|
Post by steve on Sept 18, 2011 14:00:23 GMT
magellan,
As usual you have no idea of what I think about Dessler 2010 because all you ever do is assume what I think rather than read what I think. If you actually read what I have pointed out, Dessler 2010 says that there is a very unconvincing positive feedback in the data. It's right in front of me now - I'll quote it again for you:
so you can only get excited about "smackdowns" of Dessler 2010 if you ignore what Dessler 2010 makes very clear. It's not at all surprising to me that different choices of which data to analyse will give you different results because the data is poor and covers only a small period of time.
Roy appears, on the other hand, to have so much faith in the observations that he neglects the error analysis, so emphasising the apparent disagreement between his cherry-picked choice of models and the data.
|
|
|
Post by steve on Sept 18, 2011 14:08:03 GMT
Co-author Manabe is one of the pioneers of climate modelling. So this just destroys one of the regular strawman claims that climate scientists are not interested in challenging their models with data.
|
|
|
Post by steve on Sept 18, 2011 14:16:34 GMT
Icefisher
All Dessler has done is extend Spencer's analysis and point out Spencer's cherry picking. This has identifed a flaw in Spencer's theory. No divining rods are needed because the flaw in Spencer's theory means that there is *still* no good evidence that ENSO is an important factor in the last 50 years' warming.
I have a coin-tossing model that tosses a coin 100 times. The number of heads will be 50+/-14%. My sceptic friend Rod Spender has tossed a coin ten times and only got 3 heads (40% less than half) and has used that evidence to undermine my model. Does that make Roy Spencer's error more clear?
|
|
|
Post by sigurdur on Sept 18, 2011 14:39:55 GMT
Steve: Actually, what Dr. Spencer and Dr. Dessler have shown is that models do not match observations. And they both show this.
What I feel is most important in Spencers paper is that the observations were not challenged.
We all know the models are doing a poor job. What is bad about this is that there are still those who defend the models as if they show something. What they do show is that they are wrong and need to be re-examined to find out where they screwed up.
As far as ENSO being a climate driver, the jury is still out on this. Regional climate can, it has been shown in the past, drive global climate over time. Whether ENSO provides that final push to a new phase of climatic norm is not established, but also it has not been ruled out with any degree of certainty at all.
Both Dessler and Spencer are at least looking at something. Personal bias clouds both of their papers, which is to be expected.
The lack of warming since 1998 shows that the ocean lost a lot of heat during that El Nino which has not been replaced. ARGO data since 2003 seems to show that in fact the OHC has been flat with a negative bias.
There is no question that we are in a cooling trend. How long this cooling trend will last is unknown. One thing for sure is that we sure as heck are not warming anymore.
Has the earth reached a new "balance" per se? I don't know. The models show that we haven't, while reality points to that we have direction.
I find the models intersting for discussion, as I enjoy learning. I also find the models very poor at any credible predictive value. WG1 agrees with my assesment of this. The level of certainty projected by folks who have the AGW agenda front and center is not the reality of climate science at this time.
I think that the folks who hammer this issue are doing climate science a disfavor. Their scientific pursuits have been overshaddowed by their political beliefs, and also their funding requests. There have been a lot of mouths fed with AGW research funding, and a lot of resources squandered in this.
|
|
|
Post by trbixler on Sept 18, 2011 16:29:02 GMT
Co-author Manabe is one of the pioneers of climate modelling. So this just destroys one of the regular strawman claims that climate scientists are not interested in challenging their models with data. And Manabe found them wanting just as Spencer found.
|
|
|
Post by steve on Sept 18, 2011 17:48:25 GMT
Of course they are "wanting". The meme that Spencer has got you to swallow, though, is that he is removing the veil from the eyes, calling the emperor's new clothes etc. etc. I'm afraid it is nonsense, so you pointing it out is uninteresting.
Note that the paper you have linked to gives no comfort to those hoping for lower amounts of warming at the surface, since it is primarily about the amount of tropical mid-troposphere warming. And it is still possible that the observations are wrong - this is just one paper, and they don't rule out that possibility.
|
|
|
Post by steve on Sept 18, 2011 17:53:47 GMT
No you are wrong. If you look at Dessler's paper he includes uncertainty - that *is* challenging the observations. When you include uncertainty, the models with ENSO signals show a fit. The fit is never going to be perfect because even two identical planets will show a different response because of variability.
Nope. Wrong again. Oceans have gained a lot of heat since 1998. Evidence from the XBTs and from sea level rise.
And no we are not in a cooling trend. 7 years of *possible* but highly insignificant cooling oceans does not make a cooling trend.
|
|
|
Post by sigurdur on Sept 18, 2011 18:02:14 GMT
Steve: We all know that the XBT data was suspect to say the least. And as far as sea level rise, it is not above longggggggggg term averages at all.
I will state emphatically that we are in a cooling trend. You don't want to admit it yet, but wait a few more years and you will have to admit it.
What you miss about the ENSO models that show a fit to Dessler is that they don't show a fit to anything else, and if you want to investigate their hindcast ability.....they don't have any. So that does not provide any discernable caveat of credibility to Dessler.
|
|
|
Post by sigurdur on Sept 18, 2011 18:04:06 GMT
I do think Spencer/Dessler have hit on something important. The thing is......neither of them can provide statiscally valid reasoning to prove it.
