Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace

Discussion > (A)GW Lite

How can you investigate a phenomenon by comparing its recent behaviour with its past behavior, if the past behavior includes the recent?

By examining the rate of change over time. It is a widely adopted method. I can't find any textbook prohibitions. Apparently you can't find any textbook prohibitions either. When it comes to formal logic, you state that I've been nabbed by a fallacy, but then you say:

I am not aware/knowledgeable of logical notation that can be used to demonstrate this formally, but it can surely be done.

It's not conclusive, of course, but suggestive.... ;-)

May 28, 2012 at 1:20 AM | Unregistered CommenterBBD

I give up.

This will not be resolved by us talking. You need to think over it.

May 28, 2012 at 3:03 AM | Registered Commentershub

Shub: How can you investigate a phenomenon by comparing its recent behaviour with its past behavior, if the past behavior includes the recent?

That is how moving averages work. Uncontroversial, I think.

BBD, I don't know whether or not there is a ten year oscillation, but the curve is certainly highly variable over time periods of about 10 years. Hence any claims based on periods shorter than 20-30 years are suspect and need to be made using the 'worst case' for any particular argument. If this is done, my guess is that no arguments can be sustained. My feeling is that it is unreasonble to make any assertion based upon a period of less than about 20-30 years.

I aimed my statement at both sides for fairness. I know for certain that sceptics make assertions about 10-15 year periods (there are some in this thread); but I imagine that some on the orthodox side might use similarly short periods when it suits their argument :-)

What is your view of the debate above?

I've spent the last hour playing with the graphs at WoodForTrees and I can really only see one thing clearly: a consistent upward slop after about 1970, in what looks like a straight line (kind of like the blade of a hockey stick ;-) The record before that is rather lumpy, with a slight upward bias but with the level of noise present it would take a lot of study to come to a conclusion. I defer to the researchers who have spent whole careers looking at this data; my hour with it is neither here nor there.

Unfortunately, the good faith of those researchers is questioned. Accusations of dishonesty are easily thrown; and if enough mud is thrown, some will stick, whether justified or not. This aspect of both the serious and barmy sides scepticism makes me queasy; there are some sensible people here who have serious objections to the orthodox view and I might sympathise more if seemingly everyone on the orthodox side were not smeared with claims of dishonesty. Having recently been called dishonest in this thread, for no reason that is apparent to me, it has become clearer to me how central this (successful) tactic is to the sceptic cause.

I was vacillating over whether to buy the HSI book, but from reading the first chapter I have the impression that a major component of the book involves questioning the honour of the scientists. This makes me less likely to read it. Maybe I got that wrong - any views?

May 28, 2012 at 3:06 AM | Unregistered CommenterBitBucket

"That is how moving averages work. Uncontroversial, I think."

What are you talking about?

This goes back to this graph

http://woodfortrees.org/plot/best/from:1900/to:2010.2/mean:12/plot/best/from:1900/to:2010.2/trend/plot/best/from:1950/to:2010.2/trend/plot/best/from:1975/to:2010.2/trend/plot/hadcrut4gl/from:1900/to:2010/mean:24

and drawing this inference: "The rate of warming is accelerating".

It has nothing to do with moving averages. Moving averages are calculated using adjacent values alone.

May 28, 2012 at 3:59 AM | Registered Commentershub

What are you talking about?

Er... good point! Well, moving averages do combine recent data with old data, but it isn't really the same thing. Sorry.

On the graphs, ok, I've spent another hour staring at them and I'm no wiser. Somehow drawing a line back to 1900 (ie the edge of the WfT graph) from 2010 makes me uneasy, although I can't really express why. I prefer the BEST analysis at http://berkeleyearth.org/analysis/ Using that one, drawing a line to 1900 seems a valid thing to do (ie not at the edge of the data). But I am not comfortable in concluding that, because a line from 1900 to 2010 on the BEST graph has a lesser slope than one from 1970 to 2010, this indicates acceleration; such a conclusion seems to ignore the deceleration in the 40s, 50s and 60s. It could equally be returning to trend. It all depends upon where the lines go after 2010, which of course we wont know for some time.

I don't know why I'm commenting on that anyway. I don't know the first thing about it. Shut up Bit!

May 28, 2012 at 5:13 AM | Unregistered CommenterBitBucket

shub

I give up.

Well, you haven't made your case, and you haven't shown any external references to support it. Until you do, we shall have to agree to differ. But note: you are no longer entitled to say that what I did was 'wrong'. You can think it, but you can't say it. Unless you can back it up with coherent argument and references. I don't mean to sound harsh, but that's how it works.

