New Scientist, August 13, 2008

WHAT's going to happen to the climate over the next 10 years or so? Is it time to buy that air conditioner you considered during the last heatwave? Should you rip up your garden and replant it with drought-resistant plants, or can you expect more rain -- perhaps even floods -- in your part of world? The other possibility, of course, is that your local climate will change little in the near future.

On the one hand we have weather predictions for the next few days. On the other we have climate forecasts for the very distant future. But what happens in the middle? Why don't we have forecasts for, say, 2010 or 2018? Knowing how temperature and rainfall will change over the next few years would be invaluable to many people, from farmers to the tourism industry to those in charge of our water supplies. Yet while you might think predicting how the climate will change over the next few years would be a lot easier than saying what it will be like in 2030 or 2050, it's actually harder.

Nevertheless, some meteorologists and climate scientists are now trying to make just these kinds of forecasts. It is a new and controversial field, but over the past year some groups have published the first short-term forecasts. So what are they predicting -- and can we trust their conclusions?

Underlying trends

For long-term forecasting, what matters is underlying trends, and at the moment the key trend is warming due to rising levels of greenhouse gases. Predictions made two decades ago are pretty close to the mark. In the short term, though, natural variability matters more than the underlying trend -- global warming does not mean that each year will be warmer than the preceding one.

The problem is a bit like trying to predict how the weather in New York will change over January compared with how the weather will change from January to July. It's hard to say whether the last week of January will be colder than the first, but you can confidently predict that it will be colder during January than in July.

So making forecasts is all about figuring what dominates the state of the atmosphere on various timescales. Some things, like accumulating greenhouse gases, matter over many decades while other things, like warm and cold fronts, dominate over days and months. Over periods of a few years, there's growing evidence that the oceans are the key -- and this is encouraging researchers to attempt short-term forecasts.

Ocean oscillations

"It takes the oceans a long time to heat up and cool down," says Doug Smith, who runs 10-year forecasting trials at the Met Office Hadley Centre in Exeter, Devon, the UK's official centre for climate change research. "That makes it a lot easier to predict than the atmosphere. We now think we can predict the key ocean fluctuations 10 to 20 years ahead."

The oceans are crucial because they store so much heat. It takes more than 1000 times as much energy to heat a cubic metre of water by 1 °C as it does the same volume of air. Globally, this means that if the oceans transfer just a tiny fraction of their heat energy to the lower atmosphere, there can be a big rise in surface air temperatures. Conversely, if the oceans soak up more heat from the atmosphere, there can be surface cooling.

Most of the natural variability in surface air temperature from year to year is due to heat sloshing back and forth between the oceans and atmosphere, rather than any overall loss or gain of heat by the entire planet. The state of the sea surface determines what happens, and it affects both temperature and rainfall. For instance, in recent years, climate scientists have successfully forecast droughts in west Africa and northeast Brazil several months ahead by measuring sea surface temperatures in the tropical Atlantic.

Huge influence

The long-standing droughts in Australia could be due to persistently low sea surface temperatures to the north of country, relative to warmer water in the Indian Ocean, say Wenju Cai and Tim Cowan of the CSIRO marine and atmospheric research centre in Aspendale, Victoria. Rainfall over southern and eastern Australia has been declining for half a century now, causing major problems in river systems like the Murray-Darling basin, which produces much of the country's crops.

The huge influence of sea surface temperatures has led many researchers to try to understand how they fluctuate over years and decades. Predict this, and you should be able to predict all sorts of other things as well.

The best known of these fluctuations is the El Niño-Southern Oscillation. In the tropical Pacific, cold water normally wells up near South America, while hot water piles up on the other side of the Pacific. Sometimes, for reasons poorly understood, the hot water spreads right across the surface of the Pacific in a shallow layer.

This increase in the area of warm water boosts the transfer of heat and moisture to the atmosphere, changing air circulation patterns and producing widespread consequences -- from droughts in Indonesia to floods in the Americas. An especially intense El Niño in 1998 made it one of the warmest years on record.

El Niño

Yet contrary to what you might expect, if El Niño has any long-term effect on global warming it may be to slow it down. Models suggest that by increasing heat radiation into space, it may reduce the net gain of heat by the entire planet as a result of increasing greenhouse gases. Actual measurements are not yet accurate enough to confirm this, however.

The opposite of El Niño, La Niña, results in a heat transfer from the lower atmosphere to the ocean. The strong La Niña during the early part of this year could make 2008 one of the coldest years since the early 1990s.

The El Niño cycle lasts anywhere from three to eight years, but its state cannot be predicted more than a year in advance. However, in the Pacific north of the tropics, there seems to be something akin to El Niño that lasts far longer. It is called the Pacific Decadal Oscillation and its influence is extensive, on land as well as at sea (see "Trendsetters").

The PDO was in a negative phase -- with cooler sea surface temperatures -- between the mid-1940s and the mid-1970s, and may have been partly responsible for the cooler global surface temperatures during this time. This does not mean a negative PDO has a cooling effect overall; on the contrary, it's likely that a negative PDO increases the planet's total heat by reducing heat transfer from the oceans to the atmosphere and thence into space. Since 1976, the PDO has been mostly positive again, which may have contributed to the strong warming in Alaska and the prolonged droughts in south-east Australia.

Droughts and monsoons

The other major ocean fluctuation is the Atlantic Multidecadal Oscillation, a semi-periodical change in surface temperatures north of the equator. Most researchers agree that the AMO is largely due to changes in the speed of the deep-ocean current known as the thermohaline circulation.

