It appeared to have shaken the credibility of one of the most important global warming data sets in the world. A blog-inspired campaign by amateur climate sceptics seemed to show that numerous weather stations across the US were so poorly located they could not be relied upon.
But a new scientific analysis, using data from the sceptics, has shown that, if anything, the poorly located stations underestimate warming, rather than exaggerating it.
The US temperature record uses data from thousands of weather stations spread around the country. Their accuracy was called into question following a campaign by climate sceptic Anthony Watts, an ex-weatherman who runs the influential blog WattsUpWithThat.
He set up a site called surfacestations.org for readers to post photos of poorly located weather stations, particularly in places that could be influenced by artificial heat, such as air conditioning units or car parks. The photos were compiled into a book published by the right-wing thinktank the Heartland Institute. In it, Watts wrote: "The conclusion is inescapable: The US temperature record is unreliable. And since the US record is thought to be 'the best in the world,' it follows that the global database is likely similarly compromised and unreliable."
But scientists at the National Climatic Data Centre (NCDC) in North Carolina have analysed the weather station data to see what difference poor location actually makes. Watts had ranked the stations by his estimation of the quality of their location, so Dr Matthew Menne and colleagues compared the results from high- and low-ranked stations. They described their results as "counterintuitive" – poorly located stations were actually more likely to be cooler than those in better locations. This is probably because the poorly located stations are more likely to use more up-to-date measuring equipment called Maximum-Minimum Temperature System (MMTS), which has a slight "cool" bias that is already well documented.
The complete article in the UK's Guardian newspaper is here.