreams’s criticism of my recent blog post (which you did not link. This robbed his readers of the opportunity to actually read the post he criticized.)

My post opens: Finally, in comments on my cherry-picking article, I was invited to take a look at http://rankexploits.com/musing.....-rejected/

It looks like the typing provides a correct link, readily put in to any browser.

One can ‘construe’ all sorts of things. You didn’t provide either your definition of ‘central tendency’, nor in the cited post any links to where you were obtaining it. One can guess, but if you’re making a serious criticism, I don’t think we should have to guess as to what, exactly, it is and whether the source actually said it.

Then you ignore where I follow:
In the lead paragraph, the author writes “… compared to the IPCC AR4’s projected central tendency of 2C/century for the first few decades of this century.” Again, I don’t find the IPCC saying that, and again, the author doesn’t say where the claim comes from. The nearest match I find is in the Summary for Policy Makers, p. 12, where it says “For the next two decades, a warming of about 0.2 C per decade is projected for a range of SRES emission scenarios.” The author mutated a term of precision, two decades, in to a vagary, the first few. Then the fuzzy term ‘about 0.2 C per decade’ got cast as a hard term of 2C/century. If nothing else, in reading this site, you’re not reading a reliable reporter. Once I’ve reached that point, I generally stop reading a source. There was no need to misrepresent the original. Nothing was saved or simplified.

So we’d all have been better off if I’d stopped as I usually would have. Or if you’d included the sources of your claim in the original.

In any case, where I quote you, you actually said what was quoted and the link was provided up front for people to verify what else you said as well. Starting with the opening of my note, you misrepresent what is there. Those who come over will see the reality, but it’s doubtful that everyone will come read for themselves — hence the importance of accurate representation.

I also managed to make my criticisms without loaded terms like ‘blather’, ‘jumps on’, ‘clintonian’, … or suggesting that you didn’t know how to use a dictionary. Yet you do so.

I avoid focusing on HadCrut because it currently shows the most negative trend, leading to the strongest falsification of any tests. In contrast, using UAH — which Penguindreams criticizes — does not falsify under some tests!

So? I’m not concerned about which sources agree or disagree with what estimate. If it happened that a satellite (or all of them) said that there was indeed a trend of +0.2 C/decade (+- 0.02) over the last 7 years, it’s still uninteresting as a test of the surface global mean temperature. Informative as to the nonlinearly averaged (the sensors make a nonlinear average of the temperatures) temperature centered at 13,000 feet.

Though I am entirely aware the satellites measure over a range of height, …

In your comment #5267 you said: I also compare to the satellite measurements, using the readings near the earth’s surface.

They make an average centered at around 13,000 feet (about 4 km) elevation. http://books.nap.edu/openbook......mp;page=42

Before I leave this section: in addition to complaining I used satellites at all, it appears Penguindreams also complained I use too few satellites. Penguindreams appears to have suggested that I might have chosen to include UAH ad RSS because they give the lowest trends. This guess is wrong: I included them because they are the most widely discussed at various blogs.

What I said was:
The cherry pick is that only 2 of the 4 satellite temperatures were taken, and it happens that the two are the two which show the least warming (you’d have to know about this, which is easy enough to find if you look but isn’t universal knowledge). The author gives no reason for this selection

It is true that you used only 2 of the 4 satellite records.
It is true that they are the ones that show the least warming.
It is true that you gave no reason for that selection.

The reason you now give … That the either the cherries were picked at the blogs you choose to read, rather than by you, or that a good selection was made by them for their purposes but that your purposes could be different doesn’t insulate you from cherry picking problems. If you consider the satellites meaningful, that’s one thing. Next is to use all, or to examine which ones are adressing the question at hand the best, and tell us how you concluded that.

Moving along:
For yet another illustration, including my program for doing so (after you follow a link), of how long and why one might need to average over global mean surface temperatures, see my second what is climate note. The approach there is different than the ones I’ve seen, in that I take on the question of how long one needs to average in order to decide what the climate anomaly (vs. weather) is for a given time. Suppose we want to know what the climate anomaly was for January 1990, in other words. How many months before and after would we need to average in order for modest changes in our averaging period not to substantially change the resultant? Answer turns out to be about 7 years before and after, with 10 years before and after being better. The demonstration there is casual, I noted. Anyone who’d like to make it rigorous or submit a comment on what is needed for rigor is welcome to do so.

Penguindreams–

* Comments seem open over there. But commenting here is fine.

On this:

In any case, where I quote you, you actually said what was quoted and the link was provided up front for people to verify what else you said as well

There are two parts there: The issue of mis-quoting and the issue of your posting a link.

Your link As far as I can tell, you don’t provide a usable link. On my browser, the text displays like this:

“Finally, in comments on my cherry-picking article, I was invited to take a look at http://rankexploits.com/musing.....century-st The commentator didn’t really say what I should be learning from the link or…”

The text of the link ends with ‘st’, not the “2ccentury-still-rejected”. There is no html wrapped around the text. There is no way for the visitor to click on the link or learn the full link.

If Johanthan hadn’t told me what you were discussing, I couldn’t possible figure out what post you might be discussing. Also, had you provided a link, technorati and google would have made me aware of the link in my “incoming links” section, and I would have read your article long ago.

There is no usable or comprehensible link. I did not misrepresent that.

Maybe the link appears in other browsers? If so, then I apologize for believing what is not a link on my browser is not a link for anyone. However, in that case, you might wish to be more careful in the future and make sure the link is really there.. Wrap the link in html so the browser or blogging software doesn’t “do” something to it.

Misquoting I have not suggested you misquoted me. I did, indeed, say the central tendency of the IPCC projection for the is 2C/century for the first few decades. My complaint is that you said I misrepresented AR4 as follows:

If nothing else, in reading this site, you’re not reading a reliable reporter. Once I’ve reached that point, I generally stop reading a source. There was no need to misrepresent the original. Nothing was saved or simplified.

Notwithstanding your finding a paragraph that includes the modifier “about” but uses “two” instead of “few”, “few” is accurate. This is readily apparent if you read the AR4, or are at all familar with the contents of the AR4.

You no longer seem to dispute this. Though, if you do, we can continue on that.

Note also, I did not ignore the first part of that paragraphs. I commented on your perverse jumping on “two” changed to “few” and “about” meaning heaven knows what. I provided copious evidence from AR4 to show the “about” in that paragraph expresses a very tight distribution, and “two” is taken from underlying analyses that draw from 2 or 3, and so should mean “few”.

If your issue is that you couldn’t tell my statemetn was correct because a) you didn’t previously know the number and b) I didn’t regurgitate a fully researched journal article providing links to each sentence, ok. We’ve learned something about your thought processes. We will also observe that you don’t post fully researched journal articles at your blogs and also don’t provide links for every sentence.

However, my statement about the central tendency remains entirely accurate– as I have shown.

On this:

So? I’m not concerned about which sources agree or disagree with what estimate. If it happened that a satellite (or all of them) said that there was indeed a trend of +0.2 C/decade (+- 0.02) over the last 7 years, it’s still uninteresting as a test of the surface global mean temperature. Informative as to the nonlinearly averaged (the sensors make a nonlinear average of the temperatures) temperature centered at 13,000 feet.

So what if you aren’t interested in UAH? So what if you aren’t concerned with wiether sources agree of disagree? So what if your lack of interest has some reasonable basis?

In your criticism of my post you said:
“We have to look only at global mean surface air temperature.”

First: My post contained comparisons to global mean surface air temperatures, in isolation and as a merge of three. So, this comparison of predictions to pure surface temperatures is there for those who find it interesting. The comparison is there.

Second: It is perfectly fine to discuss why you prefer the land based metrics. However, there is no rule that one can’t also compare to fiduciary measurements even if they are not the best measures. This is done all the time, everywhere in all scientific fields. It is even considered good practice, as otherwise things can go horribly wrong.

Even though the satellites do not measure the surface, they do get fuller coverage over the globe. They also don’t suffer from urban heat island. So, comparison is valuable.

In anycase, your lack of interested in UAH or RSS, for whatever reason, does not make the comparison to HadCrut, GISS or NOAA magically vanish from my blog post.

On the issue of cherry picking: You did suggest that I am cherry picking. Now you suggest this is the result of the blogs I read? Hmm… Well, it happens that I use the only two that mentioned at “Wikipedia” (http://en.wikipedia.org/wiki/S.....asurements. They are also only two mentioned at NOAA http://www.ncdc.noaa.gov/oa/cl.....h/msu.html

The reason these are the only two systems discussed at the blogs I read is these are the two most widely respected systems. I am not required to state this in every single blog post.

If you know of easily available data from two other groups — respected or not–, could you a) name them b) suggest where one might find links to readily available data. If the data are readily available, I’ll add them. That just isn’t an issue.

As for your original claim the other two satellite measurement give higher trends: How am I to know this is accurate? Heck, since you complain I don’t provide links to back up my statement… how do we even know they exist?

You not only fail to provide links to back up this claim, you fail to even mention the names of the systems. If you are going to insinuate that other peoples statements are not accurate because they didn’t spoonfeed you a link, then you should name the systems, and provide a link to support your claim. Otherwise, if you want to be treated like everyone else (which means we don’t accuse you of ‘misrepresentation’ because you didn’t provide a link) then don’t accuse others of possible “misrepresentation” because they didn’t provide the link you wish the provided.

(In the meantime, if you are at all serious about these satellite– could you name them? Then I can hunt down the data sources, if they are available on line.)

On the Moving along part: You have html wrappers, but you failed to include the link. So… I can’t read it. Your paragraph reads as if you are trying to find out how quickly we can discover january warmed compared to january. Is that right? That sounds interesting in an obscure way. Why do you wish to know this specific thing?

My interest is the whole year. Since IPCC makes projections averaged over the whole year, not month by month, I am interested in detecting changes for the whole year–which provides us more data. Also, if you check you’ll find your results iwll differ depending on whether you use the variability in GMST during periods with no stratospheric volcanic eruptions rather than those with stratospheric eruptions. The volcanic eruptions make a huge difference.

Finally, as for calling your prose blather and clintonian:
First there are more ways to load langauge than use of labels like “blather” and “clintonian”. You must certainly be awre of this. If you think your tone was level, you are mistaken.

Second, if you use the type of tenor you do in that article, and suggest that perfectly accurate statements are inaccurate simply because you a) don’t know and b) weren’t spoonfed a link, and if you try to make your case by complaining the word “few” was used (correctly) when you found a sentence that said “two”, I will indeed consider your prose clintonian blather. I will say it.

The trend over such a short period as 7.5 years will depend upon longterm forcings such as CO2, shortterm forcings such as weather, but also forcings acting over intermediate time spans (years) such as the solar cycle and ocean-atmosphere “oscillations” such as ENSO and possibly others. However, when you are trying to resolve the longterm trend over a short time period, as your headline implies that you are doing, while shortterm noise may be excluded, you certainly cannot ignore the intermediate terms since they may be biasing your result. And not doing so because the GCMs don’t do it is not a valid excuse, since in century long modeling the biases average out to be negligible (any bias causing variation is not predictable or modelable anyway, but that’s another matter).

The period you are looking at coincides with the complete drop of the last solar cycle from peak to rock bottom. The reduction in forcing which that results in is about 0.25 W/m2, and according to the findings of Camp and Tung, would cause a drop in the order of 0.15 degrees in the global mean surface temperature, see http://www.amath.washington.ed....._2007b.pdf

ENSO also represents a clear bias over the period you are considering as the ENSO index is dominantly positive, i.e. a warming el Nino, in the beginning, and dominantly negative, i.e. a cooling la Nina, at the end, see page 22 of http://www.cpc.ncep.noaa.gov/p.....ts-web.pdf What temperature drop that might translate to I do not know (there must be one), but looking at time-strength areas and comparing them with the 1997-1999 el Nino/la Nina transition, a temperature drop in your 7.5 year period of the same order as that of the solar cycle one or perhaps even more would not seem too unreasonable a guess.

The above two biases would then represent a total drop in temperature in the order of 0.3 degrees over your time period. Now there may be other intermediate trends not taken into account by me (I wouldn’t know how), but all in all it looks to me that your trend determination has actually _proven_and not rejected the IPCC expected longterm warming trend of 2 degrees per century. And of course, to really see who is right, it would be necessary to extend the period over which the trend is estimated so that known biases due to intermediate term forcings are eliminated.

Lucia,

The other satellite data series are the UW-UAH and UW-RSS series which are also available on the NOAA link: http://www.ncdc.noaa.gov/oa/cl.....h/msu.html

As you can see they are nothing more than “adjusted” versions of the RSS and UAH datasets designed to explain away the lack of tropospheric warming. There does not appear to be an lower tropospheric data which means you could not use them even if you wanted to.

The fact that PenguinDreams is ranting about “other satellite datasets” without understaning what they are demonstrates that he does not have a clue what he talking about.

Raven–
http://www.ncdc.noaa.gov/oa/cl.....h/msu.html“>NOAA says UW provides mid-troposphere data. I use lower troposphere. You can see the relatively levels in this chart from RSS

If penguindreams concern is the lower troposphere is too low, then the reason for not using two mid troposphere measurements ought to be obvious. Anyway, why would I use the UW mid-troposphere measurements when I don’t use the UAH or RSS mid-troposhere measurements?

I looked at the number in the file for UAH RSS from the NCDC link. Most (meaning nearly all) months the agree with the one at UAH’s site– except NCDC rounds. But, the most recent month disagrees.

I wonder why that July disagrees? The way the text reads, the two should be the same– they are from the same satellite, and analysized by the same group. (I’ll have to look a bit in future.)

Erik:
Sorry you got moderated.

The above two biases would then represent a total drop in temperature in the order of 0.3 degrees over your time period.

Both ENSO and the solar cycle have been examined several times here at this blog.

I discussed ENSO most recently in this post, http://rankexploits.com/musing.....l-falsify/
where I applied the correction Gavin discussed at Real Climate. When you read that post you will see that if I correct for ENSO using the value Gavin cited and illustrated in an RC blog post, the IPCC 2C/century falsifies with all instrument sets.

In contrast, if I don’t correct for ENSO, the 2C/century is not falsifying based on GISS, and using Red Corrected ordinary least squares.

The difficulty is that ENSO affects the trend, but also the scatter. The result is a 2C/century trend remains outside the uncertainty intervals consistent with the data.

Of course, many blog visitors mentioned ENSO long before Gavin discussed it’s effect on the trend since 1998, and I also looked at it here:
http://rankexploits.com/musing.....ne-orcutt/
and more qualitatively here:
http://rankexploits.com/musing.....t-the-pdo/

The solar cycle is a more difficult issue. It is not sufficient to explain the dip. However, if we treat the effect Camp and Tung suggest as being universally accepted (it is not), andbut it brings the trends into the lower regions of the uncertainty range. (You can read more here: http://rankexploits.com/musing.....ification/ )

The difficulties with doing this correction are many. While Camp and Tung are cited in the AR4, it is also clear that many climatologists disagree with their findings. In anycase, that variation appear difficult to tease out of the data. (Have you read the paper? There is a lot of fiddling going on to “find” the signal associated with variation due to the sun.)

