Requiring all grid boxes used to have no missing months is a pretty strict standard. I have Giss station data from a couple years ago and I could find only 77 stations world-wide that would qualify with perfect data through 2006. 52 of those are US. I realize they used Hadcrut instead, but it makes me wonder what percentage of global coverage they actually achieved. I think qualifying ocean grid-cells would also be pretty sparse. I suspect their coverage ended up being less than 25% of the globe. But I’m not interested enough to thoroughly check.

I downloaded a Giss table of global temperature anomalies and calculated the yearly changes. The 1998 temperature increase, although substantial, at 0.17C it wasn’t even a 2 sigma event.(+/-0.22C change in one year) There was a 2 sigma temperature decrease in 1999 that immediately wiped out all the 1998 increase and most of the 1997 increase.

There are seven 2 sigma temperature changes in the Giss record. Temperature increases in 1957 & 1977. Temperature decreases in 1890, 1964*, 1974, 1992*, and 1999. (* volcano?)

Here is a graphic. http://i30.tinypic.com/10o1vz5.gif

I don’t see how they can say there is anything being stored ‘in the pipe-line’ that is going to somehow appear a decade or so down the road.

The 32 year trend lines sure look to me like a little natural warming being alternately depressed then enhanced by the PDO every half cycle.

Kiehl, J. T. (2007), Twentieth century climate model response and climate sensitivity, Geophys. Res. Lett., 34, L22710, doi:10.1029/2007GL031383.

https://www.atmos.washington.edu/twiki/pub/Main/ClimateModelingClass/kiehl_2007GL031383.pdf

This paper (cited in Lindzen 2009, “Climate Science: Is it currently designed to answer questions?”, ArXiv) is only 4 pages long, mostly readable by non-specialists, and asks some VERY interesting questions, given that these questions are coming from an IPCC person like Jeffrey Kiehl…