I've been arguing again, this time at CA and Lucia's. There's a series of threads of which I've linked the latest. The issue arose in connection with the SH proxy paper by Gergis et al, recently famously taken back for repairs (so I can't link to it). The method issues are described in an extant paper by Neukom and Gergis, 2011.
The issue is a claim that a common practice of making a last-stage proxy selection by observed correlation with temperature in the calibration period (when instrumental readings are available) constitutes a "selection fallacy". I made myself unpopular by asking just what that is - the response seemed to be that I'd been told and anyway any right-thinking person would know.
But there have been attempts to demonstrate it, with varying effects, and whatever the "fallacy" was, it seems to be shrinking. Lucia says now that she doesn't object in principle to selecting by correlation; only to certain methods of implementing. There was an interesting suggestion at CA from mt to use MSE rather than correlation for selection.
Roman's CA example, linked above, was held to show a selection bias. But it turned out that there is in his CPS method a bias from the limited number of proxies, and selection by correlation had no more evident effect than random selection. He then gave another example with white noise and more proxies with higher S/N which did seem to show some selection effect.
Anyway, I've been doing some tinkering with methods, which I'll describe.
Monday, June 25, 2012
Friday, June 15, 2012
TempLS showed changed very little from April to May. GISS showed a rise, from 0.55 °C to 0.65 °C. GISS has been rising steadily, while other indices that showed bigger rises in March/April have paused. Time series graphs are shown here
In other news, you should check that link for the Arctic ice changes over the last week. Ice was holding up well, but has dived to a record low for this time.
Posted by Nick Stokes at 10:31 AM
Tuesday, June 12, 2012
The TempLS analysis, based on GHCNV3 land temperatures and the ERSST sea temps, showed a monthly average of 0.53°C for May, unchanged from April (April rose by a few thousandths with late data). UAH also was virtually unchanged. It was still very warm in much of NH, including the Eastern USA. There are more details at the latest temperature data page.
Below is the graph (lat/lon) of temperature distribution for May.
Posted by Nick Stokes at 10:57 AM
Thursday, June 7, 2012
I made a small and somewhat inaccurate contribution to another WUWT thread, this time on Arctic Ice volume, and a figure quoted by the Guardian which I found surprising. My error is explained here; Stoat's account of the matter is here.
Anyway, it made me realise that I should pay more attention to ice volume. Although I criticise the Guardian's 75% reduction since 1979 as a mild exaggeration, the mean reduction of 66% is still an impressive figure. So I downloaded the PIOMAS data, and put it through the anomaly visualisation program of my previous post.
Posted by Nick Stokes at 9:17 PM