In particular, you will find many many people (including bloggers) who are insistent that the effect of the solar cycle is negligible. Some blog visitors have hunted a bit for the “signal” in the few model ensembles but finding the signal elludes them.

If the dip is due to the sun, we will soon know. We are scheduled to come out of the minimum.

It appears that penguindreams is Robert Grumbine who a Physical Scientist at NCEP/EMC/Marine Modeling and Analysis Branch. Which may explain where he is coming from and his attitude. Too bad one one ever taught him to do anything but skim documents and act snotty.

I take back what I said. He is snarky enough for RC and OM.

Lucia says,
“However, if we treat the effect Camp and Tung suggest as being universally accepted (it is not)”
It is important to distinguish between empirical and theoretical/GCM solar analyses. C&T is an example of the empirical study that estimates the solar effect by looking for correlations in the data but does not offer any physical mechanism that would explain why the solar effect is that large. If the C&T study is correct then it would be evidence of a strong solar effect via an unknown mechanism that is not handled by the GCMs. This would not be good news for people arguing that CO2 sensitity must be what the models say it is because the models have taken everything into account.

Raven. I agree. And the difficulty is the models used to project don’t show that solar effect.

So, in terms of logic, it is odd to insist we must ignore what they actually predict when testing them. As far as we have been able to learn from looking at model output for GCMs that do account for the solar effect, accounts by modelers etc. according to the models the Camp&Tung solar effect is not there.

Of course, if I’m incorrect on that, and the solar effect is there, then I would be willing to give the models projection appropriate “credit” — by adjusting the projections of the ones that don’t account for the solar effect.

But it’s clearly ridiculous to give GC’s that do claim to account for the 11 year cycle an extra benefit by adjusting the empirical value downward. After all, their projections supposedly already took this into account.

It is also ridiculous to give models that don’t account for the solar effect “credit” for an effect the other models say isn’t there.

I honestly doubt the Camp and Tung effect really exists. I think it’s an artifact found by data mining. But, I could be wrong. If it is there, the temperatures will rise. But even if it’s there, the effect of the a perfect sinusoidal variation, with a start date set to maximize the ‘error’ due to neglecting the sun just isn’t all that big.

Chad tried to post comments on the “Penguindreams” post, which has some sort of comments glitch. He wrote:

Here’s what I was trying to post:

“… which you did not link. This robbed his readers of the opportunity to actually read the post he criticized.)”

There is a “link” to your post in Penguindreams’ post. I doubt you saw it as an incoming link because it’s not actually a hyperlink. Just a copy/pasted url.

Yes. And on my browser, it is an incomplete copy/paste url that cannot be clicked. Cutting and pasting results in a 404 error. Yesterday, it ended with “st”. Here’s how it looks today:

It may look fine on other browsers. But if someone wants to insist on the importance of links for determining accuracy, they should make sure their attempts at links are links.

Going to my original shows: Finally, in comments on my cherry-picking article, I was invited to take a look at http://rankexploits.com/musing.....l-rejected The reference is indeed there. You now tell me that there are browser settings that render the text improperly; that’s news to me. The text is there, and has been all along. It’s also present as a live link, as well as text, in Jonathan’s comment to which ‘cherry-picking’ is linked if someone wondered exactly what the invitation was. No, it wasn’t wrapped in html. Now that I know of the problem, I’ll look more at managing html links in my posting.

It’s ironic, though, that you say that your browser problem (perhaps shared by many, I dunno, it works on the browsers and OSs I use and nobody else has mentioned a problem) renders it impossible for your readers to follow (a search on ipcc-central-tendency-of-2ccentury brings up my blog, your blog, rabett run, and Jennifer Marohasy; and your blog’s name is clearly given in the text you show) — at the same time as you complain about me wanting to be ‘spoon fed’. IPCC reports are big things. If you want to criticize them, have at it. But let us know where it is you found the things you’re criticizing. If you’re making a serious criticism, it can only help to provide the gory details of what, exactly, you’re criticizing and where it came from.

w.r.t ‘moving along’, seems I have even more to learn about the blogosphere, wordpress, et al., than I thought. The link is http://moregrumbinescience.blo.....ate-2.html. The idea applies whether you feed it months or annual averages. If your concern is about annual averages, use those instead. The same sorts of answers come out as to time periods. Except what you show from data is monthly, not their annual average.

My concern as far as data sources goes is not affected by what conclusion you reach when using them. I phrased that badly. Regardless of whether it is the satellite that agrees best (disagrees least) with the surface air temperature trend you are testing, it isn’t (without more work than you’ve mentioned) the one that answers questions about surface air temperatures. I have no vested interest in the conclusions as such. I do care whether the test is good. A test of surface projections against surface data is better than testing surface projections against unadjusted/unexamined 13,000 foot average elevation average temperatures. I also didn’t say that you hadn’t looked at surface air temperatures, multiple misreadings by multiple people notwithstanding. You didn’t only look at surface air temperatures, I said (now with added emphasis). That’s true.

If you believe that the satellite’s nonlinear average centered at 13,000 feet reproduces the same thing as surface observations, by all means make that argument. Or show what needs to be done to the satellite observations to make them adequate proxies for the surface observations. Or cite some research that does so.

Many options here. Blindly treating them as exact proxies for surface observations, however, is definitely not comparing ‘like to like’, and is certainly not rigorous. As I did say, they’re ‘related’, but so is height and leg length. If you want to conclude about height, you have to do more than just take a measure of leg length. The fuller original is the first testing ideas note http://moregrumbinescience.blo.....eas-1.html. And yes, it is fair to say that I’m demanding about making tests. What is surprising is that with all the comments about how serious your tests are, you object to such standards.

I’d made two points regarding your using blogs as your source. One option was that you used sources that had pre-cherry picked the data. The other is that they’d selected data for their own purposes, which may not be adequate for what you’re doing. You now say that you relied on Wikipedia to make the selection for you! I’ve been getting told that you’re doing serious, rigorous, testing. Wikipedia’s purpose is not to substitute for reading the research literature when you want to criticize scientific research. At least they say not, and I agree.

Even at that, the Wikipedia article you cite says As a result, different groups that have analyzed the satellite data to calculate temperature trends have obtained a range of values. Among these groups are Remote Sensing Systems (RSS) and the University of Alabama in Huntsville (UAH). Clearly there are more than 2 groups. A few lines down, they say: “An alternative adjustment introduced by Fu et al. (2004)[5] finds trends (1979-2001) of +0.19 °C/decade when applied to the RSS data set.[6] A less regularly updated analysis is that of Vinnikov and Grody with +0.20°C per decade (1978–2004).[7]”

Your source says that there are more than two satellite analyses and mentions two of those additional analyses.

Early on, I was being told and seeing reference to, your post here being a ‘serious’ ‘rigorous’ effort. Yet many of the comments are how I shouldn’t be expecting what I would of a serious, rigorous, effort — instead, lack of sourcing, lack of explanation of why things were done one way vs. another, why certain data sets are selected and not others, why averages over atmospheric depth centering at 13,000 feet elevation are being used as if they were from the surface, etc. The most serious responses as to that matter of rigor were the exchange of laughter over whether $9.95 could be considered $10. If you’re being rigorous, no, it can’t. Whatever is involved, rigor isn’t it.

