Anyone who has an interest in exploring patterns in global temperature should take a look around WoodForTrees.org. Paul Clark, a British software developer and “practically-oriented environmentalist and conservationist” has developed an online interface that allows anyone to go examine basic longterm trends in climate time series data (including the HADCRUT3 / GISTEMP Global Temperature & HADSST2 Sea Surface Temperature, along with sunspot activity and CO2 datasets).
The interface is incredibly intuitive, and allows a variety of transformations, averaging and trend estimations within graphs. After having spent literally hours playing around on this site, I completely agree with the warnings of ‘cherry picking‘ a dataset (i.e. choosing a certain year to start the trend to exacerbate a trend). To illustrate this ‘technique’, Paul has produced this classic graph:
Which goes to show that the temperature is either: 1) falling, 2) static, 3) rising, or 4) rising ‘really fast!’ -all depending on where you place the trendline.
As John eloquently explained in this comment a few days ago, “global warming stopped after 1998” is turning into one of the most common memes of the ‘skeptics’ and ‘deniers’. Alot of their argument relies on very heavily cherry-picked data – skeptical Science also have a great in detail discussion and counterpoint to this argument here. Contrast the above graph with the longer term view (consistent across multiple datasets), showing warming between 0.13-0.17°C/decade: