A few posts ago, I used the term “‘fingerprint’” (in quotations) in connection with the big red spot that one commonly sees in the upper tropical troposphere in contour plots of projected temperature trends by latitude. On an earlier occasion, we’d talked about Ross McKitrick’s “T3″ ( here.)
A few readers contested my use of the term “fingerprint” in this context. “Fingerprints” are not a topic that I’ve covered at the blog so far nor can I say that I fully understand the operational usage of this term by climate scientists.
I get the impression that this is local dialect in climate science for something already known by a different name in conventional statistics. But without being able to work through examples from beginning to end, I find it hard to say what it corresponds to exactly. (And, of course, because it’s climate science, there’s nothing remotely resembling fully articulated examples to assist translation out of the local dialect.) Some day, I’ll try to figure it out.
IPCC TAR has a section entitled 12.4.3 Optimal Fingerprint Methods in which they refer to “optimal detection” (another term in the local dialect). For example, :
optimal detection studies of surface temperature have been extended…
Elsewhere IPCC TAR had a very peculiar appendix entitled: “Optimal Detection is Regression”, in which they say:
The detection technique that has been used in most “optimal detection” studies performed to date has several equivalent representations (Hegerl and North, 1997; Zwiers, 1999). It has recently been recognised that it can be cast as a multiple regression problem with respect to generalised least squares (Allen and Tett, 1999; see also Hasselmann, 1993, 1997)
Do “optimal fingerprints” also have something to do with multiple regression? It seems quite likely. It seems very odd that climate scientists are simply carrying out multiple regression with generalized least squares using strange and overblown terminology like “optimal detection” and “optimal fingerprints” without referring to the extensive pre-existing literature, but that seems to be what IPCC TAR says.
Lucia has a long and interesting thread on whether or not the T3 red spot is a “fingerprint” of AGW; I refer interested readers to her comments, with some of her readers contesting the idea that the red spot is an AGW fingerprint or that the seeming absence of the red spot “matters”.
For reference, here is an image shown last year at RC showing the effect of 2xCO2. (The image is entitled “2xCO2_tropical_enhance.gif”.
Figure 1. Effect of 2xCO2 (from RC).
Gavin also showed a similar T3 hot spot with large solar forcing, observing:
If the pictures are very similar despite the different forcings that implies that the pattern really has nothing to do with greenhouse gas changes, but is a more fundamental response to warming (however caused). Indeed, there is a clear physical reason why this is the case – the increase in water vapour as surface air temperature rises causes a change in the moist-adiabatic lapse rate (the decrease of temperature with height) such that the surface to mid-tropospheric gradient decreases with increasing temperature (i.e. it warms faster aloft). This is something seen in many observations and over many timescales, and is not something unique to climate models.
As a little exercise, I downloaded monthly data from UAH, made zonal averages in 10-degree bands and calculated trends for three levels (T2LT, T2 and T4) and likewise for CRU, and then made a contour plot, using altitudes of 2.5 km (740 hPA), 6.1 km (466 hPa) and 20 km (75 hPa). This resulted in the following graphic:
Figure 2. Contour of trends using CRU and UAH. (deg C/year)
Re-examining the matter, I noticed two quotes in AR4 that seem relevant (and which haven’t been mentioned at Lucia’s yet). AR4 Box 8.1 states:
GCMs find enhanced warming in the tropical upper troposphere, due to changes in the lapse rate (see Section 9.4.4).
At low latitudes, GCMs show negative lapse rate feedback because of their tendency towards a moist adiabatic lapse rate, producing amplified warming aloft.
If these are relevant quotes, then the presence/absence of Clifford the Big Red Dog might well be related more to whether water vapor/cloud impact has been adequately modeled, as opposed to the direct GHG effect.. If this interpretation is valid, then the point made by Gavin and others would seem to be an empty rhetorical point, as it seems to me that there has been more concern over the “multiplier” effect of water vapor/clouds than over the direct GHG effect. So the point remains worth discussing whether or not it is a GHG “fingerprint”. The water vapor/cloud feedbacks are very bit as relevant to understanding the matter.
As discussed on many occasions, it’s quite possible that the Big Red Dog is really there and the issue is with UAH algorithms (at some point, I’ll try to decode RSS gridcell data and do the same plot from their data.)
Having said that, the pattern, as shown above, has both points in common and that differ from the models. Stratosphere cooling and an Arctic hot spot are in common; but, in addition to the seeming absence of the Big Red Dog, there seems to be much more noticeable NH-SH asymmetry in the observations than in the models.
UPDATE: Here’s the corresponding graphic for RSS (which uses 5 levels – apparently there are some issues with the lower stratosphere values but I can’t comment on this. )
Figure 2. RSS Trends. (deg C/year)