First author

In 2000, researchers predicted that warming equatorial oceans would cause permanent El Niño conditions, which would shift tropical rainfall patterns away from the Amazon rainforest. In 2005, the rainforest did experience severe drought, but El Niño was not to blame. On page 212, climate modeller Peter Cox from the University of Exeter and his colleagues reveal that in recent years an interplay between two types of pollutant — sulphate aerosols and greenhouse gases — balanced global climate. But, Cox tells Nature, if aerosol pollution continues to decline without concomitant carbon dioxide abatement, the Amazon may not survive another 50 years.

You made an early career switch from physicist to climate modeller. Why?

In the 1980s, I worked on nuclear fusion in the effort to find a replacement energy source for declining coal and oil reserves. But as we began to realize that burning existing fossil fuels damaged the environment, I became interested in the more pressing issue of predicting future climate change.

Why did you initially include aerosols in climate models?

You can't reproduce the last century's observed climate warming, including the brief mid-century period of global cooling, without taking aerosols and greenhouse gases into account. Sulphate aerosols cause cooling both by reflecting sunlight themselves and by making clouds brighter, and so more reflective. But for the past couple of decades, the combined effect of reduced aerosol pollution and increasing greenhouse gases has led to accelerated global warming.

Are you suggesting that aerosol pollution could stave off climate change?

No. To improve air quality we need to reduce sulphur emissions. The key is to reduce CO2 emissions. Our model suggests that the Amazon rainforest won't dry significantly if we can keep atmospheric CO2 levels below about 500 parts per million. Rather perversely, one way to slow CO2 increases is to reduce the rate of deforestation in Amazonia.

What might the future of climate modelling hold?

Until now, we've focused on 'what if' scenarios — cautionary tales based on equally possible climate eventualities. Soon, we will be able to predict what will happen locally in the near term. I think climate projections will become more like weather forecasting, but instead of five days ahead, we'll predict the next 20–50 years.