The decrease recorded in the Earth's temperature between the 1940s and 1970s was caused by a sudden cooling of the oceans in the northern hemisphere. That is the verdict of climate researchers in the US and UK, who have found that that much of this cooling could have been brought about by a rapid influx of fresh Arctic water into the North Atlantic. Others, however, dispute this finding, arguing that the temperature drop can be explained by longer-term ocean phenomena and human pollution.

Although the surface of the Earth is some 0.8 °C warmer than it was at the beginning of the 20th century, that rise in temperature has not been steady. Increasing until the 1940s, it then fell slightly over the next 30 years, before climbing again from the mid-1970s onwards. This pattern is widely regarded as an underlying warming of the planet by increased concentrations of atmospheric greenhouse gases, combined with a mid-20th century cooling brought about by a rise of sulphate aerosols in the troposphere – which reflect some solar radiation – as well as fluctuations in the world's oceans that take place over decades. The fact that mid-20th century cooling was confined to the northern hemisphere seems to support the aerosol hypothesis, given that the fastest rise in emissions of sulphate aerosols was seen over North America, Europe and Asia.

However, David Thompson of Colorado State University and colleagues believe that this analysis is based on a flawed reading of changes to the temperature of the northern hemisphere's oceans. They point out that identifying trends in surface temperatures requires removing the effects of short-term events that are cyclical or random, such as the El Niño oscillation and volcanic eruptions. But they maintain that this removal usually involves smoothing out temperature fluctuations, which means that brief, but non-cyclical, changes are also erased.

Too much smoothing?

Team member John Wallace of the University of Washington in Seattle likens this to a simplification of stock market data. "To see the longer-term market trends it is better to smooth out much of the daily ups and downs," he says. "But if you smooth the data too much you do not see sudden changes such as stock market crashes."

The researchers devised a new data analysis technique that they say gets rid of the unwanted cyclical phenomena while preserving any rapid fluctuations that contribute to longer-term change. They found that the sea-surface temperatures in the northern hemisphere did not decrease gradually in the decades following the Second World War, as would be expected if aerosols were principally responsible for the northern hemisphere cooling, but that temperatures fell very rapidly – by about 0.3 °C between 1968 and 1972.

The team has found this sudden drop both in the temperature data of the northern hemisphere in isolation and in the difference between the northern and southern temperatures. It also established that it could not be a mere artefact of the measurement process, as it says is the case for a sudden drop in global sea temperatures recorded immediately after the Second World War, which they attribute to a change in the relative numbers of British and American ships at sea. The rapid dip around 1970, it says, is present in all available historical data sets of sea-surface temperature and cannot be linked to any known biases in temperature measurements.

Outpouring of cold, fresh water

As to what caused this sudden drop in temperature, Wallace says that it might have been partly due to an outpouring of relatively cold, fresh water from the Arctic into the Atlantic that was known to have taken place about the same time. But he is reluctant to attribute all of the cooling to this process. He and his colleagues found that most of the cooling took place in the Atlantic but that some also occurred in the Pacific, which would not have received any of the fresh water.

Dan Hodson of the University of Reading in the UK points out that some climate models also predict rapid hemispheric cooling as a result of fresh water being deposited in the north Atlantic. He describes the latest research, which he was not involved in, as "another piece of the puzzle in the ongoing effort to decipher the 20th century climate record" and says that it will "improve our understanding of the underlying detailed mechanisms of climate change". And he warns that "we should be especially vigilant" if Arctic fresh water can indeed cause rapid ocean cooling. He says that the freshwater released from a rapidly melting Arctic might mitigate some of the impacts of global warming but points out that a very rapid melting could have widespread negative effects such as reducing crop yields.

However, Michael Mann of Pennsylvania State University in the US is not convinced. He believes that some scientists have overestimated the significance of multidecadal ocean oscillations on global temperatures but maintains that these oscillations can explain the rapid temperature drop around 1970 identified by Thompson's team, adding that fairly abrupt temperature changes earlier in the 20th century can also be explained purely in terms of such oscillations.

The work is described in Nature.