OK, here's a synopsis ...
"Reconciling Vertical Temperature Trends Workshop October 27-29, 2003". Some of the skeptics who have posted on this topic might do well to study the presentations available there. Among other things, the whole idea of urban heat islands is completely debunked (maybe the surface temps are not so "deeply flawed" after all). Also of interest are the many expositions on the complexity of the distribution of vertical temperature changes and the degree to which there are very large and persistant differences on regional and hemispheric bases. You get very different results if you look at the atmosphere/surface anomalies in the tropics versus in the northern hemisphere, etc. So the idea of a single-valued difference expands into a whole spectrum of differences, depending on where you look for your non-surface data. The idea that there is an alternative set of non-surface measurements to be taken as opposed to the surface measurements is a myth. What does exist is a whole array of non-surface measurements which could be used - but the theory does not exist today to tell you which one to use, and how it should relate to corresponding surface temperatures. As far as the expected relation between surface and atmospheric temps, one presenter (Trenberth) went so far as to say "Given that global mean surface temperature anomalies are dominated by continental land areas in the NH, while global satellite temperature anomalies are dominated by the Tropical regions, why should we expect a relationship at all"?
(BTW "NH" in this context means "Northern Hemisphere", which is where the majority of the world's landmass lies, not "New Hampshire" which is really tiny :)
