Wednesday, August 27, 2008

Is it really so Simple?

My previous post identified a discrepancy between the global temp series from 1975 compared to the current temp series: the older data shows a remarkable post war cooling period while the latest versions do not. A simple reason for the difference (suggested in a comment by  Zeke Hausfather) is that the Newsweek temperature series from 1975 is Northern Hemisphere. The implication is that the cooling trend was a Northern Hemisphere phenomenon only. 

A fundamental and often ignored consideration is that there are an extremely large number of degrees of freedom to play with in these types of analyses. These include methods for handling missing data,  population growth, the explosion in the number of temperature sites, technology changes and so on. On the one hand is the argument that the temperature "record" is basically invariant to different approaches to these problems. On the other hand, it may be that the degrees of freedom can easily be manipulated to produce the result you want.  As a professional statistician I am predisposed to the latter view.  So let's check the temperature series separately by hemisphere:

The middle graphs is the Northern Hemisphere Temperature record. As you can see the post war cooling period that was so dramatic in the NCAR graph published in Newsweek has been flattened out.  So here it is again: the trend, which is so dramatic in 1975, is no longer even part of the historical record. How can that be?