May 28, 2012 at 9:41 AM | Unregistered CommenterBBD

BitBucket

On the graphs, ok, I've spent another hour staring at them and I'm no wiser. Somehow drawing a line back to 1900 (ie the edge of the WfT graph) from 2010 makes me uneasy, although I can't really express why.

Perhaps this expanded x-axis view will help?

We can demonstrate that the rate of warming has accelerated from this graph by comparing the decadal trends for the periods 1910 - 1940 and 1975 - 2010. I've been through all this at some length on the thread, so it might be helpful to review comments at this point. But to recap briefly:

1910 - 1940
0.14C

1975 - 2010.2
0.28C

The rate of warming has *doubled* in the later period. I hesitate to recommend anything as unscientific as eyeballing the graph, but it is quite clear that the rate of warming has accelerated in the second half of the C20th. But be blind to that! Instead, compare the trends:

Decadal trends:

1900 - 2010 = 0.09C

1950 - 2010 = 0.18C (2x 1900 - 2010 trend)

1975 - 2010 = 0.28C (3.1x 1900 - 2010 trend; 1.5x 1950 - 2010 trend)

The rate of warming has increased over the century.

Shub says:

You are falling for a logical fallacy. I am not aware/knowledgeable of logical notation that can be used to demonstrate this formally, but it can surely be done.

I don't believe he is correct to assert a logical fallacy here and it is important to note that he can't produce any supporting references. Earlier in the thread I asked him for a textbook reference specifically cautioning against examining a temperature time series in this way. Shub was unable to provide one. So it's just his opinion vs the numbers and common sense.

You will note that Shub is prone to incorrect and unsupported assertions in place of supported, coherent argument :

Even over land, there is no anthropogenic signal. Certainly there is none in the global series.

Anyway, this is done to death, so I'll leave you to review comments and consider the evidence.

May 28, 2012 at 10:20 AM | Unregistered CommenterBBD

BitBucket

Sorry - forgot:

That is how moving averages work. Uncontroversial, I think.

Try a 3rd order polynomial fit.

Accelerated rate of warming. Plain as day.

May 28, 2012 at 10:27 AM | Unregistered CommenterBBD

Same thing with HADCRUT4: 3rd order polynomial shows accelerated rate of warming. Let's try GISTEMP. Yup, same thing. NOAA global. Same again.

Who needs boring old linear trends!

May 28, 2012 at 10:50 AM | Unregistered CommenterBBD

You are wrong in drawing the conclusion you did.

May 28, 2012 at 11:01 AM | Registered Commentershub

"... back it up with coherent argument and references."

I've explained it to you in several ways, on several occasions. I am not the only one to have done so.

Your requirement that I find what I say written in a textbook, is arbitrary.

May 28, 2012 at 11:04 AM | Registered Commentershub

Your definition of 'coherent' is: written down by someone else somewhere.

I am yet stumble upon what you ask for. I probably will. Then what?

How come you have made no effort whatsoever to consider you might be wrong? Instead, you just repeat yourself.

How come you make no effort to look for it yourself? Why not present references of people using similar methods to prove acceleration, apart from the fraud IPCC SPM, that is?

May 28, 2012 at 11:11 AM | Registered Commentershub

Eh. GISTEMP link again.

May 28, 2012 at 11:26 AM | Unregistered CommenterBBD

You are wrong in drawing the conclusion you did.

Then demonstrate the error.

Your definition of 'coherent' is: written down by someone else somewhere.

My definition of coherent is logically consistent. You are conflating this with 'supported by academic references'. That's important too, but in your case we have neither a coherent argument, nor references.

How come you have made no effort whatsoever to consider you might be wrong? Instead, you just repeat yourself.

Have a good look at those 3rd order polynomials. I can't invalidate the hypothesis. Nor, it seems, can you.

All you do is claim that I'm wrong. It's up to you substantiate that claim. If you can't, you will have to stop making it. That's how it works.

fraud IPCC SPM

Considering that you have absolutely failed to make the case that warming isn't accelerating, and I have demonstrated it in three ways, you should not be accusing the IPCC of fraud. You have no basis for doing so. You will have to stop making this claim.

May 28, 2012 at 11:36 AM | Unregistered CommenterBBD

Tell you what though, Shub, I wouldn't want you to think that this has been a pointless exercise. In recognition of this very long disagreement between us, in future I'll demonstrate the accelerating rate of warming over the C20th with a comparison of the 1910 -1940 and 1975 - 2005 trends and by fitting a third order polynomial to the entire series.