As an alternate hypothesis, then, how about it’s a bit of fun by a person knowledgeable about statistics but not climate or meteorological observing systems? The time series taken are whatever is handily available (check), and no concern is displayed about the vertical structure of the atmosphere (check) when comparing data vs. the surface, and the sources used aren’t read very carefully (as in not noticing that your own source as to how many methods are used shows that it’s more than 2) nor are serious research sources necessarily used (wikipedia), check.

Sorry to have interrupted your bit of fun.

If someone would like to learn about climate, a good observational start is The Physics of Climate by J. P. Peixoto and A. H. Oort, AIP, 1992.

Dr. Grumbine, my apologies for having some fun with with your blogged response, at the time I thought it’d be funny and maybe lighten the mood around here a little with all of this bickering going on. I’ve read a few of your papers and some of your other stuff/faqs, etc, and have respect for your breadth of knowledge and usually even toned writing. So, like I said before my lame joke, I was puzzled to read your response which seemed to me to be lacking in much real substance. I understand being a stickler for accuracy is always important, but this is a blog post, not a peer reviewed paper. Actually, you bring up a peer reviewed paper, Fu et al 2004, here’s a quote from it: (they’re referring to estimates from Ramaswamy et al 2001)
“Linearly extrapolating their trends of -0.84 K decade^-1 at 20 km and -0.49 K decade^-1 at 15 km with respect to height, we obtain
a trend of -0.27 K decade^-1 at 11.8 km (200 hPa).”
$10.00 and $9.95, now, due to this small, but unnecessary rounding, is it necessary to be dismissive?

I appreciate that you’ve taken the time to write a much more even toned response in the comments here (#5354), if you and Lucia were to bicker a little less over the trivialities, I think you could offer some good critique/suggestions to help Lucia improve her analysis. It’s been slowly evolving over time, and she’s never claimed it to be bulletproof and usually always seems open to try out suggestions. (ie, I notice one of your comments on your blog mentioning Tamino’s version, after he proposed his prefferred methodology, Lucia included it)

Ok, enough of my ramblings for now, we need a brotha like Mosher around here to moderate some, and to make better jokes.

w.r.t ‘moving along’, seems I have even more to learn about the blogosphere, wordpress, et al., than I thought. The link is http://moregrumbinescience.blo.....ate-2.html. The idea applies whether you feed it months or annual averages. If your concern is about annual averages, use those instead. The same sorts of answers come out as to time periods. Except what you show from data is monthly, not their annual average.

What idea applies to what? I assume you are referring to one of the many ideas contained in your onw post which you linked? You show a lot of figures with spaghetti graphs. You compute the average temperature by including data on either side of it. The value converges.
It appears you show reader that if one averages over larger numbers of the average will converge? Adding one new data point makes less of a difference? Of course this is true.

Other than that, what, precisely are you trying to explain? And what does this have to do with testing a trend using a standard method? The standard method takes into account the uncertainty in the data used. All other things being equal, if the uncertainty in an individual monthly measurement in larger, the uncertainty intervals for the calcuated trend are larger. That makes it harder to falsify things. All things being equal, if have a shorter string of data, the uncertainty intervals are larger. Your post appears to touch on the second point.

I assume you are trying to make some larger point. However, whatever it might be, the fact that you can show the uncertainty intervals decrease in size as compute over larger and larger time intervals is irrelevant to the observation that the IPCC projected trends fall outside the whopping big uncertainty intervals that we find based on seven years of data means they are falsified based on the tests I ran.

(There are caveats. The statitical model I use might not apply. But that’s a different argument from the one you are making. )

If you want to criticize them, have at it. But let us know where it is you found the things you’re criticizing.

Robert– You are aware this is a blog right? I’ve been posting a while. So, basically you are jumping into an ongoing conversations. It’s entirely possible for you to ask for clarification if you wish. And please tell me who you are referring by “us”. Other readers have no difficulty asking questions when they think they can’t locate what I am criticizing. This is particularly true of people who find themselves jumping into a conversation.

Now, back to your endless complaint about the satellite comparisons:

Regardless of whether it is the satellite that agrees best (disagrees least) with the surface air temperature trend you are testing, it isn’t (without more work than you’ve mentioned) the one that answers questions about surface air temperatures.

Robert. Let me repeat the points I made earlier. They are:
a) I tested the projections against land based data sets. These are GISS temp, HadCrut and NOAA.
b) You seem to think I didn’t. (At a minimum, you don’t acknowledge that I did, indeed do this and keep returning to some point about the satellite sets.)
c) Comparing the projections to GISStemp, HadCrut and NOAA answer the question about surface air temperatures.

So, the comparison you insist must be done was done in that post. Period.

As for your discussion about the difference between what the satellites obvserve and what the surface instruments observe:All the regular at this blog know this, and knew that before you dived in here.

It’s also irrelevant to the fact that the projections are compared to surface measurements in the post.

I’d made two points regarding your using blogs as your source.

Where have I used blogs as my source? Nothing in the article you criticized drew from other blogs. I did mention I include the satellitte measurements because they are discussed at blogs. This means that readers are interested in those results.

I do respond to reader interest. I suspect you will find yourself doing the same.

I invited you to name the other two sources of satellite observations you tell us exist. You did not do so. (As it turns out, you still have not done so– more later..)

You now say that you relied on Wikipedia to make the selection for you!

I didn’t say I used wikipedia. Nor did I say they made my selection for me. Also, it’s interesting you edit out the fact that I also mentioned the two I used are the only appropriate ones mentioned at NOAA– providing a link to that source..

But yes, I do read blogs, wikipedia, newspapers NOAA, GISS’s web site etc. to find sources that people wish to examine. As a result, my analysis generally include 5 data sets, where most other bloggers use only 1.

If you are aware of other sets, let me know. I will add them. This is not a problem for me.

On to the issue of the four satellite data sets
You keep insisting there are four satellite data sets.

After I provided you with a link to Wikipedia, you read that, came back and now tell me you found the two sources I do not use discussed at wikipedia.

However, it appears you came up with the number “four” by mistaking two journal articles cited by Wikipedia are “data sets”. Fu et al. (2004) is a journal article discussing an analysis. (The reference is: Fu, Qiang; et al. (2004). “Contribution of stratospheric cooling to satellite-inferred tropospheric temperature trends”. Nature 429 (6987): 55–58. doi:10.1038/nature02524 Had you clicked the link at Wikipedia, you would have discovered this. You may click it how: pdf. ) It was published in 2004. Clearly, this data cannot be used to compute trends from 2001-july 2008.

Vinnikov, Konstantin Y. (2006). “Temperature trends at the surface and in the troposphere”. Journal of Geophysical Research 111: D03106. doi:10.1029/2005JD006392. is also not an available data set. Feel free to click the link and read the article. pdf, also a journal article.

As far as I can tell, neither is an up to date readily available data set; they are journal article providing results of their analysis for a time period that predates the AR4 projections! As such, neither can be used to test trends from from 2001 -2008 for obvious reasons.

So, wikipedia does not mention more than two available data sets.

Wikipedia and NOAA each individually describe precisely two data sets for lower tropospheric temperatures: RSS and UAH.

The reasons blogs, newspaper articles, etc. use these and only these two data sets, is they are both highly respected data sets, they are updated, they are available. Maybe there are others sources that provide up to date observations for lower tropospheric data.

