Re: [Phys-l] global temperatures

[John Denker <jsd@av8n.com>]

On 04/04/2009 08:02 AM, Rick Tarara wrote in part:
> We need to keep the questions clear here.  The polar melting (and there has 
> been some controversy on that lately as well),    [1]

Controversy?  There's also controversy about evolution, and controversy
about whether astronauts landed on the moon, and controversy about 
whether Elvis is dead.  Controversy is not evidence.  Please let's
stick to the evidence.

See also
  http://www.reuters.com/article/latestCrisis/idUSN03361051

> the last 10 years when the global temperature has not increased.    [2]

How do you know that?  Just because Michael Crichton said so
doesn't make it so.

This is an interesting bit of spin:  Take a point [1] that is well 
supported by the evidence and say it is controversial ... and then
take a point [2] that is not supported by the evidence and blurt 
it out as if it were a God-given fact.  Most people on this list 
are not fooled by this sort of spin.  The facts are not changed by 
this sort of spin.

In science we learn to deal with data that has fluctuations.  The 
fluctuations in the temperature record are such that over any 10 
year period ΔT/Δt is not a reliable estimate of the actual trend.  
More careful analysis is required ... maybe something as "sophisticated"
as the curve-fitting that is taught in introductory physics courses.
If you take a 5-year moving average, so as to get rid of the worst 
of the fluctuations, the point for 2008 is markedly higher than the 
point for 1998.
  http://data.giss.nasa.gov/gistemp/graphs/Fig.A2.lrg.gif

Yes, you can find wiggles on top of the trend.  But the trend is 
still there.


> None of these questions should be taken as nay-saying or blind skepticism, 
> rather (I think) legitimate questions about the MODELS which we are (forced) 
> to use for future planning.  All data, including a decade long pause in 
> temperature increases, are tests for these models.  How well do they 
> accommodate such data?

Twenty years ago, Hansen et al. used their model to make predictions. 
The model's 20-year prediction "accommodates" current data reasonably 
well.  
  http://pubs.giss.nasa.gov/abstracts/2006/Hansen_etal_1.html

You can find stretches where the data is below the trend for two or
three years at a time, but mostly it just follows the trend. If you 
have a model that you think is better, show us the model and explain 
why it is better.

There are well-established statistical techniques for validating the
predictive power of such models.