Would you be happy with that?

May 28, 2012 at 11:52 AM | Unregistered CommenterBBD

It was the synthesis report, not the SPM.

I looked at the mohyu (?sp) graph you linked to. The temps don't even look like anything I recognize (I am not talking about the polynomial fit). I don't know why.

We can let it rest for a while, and come back if/when something comes up.

May 28, 2012 at 12:34 PM | Registered Commentershub

BBD, the moyhu graphs are nicer than those at WfT :-)

The trouble I think I have is that the word 'accelerating' implies that the rate of change is still increasing. To me it looks like a straight line after 1970. But maybe I'm also guilty of drawing conclusions from too short a baseline.

I'd suggest a compromise: "It has accelerated". It may really be accelerating still but we wont know for 20-30 years, by which time it will be too late to change. Hence in my mind it is best to attack the issue now, not to wait.

I find it a bit comical that with all our modern instrumentation measuring the temperature all over the globe and with satellites circling in space, we have trouble agreeing upon the state of the global climate in the last century, and yet the MWP was apparently as warm as it is now. How dow we know that? Well we have some tales of Vikings and Vinland and a few huts in Greenland and thats about it...

May 28, 2012 at 2:16 PM | Unregistered CommenterBitBucket

Shub

Moyhu defaults to annual means, but the data is direct from the original source. Here's HADCRUT4 (Moyhu) vs (Wft). Same curve. No funny business ;-)

May 28, 2012 at 3:31 PM | Unregistered CommenterBBD

It is not the same curve. Moyhu is plotting from 1900 to 2011. WFT plots from 1850.

I am looking at HADCRUT4 in my eclipse window right now.

May 28, 2012 at 4:11 PM | Registered Commentershub

BitBucket

Yes, let's say 'it has accelerated'. Indeed, modelled projections suggest that further acceleration in the rate of warming is several decades off.

The point here is to counter the meme that 'warming in the first half of the C20th was the same as in the second (so AGW is a myth etc)'. We can see that this is not true by comparing the respective trends for the two warming periods using several data-sets (not just HADCRUT3) and from the full-series polynomial fits. I'll be using those a lot more from now on. WfT doesn't do polynomials, which in part explains my historical reliance on linear OLS trends ;-). And Moyhu is prettier, of course.

Just one more thing. The 'MWP' still needs to be treated with caution. My understanding is that it wasn't a single, synchronous global event but a series of regional events spanning several centuries (even the dates are vague, but let's say ~900CE - 1300CE). So while it is probably correct to say that some regional warming events may have been as warm as the mid-C20th or possibly warmer, we are comparing *regional* temperatures to a modern *global average* temperature. This needs to be kept in mind.

May 28, 2012 at 4:19 PM | Unregistered CommenterBBD

Shub

Use the blue and pink bars below the x-axis to scale it until you see the full time-series. Or set '1840 - 2020' or similar in the x-axis value field on the right.

May 28, 2012 at 4:22 PM | Unregistered CommenterBBD

What is your 'eclipse window; btw?

May 28, 2012 at 4:24 PM | Unregistered CommenterBBD

BBD, yes that was my point: we have oodles of data about here and now and can still manage to dispute current global temperatures etc. Yet with just a little anecdotal evidence from the North Atlantic and some proxy data that they otherwise do their best to discredit, sceptics claim a global MWP (or do they? maybe they don't claim it was global, in which case it is irrelevant).

May 28, 2012 at 6:04 PM | Unregistered CommenterBitBucket

The point here is to counter the meme that 'warming in the first half of the C20th was the same as in the second (so AGW is a myth etc)'

How do you 'counter' something that is true? By making false claims, I suppose.

I use Eclipse to run R.

May 29, 2012 at 1:50 AM | Registered Commentershub

Listen , you guys. Though I think you know this anyway.

Given a time series but lacking a model for what generated it, you can draw pretty much whatever conclusions you want, depending on what model you dream up (or which you don't even consciously dream up but is implicit in your arguments).

If you have a model, you can draw conclusions from the time series which are valid to the extent your model is valid. Without a defined and valid model, you are in the realm of debating the number of angels dancing on the head of a pin.

Myself - I tend to think there are some things that are inherently incapable of being modelled - they are chaotic and don't have definable statistics at whatever level/timescale you use to look at them. I have a feeling that the climate falls into this category. Which makes me believe:

- You fellows are unlikely to reach agreement
- Thinking models can be used to predict climate (eg Met Office) is an exercise in self-delusion.

May 29, 2012 at 9:36 AM | Registered CommenterMartin A