IPCC comparison

[Patrick LEE]

Originally Posted by mchandler

I do not believe there is any complex issue in comparing surface air temperatures amongst models. The variable "surface air temperature" is the value used in Chapter 10 IPCC AR4 WG1 and is comparable to the variable by that name in EdGCM/Model II.

Comparing surface air temperatures amongst models is so complicated that we all just ignore the complications.
The definitions of surface air temperature in different GCMs can be different. Even using the same definition, different vertical resolutions among models will lead to different surface air temperatures (becasue they are defined at different altitudes).

The conceptual error of surface air temperatures as defined in GCMs are, generally, larger in models with low vertical resolution (such as Model I and Model II) -- For simplicity, just ignore the conceptual difficulties of the measured surface air temperature for this discussion and think about the 2-meter temperature as the measured air temperatue.
Because the surface air temperature defined in GCMs are defined at too high in altitude (compared to the measured surface air temperature), people introduced the concept of "near surface air temperatures" into GCMs in order to reduce the conceptual error. Personally, I do not like the near surface air temperature because I think it is ugly and dirty. But the models with 12 vertical (dynamic) layers or fewer do need the help of the near surface air temperature. For models with very high vertical resolutions, we should not use the dirty near surface air temperature. As the computers become more and more powerful, the concept of near surface temperature may go way.

Evey thing in the real world is very complicated. Only the classroom exercises are simple.

Patrick.

P.S. I copy and paste part of the two emails exchanged between myself and Gary Russell below so that one can see how dirty the near surface aur temperature is. In the emails, TAS stands for the surface air temperature, TS stands for the near surface air temperaure, and TG is the composite ground temperature. (Warming: The discussion is stopped in the middle because I have no time to write. Hence, please do not take this as the final words on this topics.)

On 25 Feb, 2008. I wrote in my emails:
......
Also, I think it may be better to use a parameter 'RATIO' for diagnostic purposes:
Hence
TS = RATIO * TG + (1. - RATIO) TAS (0 < RATIO < 1)
instead of
TS = .5 * TG + .5 * TAS
....

Gary relied on 26 Feb, 2008:
....
TAS, the surface air temperature, is really calculated too high up in
the atmosphere; it is used and is accumulated in subroutine SURFCE.
I defined TS, the near surface air temperature, as: RATIO*TG +
(1-RATIO)*TAS. As you indicate, this RATIO does not have to be .5 .
I chose it for simplicity and becasue it matches near surface air
temperature observations reasonably well. A while ago, .5 was better
than either .4 or .6 , but this could be cheked again. I suppose it
could also have a seasonal and location dependency. If RATIO is a
global constant, it may be that .5 is not the optimal number.
.....