Nick Stokes has been messing around with another regression of the Antarctic, he’s done some good stuff but I believe is seeing some of the difficulty of spatial trend containment using PCA methods. It’s fairly interesting:
Nick Stokes notes of his methods calculation and without CI’s or valdation :
“Again a pattern somewhat like that of O10, but with higher trends everywhere. The continent trend of 0.18 C/decade is very high.”
I believe Nick has noted that he needs to do further analyses (senstivity testing?) on this method, but he already is going to apply it to the West Antarctica alone per Steig’s suggestion. I do hope Nick does the proper sensitivity testing as too often, when a “good” result pops up immediately, motivation for further testing is dampened.
I think Nick is mostly just playing around now. We did this as well. Trying a bunch of random crap helps make the information make sense, as you get to see how the results respond to different methods. That helps you realize what is important and what is not. I imagine it will be quite some time before Nick settles on a validation method.
Ryan’s right – I just at this stage wanted to see what TempLS could do. I’ll produce some testing soon, but as Ryan says, getting the right test needs thought. So I’m not suggesting these results should be set against the better tested results of S09 and O10. There’ll be more.
And, of course, once achieving statistical significance, we all know an Antarctic decadal T-trend of, e.g., 0.05-0.2 C (whatever) will also be physically significant.
Should these untested results be published onto the internet, without validation or at least good documention, as part of a “just playing” around exercise? This question is not posed as a legal issue or even one of internet protocol, but in the view of the Eric Steig hand waving critiques of O(10) we have seen recently. Using an untested method and obtaining an unexpected result for the Antarctica continent and then applying it to West Antarctica seems like a double case of bad judgment to me. How has the method in question performed against other methods in other global areas?
No need to worry – yet. Let us see how fast we get verification/validation and how we get it. Just wondering also whether we would have receieved an early report of trends if those trends had been unexpectedly low.
the Air Vent 
Seems like there’s little a priori basis for the weighting decisions there …
Nick Stokes notes of his methods calculation and without CI’s or valdation :
“Again a pattern somewhat like that of O10, but with higher trends everywhere. The continent trend of 0.18 C/decade is very high.”
I believe Nick has noted that he needs to do further analyses (senstivity testing?) on this method, but he already is going to apply it to the West Antarctica alone per Steig’s suggestion. I do hope Nick does the proper sensitivity testing as too often, when a “good” result pops up immediately, motivation for further testing is dampened.
Is TempLS a method of Nick Stokes and without significant testing to show robustness?
Kenneth,
I think Nick is mostly just playing around now. We did this as well. Trying a bunch of random crap helps make the information make sense, as you get to see how the results respond to different methods. That helps you realize what is important and what is not. I imagine it will be quite some time before Nick settles on a validation method.
Ryan’s right – I just at this stage wanted to see what TempLS could do. I’ll produce some testing soon, but as Ryan says, getting the right test needs thought. So I’m not suggesting these results should be set against the better tested results of S09 and O10. There’ll be more.
And, of course, once achieving statistical significance, we all know an Antarctic decadal T-trend of, e.g., 0.05-0.2 C (whatever) will also be physically significant.
Should these untested results be published onto the internet, without validation or at least good documention, as part of a “just playing” around exercise? This question is not posed as a legal issue or even one of internet protocol, but in the view of the Eric Steig hand waving critiques of O(10) we have seen recently. Using an untested method and obtaining an unexpected result for the Antarctica continent and then applying it to West Antarctica seems like a double case of bad judgment to me. How has the method in question performed against other methods in other global areas?
#6 and #7, no need to get worried already.
No need to worry – yet. Let us see how fast we get verification/validation and how we get it. Just wondering also whether we would have receieved an early report of trends if those trends had been unexpectedly low.
Thanks the info Mr.
From: Alieph