104 replies to this topic

[Little Fish]

Br Cornelius, on 08 December 2012 - 11:09 AM, said:

Undoubtedly the rate of warming over the last 15 years has slowed, but it certainly cannot be used as evidence that the theory of AGW is invalidated.

and that has been the problem with this discussion, you and doug have misrepresented the point being made, perhaps because you just regurgitate talking points from SKS.

the last 15 years of data show a discrepancy in the expected warming rate produced by gcm models. this is according to NOAA and the top climate modelers.
http://www.unexplain...5

"Near-zero and even negative trends are common for intervals of a decade or less in the simulations, due to the model’s internal climate variability. The simulations rule out (at the 95% level) zero trends for intervals of 15 yr or more, suggesting that an observed absence of warming of this duration is needed to create a discrepancy with the expected present-day warming rate"
http://www1.ncdc.noa...2008-lo-rez.pdf

Edited by Little Fish, 08 December 2012 - 11:59 AM.

[Br Cornelius]

Models are not reality - they are guides to the future. They are imperfect and they were never intended to track real weather on a year by year basis. They are guides to trends.

It is you Little Fish who are placing to much importance on a model. Models are what they are - reality is what we live with.

The fact that you are basing your observation on a skewing outlier (which by the way is the warmest year ever recorded) which distorts the trend in the short term only means that you don't actually understand that reality matches the models rather better than you try to make out.

You have demonstrated nothing.

Br Cornelius

Edited by Br Cornelius, 08 December 2012 - 12:21 PM.

[Doug1o29]

Little Fish, on 08 December 2012 - 11:57 AM, said:

and that has been the problem with this discussion, you and doug have misrepresented the point being made, perhaps because you just regurgitate talking points from SKS.

In case you missed it, I am a dendrochronologist.  Statistical analysis is what I do for a living.  I have the equivalent of a Masters degree in statistics.

I do not use anybody else's "talking points" because I don't trust anybody else to get it right.  I run the tests myself.  For the same reason, I do not trust your opinion or those of your sources, whatever they are, without access to the analysis that led to their conclusions.  If you don't post that, you have not presented any evidence to support your contentions.  The work of climate scientists must be double-checked every bit as much as that of Anthony Watts.  Only then can I separate the wheat from the chaff.

By now, you should realize that I expect you to acquire the knowledge needed to double-check your statements yourself.  Since we started "discussing" things on UM, a half-dozen students in my department have started and completed masters degrees, one has done the same with a doctorate and another will defend his thesis, probably in February.  There has been more than enough time for you to become an expert in climatology and make a significant contribution to climate knowledge.  What am I to expect of someone too lazy to read a book on the subject while others get advanced degrees?
Doug

[Doug1o29]

Little Fish, on 06 December 2012 - 05:18 PM, said:

and doug if you are going to claim that HADCRU3 has a warming trend over that period, to be truthful you should mention that it is statisically INSIGNIFICANT, meaning the trend IS flat.

"Insignificant" does not mean zero.  It means that whatever the slope is, statistically it can't be distinguished from zero.  Lack of proof that a rate of slope is not zero does not mean that it IS zero.  And that cuts both directions:  If I point out that the trend in your fifteen-year dataset is +0.0485 degrees per year, that is still within the error limits of your data.  And seeing as that is closer to the mean estimate than it is to zero, it is actually a better estimate, albeit, not much better.  Even using your own methods, you have arrived at an incorrect result.

And because you cannot show that there is a fifteen-year period with a zero slope, you have not invalidated NASA's global climate model.

BTW:  that +0.0485 figure is the 1996-2011 average rise in global temps based on the NCDC dataset.  It is +0.0483 degrees C. during the period 1997-2011.  And both are better estimates than zero, even if I didn't remove autocorrelation or correct for small sample size and even if they aren't much better than zero.
Doug

P.S.:  I can run the RSS, Hadley and NCDC datasets side-by-side if you like.  I could also remove autocorrelation and correct for sample size and post the results here.  And I could also use monthly figures if you like.  Then we'll have exact figures to argue about.  Want to make a friendly wager on the results?

One problem with autocorrelation, though.  The solar cycle has a seven-to-eleven year lag.  To remove autocorrelation from nine years earlier, I will have to add nine years to the dataset.  That has some implications for your claims.