If you actually know of any, please mention them. But next time, make sure the sources you mention are regularly updated data sets, available to the general public. Please don’t mistake static journal articles published in 2004 or 2006 for sources of data that can be used to test trends from 2001-2008!

As for your final comments about not having discussed the difference between the lower troposphere and the surface, all I can say is this: Not every blog article discusses every possible thing in every single post. Even journal articles don’t discuss everything in a single article.

And yes, as for your complaints about lack of rigor: I describe this as a blog. It is not a journal article. Blogs are conversational media. If you don’t like that or think yours is something else, fine.

Oh, and do drop by and explain the atmospheric boundary layer sometime!

As always, good luck with your blog!

MikeN: I seldom use the ‘Dr.’ title and certainly not in blogs.

The $10 vs. $9.95 was helpful, hence my comment about most substantive. I’d been told that this was a serious, rigorous, source. And when I see comments about having falsified peer-reviewed research, I look for serious and rigorous work.

That’s not what we have here. It’s just a blog. It’s just a venue for some conversation. And so on for comments that have been made here. That’s fine too. So the next time someone says it’s more than that, say that work that was peer-reviewed is somehow falsified by this blog, I’ll just point them to the many comments here about it being unreasonable to hold it to serious standards.

The papers of mine you mention having read and liked (thanks), all received much more critical comments, including sometimes outright hostile, than my comments. Such is life if you’re going for serious. And none of my papers (so far have said that other people were wrong in their published research. I expect that when I do, the review process will be more challenging. (A snark in one of my papers, about model testing having not been rigorous enough to date, has resulted in some new work on measures to apply to those models. At least the authors can be read that way. I think it more likely that they’d planned on setting up the new measures anyhow, and my snark was a convenient way for them to note that they’re not the only ones who saw a problem in what had gone before.)

For the purpose of the ‘just blogging’, ‘just having a conversation’, it’s fine to ignore that the surface of the earth does not average 13,000 feet elevation. Or that the NCDC, UK, GISS data sets do include the ocean, not just land (notwithstanding Lucia’s frequent comment that they are land-based; See the graphic at http://hadobs.metoffice.com/hadcrut3/index.html for where the HADCRUT analysis has data, for instance. I’m not a geographer, but it does look like rather a lot of non-land is included.)

But, as this is a venue in which such differences constitute bickering about trivialities, it really isn’t one for me. (While that’s your phrase, rather than hers, her responses fit that description well.)

A recent comment, whose correction I accepted, was noting that 1824 was a better date than 1827 for Fourier’s discovery of the greenhouse effect. (Greenhouse misnomer article). Correction accepted, not dismissed in a wave about trivialities. (Does the greenhouse effect change nature if I got the publication date of the original wrong by 3 years?)

Over at my blog the next week or so I’ll be summarizing the simplest meaningful climate model series of posts, and then starting to look at the vertical structure of the atmosphere as we try to see why the surface is so much warmer than computed from that simplest model.

Robert:

For the purpose of the ‘just blogging’, ‘just having a conversation’, it’s fine to ignore that the surface of the earth does not average 13,000 feet elevation.

My tests do not ignore that the surface of the earth does not average 13,000 feet elevation. The precise comparison you insist should be made is made in the article you criticize. The projections are for the full surface– land and ocean. I compare projections too three surface based sources which, you correctly observe include ocean.

It is odd for you to claim you don’t want to bicker about trivialities. Are you really going to continue to avoid that point that the comparison in the article you criticize contains precisely the comparison you insist must be done. And now your reason is because I used “land” instead of surface in a sentence? Wow!

If your comments were anything other than trivialities of this sort or remotely correct, I would accept them. So far everything you say appears to be either a quibble over word choice, wrong or unsupported.

I realize you have convinced yourself your points are substantive, but would anyone else accept them?

Lucia, take it up with that person who recently wrote in #5357:
Robert. Let me repeat the points I made earlier. They are:
a) I tested the projections against land based data sets. These are GISS temp, HadCrut and NOAA.

So…In other words, yes. You are going to stick to quibbling and trivialities rather than substance. -W

Robert, please don’t leave us just yet. Please help us locate those satellite data sets that Lucia fails to include in her analysis…

If I may adapt the old joke:
Robert Grumbine made a number of criticisms that were substantive and correct. However, the ones that were substantive were not correct and the ones that were correct were not substantive.

On a more serious point, Dr. Grumbine is clearly a serious scientist with quite respectable academic credentials in this area. What then explains his “scattershot mess” (as Lucia described it) in both the posting on his own blog and his responses here? He has missed all the valid weaknesses in the Lucia’s statistical approach and focused on semantics – often while in error. Particularly weak are his cherry-picking comments including the identification of the mysterious additional two satellite record. (aka “…The curious case of Dr, Grumbine and the missing satellites…”). He also appears to rather foolishly presume that the climate v weather difference is not well known to the proprietor (and readers) of this blog even though it has been extensively debated here (including a very interesting exchange with Gavin Schmidt). And when challenged, he resorts to insignificant semantics and thinly veiled ad-hominems.

A friendly note to Dr. Grumbine – the skeptics / luke-warmers who inhabit blogs such as these are generally quite well versed in AGW theory. They are regular readers of Real Climate, Tamino and other “mainstream” sites (as well as Climate Audit, of course) and are also quite familiar with AR4. Most are not formally trained climatologists, but a significant number have advanced degrees in other math/statistical/scientific fields. Second tier arguments are not going to fly.

In fairness, blogs are not easy to navigate in a logical, thematic order so Dr. Grumbine is likely unaware of much of the background to the current discussion. My suggestion is for him (and other new visitors who would like to add comments) is to spend a little time reading some of the earlier discussions (including the exchange with Dr. Schmidt) that will give some necessary grounding to the conversation.

Hi Robert, thanks for the response. I think you might have missed what I was getting at in quoting the Fu et al paper with respect to the $9.95 and $10.00.
Notice that after they do their simple linear extrapolation, they round up a little to get their -0.27K, similar to 9.95 rounded up to 10. It’s small, but unnecessary. (and doesn’t it happen to very slightly increase their proposed effect?)

I guess in mentioning it earlier, I was hoping you might comment on whether or not this choice was rigorous; and whether or not it is substantial enough to be dismissive of their anaylsis, or just in the realm of small potatoes?

Anyway, as far as the satellite/surface thing, I understand the points you’ve made, but Lucia’s analysis is an array of tests, the satellites being used only on some of the tests. She’s baked a bunch of pizzas with different toppings, using four different cooking methods. Your complaint about using anchovies/dried squid(RSS/UAH lower) as a topping and calling her pizzas bad, doesn’t change the flavor of the pepperoni topped(HadCRUT) or jalapeno(GISS?) pizzas, or the flavor of the 3 topping merged surface pizzas.

Ok, sorry for the analogy, I usually don’t like them, but I’m on a diet and I’m hungry, so this one came to mind.

Ah. Different thing Mike. They started with only 2 digits precision in doing their extrapolation, the -0.84 and -0.49. They were then obligated to end with only 2 digits, at they did with the -0.27. They don’t get to create more precision than they started with.

As to the pizzas… well, proxy data can be useful, but they have to be treated as proxy, not direct. If one wants to understand how to make a good Chicago style pizza, you have to study Chicago style pizza. Study of New York style is informative about pizza in general, and one might understand Chicago style better for knowing what is different between it and New York. But tossing a some New York styles in with the Chicago style pizzas to your analyzer for Chicago style goodness, without adjustment for the differences due to them being New York … not so good.

