What Is The Average Global Temperature? asks Adam Retchless
Adam's question is a good one.
In brief, and very much oversimplified:
the global average, or Global Mean Surface Temperature, GMST, is an integration of direct measurements in modern times, or proxy1 data from historic and prehistoric times. It is calculated by making allowances for localised effects such as the 'urban island'2 effect. From all of the data that might be used, data sets are selected3 that are the most likely to have been properly and accurately recorded, that are as free from error as humanly possible and that contain no unaccountable deviations from contemporaneous data. These are integrated into regional datasets, which are then integrated into global datasets.
It turns out that we do not need to plot averages of global temperature to demonstrate the fact of global warming. What can be more readily averaged is warming trends. In a nutshell, if localised direct measurements and proxies show the same temperature trend at scattered locations across the planet and they all fit the same time scale, then those trends strongly indicate a global trend within that time frame.
It is the non contemporaneity of scattered records which shows that the so called 'medieval warm period' and the 'little ice age' were very localised events. The warming and the cooling trends did not appear at the same places in the same years across the entire planet, across the entire northern hemisphere, or even across the entire northern temperate zone.
Much of the data picked up by global warming deniers is modern or historical localised data. The famous 'hockey stick' graph integrates contemporaneous data from across the globe. Its detractors cherry pick data from different regions in different years within a time slice and claim that this negates the whole graph. It doesn't. There was no global medieval warming period, and no global little ice age. The debate on this resembles the rather silly notion that heavy snow falling in northern winter 2009/2010 demonstrated global cooling - at the same time as there were heat waves and drought in New Zealand and Australia!
 - temperature proxies can be human or natural records. Human records include government records, personal observations as in diaries, and ship's logs. Natural records include e.g. tree rings, sediment cores, peat bog cores and ice cores. The correlation between oxygen isotope records, sea levels and temperature is strong enough to infer any one proxy from the other two.
 - the urban island effect is the effect whereby urban areas tend to exhibit higher temperatures than their immediate rural surroundings.
White Roofs Could Reduce Urban Heat Island Effect
- this procedure does not equate to selecting datasets to fit the graph. The datasets are selected for scientific quality before the math is ever done to create the graph. This simple fact is obviously not known to the people who makes such a big fuss about the stolen emails.
The goal is not to hide the data, but to accurately represent the real state of global temperatures. Anyone who says otherwise is ignorant or dishonest.http://scienceblogs.com/tfk/2009/11/stolen_emails_climate_change_a.php
Related reading from The Chatter Box:
Global Cooling: A Good Story, If Only It Were True.
Global Cooling : How Wrong Can You Get?
Global Cooling : Beyond Parochialism
Join The Navy : See A Nuke
Himalayan Hype : Reading Between The Lines
How To Model A Smoking Gun