Summer mean temperatures, including data for 2011, have risen inexorably above the 1951-1980 average, unpublished research by NASA's James Hansen shows.

Hansen defined extremely hot summers as those exceeding three standard deviations from the average 1951-1980 climate.

One standard deviation captures about two thirds of observations compared with the seasonal summer average, and three standard deviations nearly 99.9 per cent.

In other words, a summer more than three standard deviations from the average should barely ever happen.

Yet they are occurring, in Europe in 2003, in eastern Europe in 2007, Russia and the Middle East in 2010 and Texas in early 2011.

The argument is that such increasing frequency can't be explained only by natural variation in the weather.

And the problem is growing, with Hansen's research showing average summer temperatures globally are up by 1 standard deviation since 2001 compared with 1951-1961 in a steadily rising trend.

He forecasts by 2050, assuming present trends of growth in fossil fuel carbon emissions, "extremely hot" summers in 1951-1980 terms would become the norm.