Then if someone wants to tell me that somebody else is wrong about Chicago style pizzas because when the first person ate a New York style pizza they didn’t see the things the latter had to say about Chicago’s … not very persuasive.

Since what’s of interest here instead is just to throw pizza-like things into the hopper, the radiosonde network also gives figures at depths through the atmosphere (should be somewhere at NCDC), and you can get past and current global meteorological analyses from http://nomads6.ncdc.noaa.gov/ for surface and standard pressure levels.

Lucia: If the difference between the globe and land is trivial to you, then it’s quite likely that very little I’ve ever written or will write will be nontrivial in your eyes. See also your #5319.

It is perhaps an attempt to loosen up, but I am not convinced the communication becomes more effective as the pastry metaphors at thrown around.

Could the participants kindly consider to return to a less allegorical language?

Cassanders
Ek writu i RunoR….

Robert–
The difference between globe and land is not trivial. That’s why I compare projections for the full surface of globe to projections for the full surface of the globe.

What you are doing, which is pretending a mistaken word in a comment modifies the comparison or it’s meaning is simply bizarre.

Thanks Lucia for giving me the opportunity to meet yet another (choose your word) .
Already a name like “Penguindream” suggests pink elephants and dancing fairies .
But then the prose of the said Penguin , what an unsubstantial , wordy , boring mess .
It takes so much space and bandwidth for so little content that it is really a waste for people who browse here for the data and statistical analysis .
I would wish that the Penguin stops posting or posts only somewhere in the wilderness for other Penguins .

In any case as far as I am concerned , I know that when you write that you make an average of land based temperatures , you actually make an approximation of a space integral over the whole surface of the Earth and an approximation of a time integral over 1 year .
Of course supposing ergodicity
I wouldn’t waste my and your time by pointing out that “average” , “land” and “based” words should have explanations attached to them .
Yet as English is actually not my natural language , I probably should have
I suspect that our Penguin , very much alike M.Mann , doesn’t understand much about statistics or english for that matter .

On a more serious level .
In the frame of your research , I suggest you to read (if you have not yet read it) : http://adsabs.harvard.edu/abs/2006JHyd..324..239K
It is an excellent paper fully relevant to the question whether classical statistical methods (f.ex AR1 residuals) apply to the selfcorrelated scaling processes like climate or hydrology .
I know that this discussion has not yet taken place here but it will have to come one day or the other .

The problem Dr. Grumbine has with your analysis Lucia is that you are asking the question in the first place.

You and I and everyone are not allowed to double-check or verify the central tenets of global warming theory. You and I and everyone are just supposed to believe.

Global warming believers are so invested in the accepting the theory, that they do not a verification process or “scientific proof process” to be undertaken (since there is a chance the theory will fail those tests.) That is a risk they do not want to face.

This has happened before in the history of science so it isn’t reallu unsurprising. Human beings have a very strong inate need to be “right” about things. I suppose that is what we are seeing.

But scientific progress occurred throughout history when the scientific process was followed and the human need to be right was suppressed in favour of the actual facts. Theory-experiment-observe-measure data objectively with statistics-accept or revise theory-reproduce-reproduce again-scientific fact (to be revised again when new theory and data is produced.)

Lucia,

Your discussion with Robert Grumbine and review of his site has gotten me thinking. This usually leads to some “dumb” questions that help me organize my thoughts better.

question: Can we use this short term trend to test the IPCC projection? Now, I have no problem with the length of the trend itself, but am rather drawn to the question if this particular trend can be used. It seems to me that as long as the short trend is consistant with the greater trend from the observations, that it should be ok to use. But if for some reason the trend isn’t consistant with the larger set of observations it shouldn’t be used.

Is that something which can be considered to determine the validity of your test? I mean, if the trend isn’t consistant with the larger data set, what value does it have when testing the constistancy of a model?

Raphael–
The issue is the uncertainty intervals. The shorter the trend, the larger the uncertainty intervals.

We can come up with analogs outside climate. Suppose you want to determine the average height of 18 year old Brazilian male.

Can you estimate it based by selecting 10 Brazilian 18 year old men at random and measuring them? Sure. But your uncertainty is likely to be fairly large. I don’t know what the average height of 18 year old Brazilian men is, but maybe you’d measure and figure out it’s 5′ 9″ ±2 inches. Meaning, you think the height is about 5′ 9″, but your uncertainty intervals is 2 inches.

Can you get a more precise estimate by measuring 100 18 year old Brazilian men? Sure. Maybe after measuring 100 men, you’d discover the height was 5′ 8″ ±0.7″. So, your first estimate was high — but so was your uncertainty.

Now, as for testing a hypothesis: If someone has told you the average height of a Brazillian man was 6′ (6 feet) , would you have to wait until you’d measured 100 men? Or would you already know when you’d measured 10? Obviously, you’d be able to rule out 6′ tall quite quickly– because it’s way off. On the other hand, if someone had a theory they were 5′ 10″, you would need to take a lot of measurements before you could say they were off by 2″.

Also, if, for some reason, you needed to know the height to within 0.5″, you’d need to take more than 100 measurements!

Now back to climate:
The reason why many firm claims that we need 30 years make no sense, is that statement that we “need” 30 years isn’t connected to any particular idea of how precisely we need to know something. Obviously, we don’t need 30 years of data to figure out that the average temperature in Chicago during January is colder than the average temperature in El Salvador in January.

Even if we had no knowledge of physics, and no previous data because we were 7 years old when we moved to Chicago from El Salvador with our family, we would know very quickly! (Guess who moved to Chicago at the age of 7. )

I didn’t have a problem the sample size in question.

To expand on your example:

“Meta knowledge” says the average height is 5′ 8″ ±0.7″
The claim says the average height is 6′
When we go to take our random sample, we decide to take the next 10 brazillian men that come through the entrance at a dance club. That should give us a random selection.

However, the next 10 brazillian men through the door were Brazillian little people. If their heights fall outside the 95% range of brazillian men 5′ 8″ ±0.7″, should they be used to test the claim that the average height is 6′? Or perhaps I should ask if the sample rejects the “meta knowledge” in the same test we would use to test the “claim”, does the test have value?

Raphael–
Ahhh! Ok. Yes. There are a bunch of questions– and some do relate to sample size:

1) Was your sample randomly selected over all.
2) Are adjacent samples independent even if the batch over all was independent?

So with the Brazilian men, maybe someone picked the individuals randomely, but you drive from town to town, and measure everyone in one region first. Maybe as a result you got all the short ones, and the tall ones were elsewhere.

So, this is a problem.

With weather data, we obviously get todays weather, tomorrows, the next days etc. We also know weather patterns persist, so there is correlation. That’s a problem!

In principle, we correct that by using the “Red Noise” correct. But it’s not entirely clear that’s enough. So, you see I then added a second method that assumes weather is red noise, and then assumed measurements are white noise.

So, is this enough? We can’t know based on the data alone! In fact, we would need centuries of data to know whether the red noise correction is enough or too much!

But what I have done is look at data during the previous volcano free period and it’s properties are close to “now”, so that suggests “now” might not be a batch of “Brazilian short people”. (Still, the previous period is short. So, it’s not a sure thing!)

So, then we can try to turn to the models to see what they tell us. Of course, the problem is the models might not get the “weather noise” right. So, we have a problem.

I’m currently looking at the models. It’s pretty clear that, at least collectively they don’t get the weather noise right. Or at least, they don’t get sufficiently “right” to improve on our lack of knowledge based on the data. They all disagree with each other in many important ways. (And it’s not just a tiny disagreement.)