There is another little detail that I just noted:  the Pacific Decadal Oscillation is a function of the derivative of global mean temperature.  Which means that when the PDO is in its "warm" phase, global mean temperatures are rising and when it's in its "cool" phase, they are falling.  And seeing as there is now a way to forecast values for the PDO three to four years ahead, we can also forecast global warming that far ahead.  No climate models.  Just sea temperature measurements and old-fashioned statistics.  Sounds like a fun project.

One good thing about our discussions:  whenever I get PO'd with you and look up the actual studies, I always learn something new.  Usually it's the precise reason your arguments don't work, but once in awhile I learn something I didn't know existed (like the PDO thing above).
Doug

Edited by Doug1o29, 08 December 2012 - 09:18 PM.

[Br Cornelius]

Well worth reading;


I say, if your not terrified at this stage - your a fool.

Br Cornelius

[Little Fish]

Doug1o29, on 08 December 2012 - 08:47 PM, said:

you cannot show that there is a fifteen-year period with a zero slope, you have not invalidated NASA's global climate model.

more than 15 years, zero slope:
http://www.woodfortr...2gl/from:1997.1

[Br Cornelius]

Little Fish, on 09 December 2012 - 05:09 PM, said:

more than 15 years, zero slope:
http://www.woodfortr...2gl/from:1997.1

Keep ignoring an experts opinion and repeating the same false claim

Consistency is this case is not a virtue

I can get that trend line to jump all over the place by varying the start date from 1995 to 1997 and careful selection of the dataset. What have I proved - nothing !!!

Br Cornelius

Edited by Br Cornelius, 09 December 2012 - 05:39 PM.

[Doug1o29]

Little Fish, on 09 December 2012 - 05:09 PM, said:

more than 15 years, zero slope:
http://www.woodfortr...2gl/from:1997.1

And that is exactly why you need to learn to do arithmetic.

The Hadcrut3 data set shows an average temperature rise of +0.0372 degrees C. over the years 1997 to 2011.  That's based on an average of the actual numbers from the Hadcrut3 dataset.  The graph you referred to shows a zero rate of slope for the Hadcrut3 data set (red line) during that time.  That is because the graph does not give enough detail to show that.  And that's why your pretty pictures don't reflect reality.  I'll see if I can dig up the RSS dataset and post the numbers for it, too.
Doug

P.S.:  Note that the NCDC dataset (+0.0485 degrees average annual increase) and the Hadcrut3 dataset (+0.0372 degrees average annual rise) do not exactly agree.  However, they both agree that temps rose over that interval.  Both are estimates of the same reality.

Little Fish, why don't you actually post something that supports your contentions instead of trying to cherry-pick somebody else's data?
Doug

P.P.S.:  I found the RSS datasets (There are two - lower and upper stratosphere.).  Calculating the average world-wide is going to be a challenge as there are nine different subsets that have to be combined.  However, I calculated the change in lower stratospheric temperature over the United States for the1997-2011 period.  The average rate of temperature increase over that period was +0.0129 degrees C.  I decided to check for the level of significance and got a surprise:  F(1,15)=37.392, highly significant F(0.95,1,15) = 4.600.  I'll be back after lunch and check the significance levels for the NCDC and Hadcrut3 datasets.
Doug

Edited by Doug1o29, 09 December 2012 - 06:24 PM.

[Little Fish]

Doug1o29, on 09 December 2012 - 05:33 PM, said:

The Hadcrut3 data set shows an average temperature rise of +0.0372 degrees C. over the years 1997 to 2011.

but what does it show from the period I specified (April 1997 to present)?
http://www.woodfortr...m:1997.25/trend

[Br Cornelius]

Little Fish, on 09 December 2012 - 06:04 PM, said:

but what does it show from the period I specified (April 1997 to present)?
http://www.woodfortr...m:1997.25/trend

What you have demonstrated is that you cannot determine a meaningful trend over such a short time series since altering it by 1-2 years either side completely changes the outcome. That is the very essence of misunderstanding the purpose of stats.