I think the uproar over Spencers paper in the AGW community does show that the importance of Spencers potential findings are more on spot than Dessler.
|
|
|
Post by trbixler on Sept 18, 2011 18:24:57 GMT
Check the slope. "Doing the same regression with 4-month lagged relationships (which both Dessler and SB agree to be more significant than the instantaneous relationship), the sign of the slope is reversed. Whereas Dessler 2010 had reported a slope of 0.54 +- 0.72 (2σ) W/m2/K, the regression with lagged variables is -0.90 +- 0.95 w/m2/K and has better diagnostics. [Update Sep 8 – Nick Stokes observes that this reversal of sign may be a phase phenomenon. This is something that needs to be examined as I haven’t handled this data before. However, please note that a sign reversal also results on alternative grounds merely from using CERES clear sky data instead of ERA clear sky data, the latter being used in Dessler 2010 without an explanation for the variation. See here.)" climateaudit.org/2011/09/06/the-stone-in-trenberths-shoe/
|
|
|
Post by icefisher on Sept 18, 2011 18:34:18 GMT
All Dessler has done is extend Spencer's analysis and point out Spencer's cherry picking. This has identifed a flaw in Spencer's theory. No divining rods are needed because the flaw in Spencer's theory means that there is *still* no good evidence that ENSO is an important factor in the last 50 years' warming.
What both Dessler and Spencer show is that after 16 years of statitistically insignificant warming that the models do not correctly model the physics of the atmosphere sufficiently to lend a shred of credibility to the models. You will need to explain where any certainty that the models are right comes from.
I have a coin-tossing model that tosses a coin 100 times. The number of heads will be 50+/-14%. My sceptic friend Rod Spender has tossed a coin ten times and only got 3 heads (40% less than half) and has used that evidence to undermine my model. Does that make Roy Spencer's error more clear?
Models are parameterized Steve. There is zero relationship between the models tossing the coin and Spencer's observations. What it shows is you run the model for 10 years and it comes up wrong because nature is tossing that coin not Spencer.
All Spencer is doing is applying model assumptions to nature's response and demonstrating that the models are not doing a good job. Thats true! The only thing you are complaining about is he quantified his findings and you don't want to believe the results. Dessler indeed confirmed the quantification.
Right now where we are on the road to flight is still well before 1903, every plane has crashed without staying airborn.
Some bright brothers need to come along and find the right climate controls.
You are just arguing the controls are in place and a rare cross wind just happened to come up and foul up the flights of otherwise worthy designs.
In fact Trenberth is investing all in a campaign to ensure no new designs enter the fray until his design has a chance to prove itself.
Fortunately, such efforts in 1903 were unsuccessful. I have seen this kind of behavior played out 100's of times. Most of the time how far off the establishment's estimate is is underestimated. Its no different than the recent mortgage crisis. The establishment always argues all they need are a few trillion more dollars and the problem goes away.
|
|
|
Post by steve on Sept 19, 2011 17:34:42 GMT
Icefisher,
My coin-tossing model is also parametrized - I don't model the dynamics of coins flying through the air! But that's not the point. The "relationship" between mine and Spencer's examples is the differing timescales of the statistical analysis. Averaging over longer periods will give you different statistical results depending on the situation. It is easy to do it right (eg. turning a century of model data into 10 different decades and looking at the spread, perhaps). Spencer did it wrong (this was his number 1 obvious error but only number 3 in terms of importance to the result).
|
|
|
Post by sigurdur on Sept 19, 2011 17:37:26 GMT
Icefisher, My coin-tossing model is also parametrized - I don't model the dynamics of coins flying through the air! But that's not the point. The "relationship" between mine and Spencer's examples is the differing timescales of the statistical analysis. Averaging over longer periods will give you different statistical results depending on the situation. It is easy to do it right (eg. turning a century of model data into 10 different decades and looking at the spread, perhaps). Spencer did it wrong (this was his number 1 obvious error but only number 3 in terms of importance to the result). Steve: Spencer "may" have made an error there, but Dessler does no better with his analysis. My point has been that neither Spencer NOR Dessler has proved a point, but at least they are looking at it. And that is important!
|
|
|
Post by steve on Sept 19, 2011 17:38:09 GMT
sigurdur,
So you have vague memories that some XBT data was suspect and thereby reject all its results. But you can't reject the sea level data that validates the XBT data so you simply compare it to sea levels of 125000 years or more ago.
In reality, the XBT data (after correction for the identifiable biases) and the recent sea level data are consistent enough to disprove your claim that 1998 was the peak in ocean heat content.
|
|
|
Post by steve on Sept 19, 2011 17:43:15 GMT
There's nothing particularly wrong with Dessler's paper - he shows that cloud feedbacks are hard to measure on 10 years of data, and in any case their influence is small compared with variability due to SST changes. His analysis could be repeated with different choices of dataset and appear to show the same thing.
Spencer has proven he can cherry-pick models and create a splash in the internetosphere.
|
|