So, that’s a long way to say: We don’t entirely know. I think the best we can do is use the properties of the data we have.

Wow, I actually managed to follow all of that.

But, I don’t think that was where my thoughts were headed.

Um… Next dumb question.

Would there be any value in using a non-skilled projection, (say a calculated trend for the start of the dataset to 2000) and using the same test against that? Or using the same test against the trend for the complete dataset?

Rafael–
I’m not sure precisely what you mean. The reason I test 2C/century, is that projection is for the current time period. We don’t expect 2C/century in, say, 1950-1960. So, there is no point is testing for that, back t hen.

Can we calculate a trend from 1880-2000? Sure. It’s done all the time. And, I could tell you uncertainty intervals, based on assumptions about the type of correlation in the data.

But, it doesn’t help much for the shorter period primarily because if we were to look at properties of the variability, it’s obvious the volcano’s make a difference. And, we haven’t had any!

That volcano eruption affects the properties of the deviations from the trend, and that matters in the calculation!

Sorry, when my thoughts aren’t well organized, I have trouble getting the point across. When my thoughts are well organized, I usually don’t need to ask the question.

Maybe the best way to handle this is to clear my mind and type out the line of thought I am taking. With any luck it’ll make sense. :-/

*takes a deep breath and clears his mind*

Set. Sub-set. Model.

From the previous example on the height of brazillian men.

Set: All braziliian men: 5′ 8″ ±0.7″
Sub-set: Brazillian little people surveyed in a dance club.
Model: Say I create a model which estimates height of brazillian men based on genetics and mating habits.

Using the subset you reject my model.
I say that the subset was not an accurate representation of the set.
You say you have sampled other dance clubs and while on the low side of samples of 10 brazillian men who enter dance clubs, the sample used is consistant with the other samples.

Maybe brazllian men who go to dance clubs are shorter than the average brazillian?

Which really doesn’t make a whole heck of a lot of sense to me. As it would imply that vocanic periods have a net effect of causing the temperatures to be higher than they would be without them. :/

Which leads me to conclude that the brazillian men who go do dance clubs are not shorter than average. (Unless of course volcanoes really do cause temeperatures to increase more than they would without them)

Raphael–
The problem with the volcanos eruptions is the externally applied forcing causes the temperatures to drop then rise. This increases variability compared to when there is no volcano eruptions. In a statistical calculation this cause our estimate of the uncertainty (the bit in the ±1C/century ) to rise.
But using periods with volcanic eruptions to estimate variability due to the non-linearities in the unforced weather system is a bit like comparing variability in surface height in a pool driven by a wave machine to that in a pool not driven by a wave machine.

On the dance club: Yes. If there was a dispute over sampling in a dance club, we would best resolve that by not sampling in a dance club. That’s an issue of bias, and we do need to try our best to avoid that. (If we can’t avoid it, we do get to argue about it. It’s valid to investigate possible biases, estimate their likely magnitude etc.)

Lucia,

Thanks for the patience with the dumb questions. You said our current sample of brazillian men was on the low side of samples of brazillian men entering clubs. errr… I mean you said that the current temperature trend was on the low side of other samples from non-volcanic periods.

Would adjusting for solar account for it being on the lower side of the “non-volcanic norm”?

Raphael–
No. I’m not saying the current trend is on the low side from other samples of non-volcanic periods. (Or didn’t mean to if I did.)

The current non-volcanic period has roughly the same variability monthly temperatures relative to an trend fit as other non-volcanic periods. In contrast, volcanic periods have must larger variabilities. Also, if you calculate all 8 year trends during periods with no volcanic eruptions, those are larger than during periods with volcanic eruptions.

So, basically, volcanoes introduce “noise”. (I guess if we were going for the height analogy, a country with more immigrants fro different places — some tall, some short–might have more “height noise”.)

But, they don’t affect the long term trend much. (The do affect the short term trend a lot– that means when there are eruptions it’s harder to estimate the long term trend using short term data.

The magnitude of this variability affects of ability to estimate the mean trend.

I think I missed my turn off, ’cause I am lost.

N ( x , y )

My question was is our current sample consistant with past samples.
You clarified that our sample would need to be compared with other non-volcanic samples. You state that y for our sample is “roughly the same” as other non-volcanic samples. What about x? Does it matter?

Raphael–
It depends what “x” is. According to the models and the various theories underlying the models the temperature trend in the 30s is not supposed to match the trend now.

I think in your question “x” is the trend.

The original question is can we estimate a trend (x) using smaller time samples: The answer is yes. However, there are larger uncertainty intervals for smaller time samples.

The argument that we can’t use smaller time samples is founded on a number of things. One of them is the variability of “weather noise” is very large, and so our confidence intervals are so large we can’t get decent estimates. I think you are calling the “variability”, “y”. Those are similar during the two periods– and the dispute is fundamentally over y.

So, we can estimate “x”. We can distinguish “x” from some incorrect theories about “x”. The argument that we can’t do it is founded (in part) on the claim “y” is larger than it is. But, evidence suggests “y”– based on current weather– is the same value as “y” in other eras with no volcanic eruptions.

This is all getting very confusing.

Lucia,

If you think it is starting get confusing, you should try being in my head.

Ok, when I re-read, “the temperature trend from the 30′s is not supposed to match the trend now,” I thought about what I would ask my son if he said the same thing (to try to stimulate his grey matter). “Really? Why not? How should it differ? Does it differ in the way expected?” I think that last question is the one I am having.

Rafael–
The surface temperatures are supposed to be rising because of the increase in GHGs. That’s the theory.

Lucia,

Sorry, sometimes I confuse myself. Someplace along the way I got to thinking the theory says that temperatures go up if the sensitivity to GHG’s, after feedbacks, is greater than 0.

Robert (#5375), good point, I shouldn’t have used “unnecessary” wrt the Fu et al round-up, I was straining to look for a similarity to use for my nitpick caricature.

As far as the pizzas, I guess I wasn’t very good with my analogy judging from your response. In my version, when I said, “four different cooking methods” I was referring to her statistical models, which she doesn’t mix. I had thought that your earlier qualm was that in some of the tests, she’s mixing the toppings(GMST estimates/proxies). In your version of the pizza analogy, if I’m reading it right, she’s mixing the cooking methods (New York/Chicago) and I think my “toppings” transformed into your “cooking methods”. If that’s the case, the point I was trying to illustrate with my analogy (but now converted into the terms of your analogy) is that if you don’t like the New York/Chicago style mixed pizzas, that’s ok, but it doesn’t change the tests of the isolated New York style (only) pizzas.

Sorry, I promise not to use any more analogies for awhile..

You mentioned that 7 years perhaps isn’t a statistically significant interval, but considering that AGW theory was based on about only twice that number when James Hansen made his appeal to congress, I’d say it’s pretty close.

What would be a statistically significant interval, assuming this trend stays the same or only slightly warms/cools for x amount of years in regards to severely weakening or disproving AGW – my thoughts are that it already has begun weakning the argument and we might start seeing friendly neighborhood GISS henchmen lighting fires near strategic temp sites if things don’t pick up soon.

Mike: Analogies are dangerous indeed. The statistical methods are tools, nothing more, nothing less. Back to analogy land, they’re food processors. They have settings, and you can throw stuff in to them. If you throw in good ingredients and use good settings, you can make a wonderful dish. If you use a poor bunch of ingredients and poor settings, what comes out doesn’t become good by virtue of them being a spiffy set of processors. My argument is not about whether the processors were spiffy or not, but about the ingredients and settings (in particular, the use of the ‘ignore that 7 years is unlikely to represent climate’ button).
Less analogized, see below in response to Eric.