Br Cornelius

[Little Fish]

Br Cornelius, on 09 December 2012 - 05:30 PM, said:

I can get that trend line to jump all over the place by varying the start date from 1995 to 1997 and careful selection of the dataset. What have I proved - nothing !!!

but this is what the top expert climate modelers and NOAA say:

"Near-zero and even negative trends are common for intervals of a decade or less in the simulations, due to the model’s internal climate variability. The simulations rule out (at the 95% level) zero trends for intervals of 15 yr or more, suggesting that an observed absence of warming of this duration is needed to create a discrepancy with the expected present-day warming rate"

the point you do not grasp (and as explained to you eight times previously), is that according to the top climate modelers and NOAA, there should be no periods of 15 years where the trend is flat (otherwise there is a discrepancy with their gcm models)

[Br Cornelius]

Little Fish, on 09 December 2012 - 06:12 PM, said:

but this is what the top expert climate modelers and NOAA say:

"Near-zero and even negative trends are common for intervals of a decade or less in the simulations, due to the model’s internal climate variability. The simulations rule out (at the 95% level) zero trends for intervals of 15 yr or more, suggesting that an observed absence of warming of this duration is needed to create a discrepancy with the expected present-day warming rate"

the point you do not grasp (and as explained to you eight times previously), is that according to the top climate modelers and NOAA, there should be no periods of 15 years where the trend is flat (otherwise there is a discrepancy with their gcm models)

And the point myself and Doug have tried to make you understand is that the flat line is a statistical artifact of the start point you have used. Your point is invalid because it is factually wrong. I can make the trend anything I want by simply moving the start and end points in a range of 1-2 years. i choose not to do it because to do so would be dishonest of me.

There has been no period of zero temperature growth.

Repeating a false assertion will never make it true Little Fish. It is a standard arguing point with you to focus in on a detail in order to obscure the greater picture, it hasn't work before and it won't this time. Can I suggest you go away and learn some basic stats so that you can actually make a meaningful contribution to the debate about temperature trends.

Br Cornelius

Edited by Br Cornelius, 09 December 2012 - 06:21 PM.

[Little Fish]

Br Cornelius, on 09 December 2012 - 06:08 PM, said:

What you have demonstrated is that you cannot determine a meaningful trend over such a short time series since altering it by 1-2 years either side completely changes the outcome. That is the very essence of misunderstanding the purpose of stats.

Br Cornelius

if you adjust the start point a little bit one way (we're not talking "1-2 years") and you get a tiny up trend, then adjust the start point a little bit the other way and you get a tiny down trend, then what you have found is a flat trend over that period. the period happens to be over 15 years which is what the top climate modelers and NOAA say is required to find a discrepancy in the models.

it's fun watching you dance around this.

Edited by Little Fish, 09 December 2012 - 06:24 PM.

[Br Cornelius]

Little Fish, on 09 December 2012 - 06:22 PM, said:

if you adjust the start point a little bit one way (we're not talking "1-2 years") and you get a tiny up trend, then adjust the start point a little bit the other way and you get a tiny down trend, then what you have found is a flat trend over that period. the period happens to be over 15 years which is what the top climate modelers and NOAA say is required to find a discrepancy in the models.

it's fun watching you dance around this.

It does not - it proves simply that you have inadequate data to draw a statistically meaningful conclusion. The temperature record did not start in 1997 and no amount of self blinkering can change that. If the error component in a dataset is to large then all you have is error and no actual useful data. This is why stats has techniques to deal with it and simply averaging the data over 15 years is not one of them.

Watching you display your ignorance is painful.

Br Cornelius

Edited by Br Cornelius, 09 December 2012 - 06:40 PM.

[Little Fish]

Br Cornelius, on 09 December 2012 - 06:37 PM, said:

It does not - it proves simply that you have inadequate data to draw a statistically meaningful conclusion. The temperature record did not start in 1997 and no amount of self blinkering can change that. If the error component in a dataset is to large then all you have is error and no actual useful data. This is why stats has techniques to deal with it and simply averaging the data over 15 years is not one of them.

Watching you display your ignorance is painful.

Br Cornelius

"an observed absence of warming of this duration <15 years> is needed to create a discrepancy" - these are the words of the top climate modelers and NOAA. stop being a King Cnut and accept it.