General: I’ve opened up a ‘question place’, http://moregrumbinescience.blo.....ace-2.html for science questions. See also my current draft of comment policy at http://moregrumbinescience.blo.....olicy.html

Eric: I see folks on blogs refer to AGW theory, but it isn’t one I know from scientific sources. Could you, or someone else, tell me what it is that theory is supposed to be. (Preferably also what scientific source you saw it in.)

A major point, which I see I didn’t write explicitly in either of my testing ideas notes http://moregrumbinescience.blo.....eas-1.html and http://moregrumbinescience.blo.....eas-2.html is that my basic objections are physical, not statistical.

The physical issue is ‘what is climate’, which I started giving takes on in http://moregrumbinescience.blo.....imate.html and http://moregrumbinescience.blo.....imate.html. Though statistics are apparent, what’s important to me is the physical side — climate is the part of the system which is slowly varying, weather is the part that is rapidly varying. Unfortunately, we only ever directly observe weather — instantaneous measures of what’s happening.

If what you want to study is weather, you’re set, just take the observations and have at it. (Which still leaves you a very complex problem!) If you want to study climate, you have to do more work than that. So say I, Lucia disagrees. What I, and (more importantly) the people who do the observational work on climate think is that you have to disentangle the two. If you’re interested in trends in surface temperature, for instance, you need enough data so that the magnitude of your trend doesn’t depend sensitively on how long your record is (among other things). If it does, you’re looking at weather (still).

How long is ‘long enough’ … well, that’s what I was addressing in the two posts. It’s incomplete and not rigorous, but it is, at least, a start. The conventional answer in the field is ’30 years’, which goes back to at least the early 1900s. I’ve taken a side project to see if I can find the origins of that conventional 30 year period. That it’s stood to use for a century, though, suggests a pretty good value.

But let’s make another tour of the trend issue (I was looking at average rather than trend, but several others have looked specifically to trend, including stoat, tamino, and realclimate). Yesterday the high in my area was 10 F below normal. Four days earlier it was 10 F above normal. Clearly the climate will drop below absolute zero in another few months (5 F/day, ca. 500 F to drop, 3-4 months). Except that those weren’t climate observations, they were weather observations. The former was while a front had us in the warm sector, the latter observation was while Hanna was pulling colder air down. If you want a trend, you need a long enough period that the timing of fronts and storms doesn’t change your answer. Today will probably be close to normal, dropping the trend from 5 F/day to 2 F/day. A one day change in record length can’t have that kind of effect on the trend of climate.

So now you (a figurative you) decide that you’ll avoid that sort of problem by using global averages (if the front is making me warm, it’s making someone else cool, so it averages out) and, say, average a month at a time. But look up at the plot of monthly temperature anomalies. They change by several tenths of a degree C from month to month. Your trends for ‘climate’ are then changing by plus or minus 500 C per century (century being Lucia’s preferred time unit), from month to month. This is still, clearly (to me and folks in the field) ‘weather’. And, clearly, you’d need quite a lot of months to get that sort of noise to average out, even more so when you’re arguing about signals of only about 2 C/century.

So then you try, say, about a decade, as Spencer and Christy did in their first major MSU publication. And you get burned by the fact that the record starts near a major El Nino (warm event) and ends near a major cool event (Pinatubo). Had they used the standard 30 year period, this wouldn’t have been a major problem. On the other hand, it would also mean that they’d only now be publishing their first paper. If you’re going to violate the standard, you have to be very careful about all those sorts of things that can give you false results, including: volcanoes, solar variation, ENSO, AAO, AO, MJO, QBO, PNA, AACW, PDO, … plus how all of them affect the different observing and analysis systems. (Not least, you should know all the things I left off the list — that list is far from complete, just things I happen to know of while working in areas not terribly concerned with most of them.) If you then want to compare a trend from your analysis of a short record of temperatures with a trend from a short record of model output, you also have to become expert enough about the models to know whether they can honestly be expected to model those various things, and if not, whether and how that affects their ability both to get the trend for your short period right, and how it affects their ability to get the long period changes right.

Of course that ‘have to’ is only one in the sense of ‘in order to be rigorous by my notions’. You’ll see from the testing ideas 1 note (an example that has nothing much to do with here) that this can be rather demanding. In contrast with rigor, I’ve been reminded that this is just a blog. For ‘just a blog’ purposes, the fact that my back yard was well above the models’ projections for last Wednesday, or that it was well below yesterday, or that the cooling trend was so enormous between the two, are all ‘sufficient’ to lead people to disregard the models. I’ve been told this a number of times over the years (the other people using their own back yards for the ‘disproof’).

So what does all this say about whether climate models can be tested? Not much. There are a lot of aspects to climate models — global mean surface air temperature is far, far from the only variable present. There are temperatures for the ground, sea surface (vs. atmosphere), pressure levels through the troposphere and stratosphere, elevations of the pressure levels (relates to temperature in a vertically averaged sense), phenomena like ENSO that the models should be able to treat with some degree of skill, and on for quite a long list. The different variables have different observing problems associated (all observations have problems!), different variances, different expectable degrees of skill from the models, etc., etc. While global mean surface air temperature is not a good variable for making a test with only 7 years of data, there are rafts of other candidates to look at. So look at them, and verify that 7 years (or whatever) is sufficient to make a good statement about the climate with. Then have at it.

Robert–
The observation that this is a blog was in response to your complaint that people might have difficulty finding the IPCC AR4 because I didn’t provide a link. Because this is a blog, people who have difficulty using Google to find the documents, can ask for links in comments.

[...] IPCC Central Tendency of 2C/century: Still rejected [...]

Lucia,

Not to pick nits, but you did say, “And yes, as for your complaints about lack of rigor: I describe this as a blog. It is not a journal article. Blogs are conversational media. If you don’t like that or think yours is something else, fine. This seems to suggest you had used the observation “this is a blog” in response to more than just linking issue. That being said, it also suggests only lacking the rigor of a journal article and says nothing about lacking rigor, which is what Robert is inferring.

Raphael–
Fair enough. Though, the reason I referred to conversational media is Robert is complaining about not providing links, not explaining every single possible choice inside the “box” of one blog post.

Blogs are conversational media involving the author and mostly regular readers. They provide comments blogs. For that reason, the posts may not be “complete” in the sense that stand alone journal articles may be, and some choices discussed in past blog posts may not be discussed in the current one.

Oddly, book chapters do the same things. They often assume the reader has read previous chapters to understand the current one, and don’t necessarily tell “link back” to the precise page where a concept was first introduced.

In my opinion, the conversational style of blogs has little to do with the rigor of analysis. That’s basically what I mean when I respond with “this is a blog” to those sorts of complaints.

[...] I compare IPCC projections for global mean surface temperature (GMST) to observations, someone suggests I do the comparison a different way. Today, I’ll do the comparison yet [...]

Science Dude—my link is from here:

http://rankexploits.com/musing.....-rejected/

Basically and analysis using GISTemp and UAH to show the IPCC’s forecast for temp increase to be falsified

[...] more data. We can see if, over time the results of this method line up with the results of the “Red-Weather noise + White measurement noise” method discussed in July (which by the way, I still haven’t explained how to implement, and [...]

[...] IPCC Central Tendency of 2C/century: Still rejected [...]