When the circulation speeds up, more warm water from the tropics moves up into the North Atlantic, transferring huge amounts of heat to the air as it goes -- the positive phase of the AMO. When a slowing circulation pushes the AMO into a negative phase, more warm water stays in the tropics, and surface temperatures fall in Europe and the eastern side of North America.

The AMO was in a negative phase from the late 1960s until the mid-1990s. We're now in the middle of a positive phase again, which may have contributed to the very rapid warming in the Arctic in recent years and the dramatic fall in the extent of its sea ice during the summer. The effect of a positive AMO on the planet's overall heat budget is not clear, but it may speed up global warming since less ice cover means less solar radiation is reflected back into space.

What's more, there is growing evidence linking the AMO to climatic trends on land, even in areas far from the Atlantic. Decades-long fluctuations in the intensity of the Indian monsoon rains, droughts in the region of west Africa called the Sahel and even the numbers of Atlantic hurricanes all seem to depend on the AMO. Droughts in the western US, including the 1930s Dust Bowl and low river levels in the 1990s, all happened during its positive phase.

The big picture

So what does the future hold? If the AMO stays positive in the coming decade, it will increase summer rainfall over India and the Sahel -- and increase Atlantic hurricane activity. However, the AMO may be poised to turn negative, says Rowan Sutton of the Walker Institute at Reading University, UK, who has studied the phenomenon in detail.

The thermohaline circulation is driven by the sinking of cold, salty water near the Arctic. Its strength, and thus the phase of the AMO, seems to depend on what happens in the waters between Greenland and Scandinavia. "There is evidence that we can sometimes predict the changes up to 10 years ahead," Sutton says.

Meanwhile, the PDO has already been negative for the past couple of years. If both ocean fluctuations were to be in a negative phase over the next few years, things will be very different.

For starters, there will be a slowdown in the rapid warming seen around the Arctic and North Atlantic in recent years. The rapid fall in the extent of sea ice in summer -- which has been happening much faster than predicted -- could slow and perhaps even reverse.

A temporary respite

Droughts could return to India and the Sahel, but for the parched American west there could be a desperately needed respite. On current trends, the great reservoirs on the Colorado river that sustain western cities like San Diego and Phoenix could be dry within a decade. "We are stunned at the magnitude of the problem, and how fast it is coming at us," says Tim Barnett of the Scripps Institution of Oceanography in San Diego. If the AMO enters a negative phase, then the river may live on.

Some also predict a decline in hurricane activity in the Atlantic. But Michael Mann of Penn State University says some things attributed to the AMO are more likely a result of global warming. He thinks the AMO has little influence on tropical sea temperatures, so he predicts that Atlantic hurricanes will intensify even if the AMO is negative. We will have to wait and see.

That's the broad-brush picture. In Europe, several groups are trying to model exactly what might happen over the next 10 years or so. Smith's team produced the first such forecast last year. It suggests that surface air temperatures will remain steady for the next six years or so as cooler sea surface temperatures keep the lower atmosphere cool despite ever higher greenhouse gas levels.

But this respite won't last. Smith expects surface temperatures will start to rise again by 2014, and that they will go into overdrive with a string of record highs at the end of the next decade if both the major ocean oscillations kick back into positive phases.

Global cooling

Earlier this year another group, headed by Noel Keenlyside of the Leibniz Institute for Marine Sciences in Kiel, Germany, produced an even more striking forecast: "Global surface temperature may not increase over the next decade."

Climate change deniers promptly proclaimed that we could expect "more global cooling ahead". But surface temperature is not the same as the overall heat content of the planet. Since the 1960s, 90% of the excess heat due to higher greenhouse gas levels has gone into the oceans, 7% into land and ice, and just 3% into warming the atmosphere. Even if the lower atmosphere doesn't warm in the next few years, that's no reason for comfort so long as the strong warming trend in the oceans continues. In the long run, warmer oceans inevitably mean a warmer atmosphere.

In any case, many climate researchers don't think the Keenlyside forecast is right. "The way they try to predict the AMO is almost guaranteed to give you the wrong answer," says Gavin Schmidt of NASA's Goddard Center for Space Studies in New York.

The issue is where the models start from. If you're trying to forecast natural variability, conditions in the seas in the model must match those in the real world. Thanks to a network of undersea sensors, we are now starting to get good data on both temperature and salinity levels in the crucial upper layers of the Atlantic. However, feeding this information into models is not as easy as it might seem and -- unlike Smith's team -- the Keenlyside team put only sea surface temperatures into their model.

Completely wrong

"If you match the sea surface temperatures but not the salinity values, the water density will be completely wrong," says Schmidt. Yet this density is crucial for determining the state of the thermohaline circulation and hence the AMO.

Some of the scientists who write for the RealClimate blog are so sure that the forecast is wrong that they offered the Keenlyside team a bet of €2500 that the average surface temperatures for 2005 to 2015 will, contrary to the team's forecast, turn out be higher than during 1994 to 2004. The team has not accepted the bet.

Even if the various "model initialization" problems can be solved, is it really possible to predict how the oceans will behave so far in advance? According to David Battisti at the University of Washington in Seattle, who specializes in studying natural variability, there's a growing consensus that the PDO is just the mid-latitude "debris" left by the past two or three El Niños or La Niñas. If this is right, it means the PDO cannot be predicted long in advance.

No predictability

"There is no predictability in the Pacific," he says. "If there's any hope for predicting natural variability, it's in the Atlantic." Even there, Battisti thinks it will only be possible to make accurate decadal forecasts for tropical regions where there is far less variability from year to year than in higher latitudes.

What's more, there is another aspect of natural variability that cannot be predicted years in advance: volcanoes. Major volcanic eruptions like those of El Chichón in 1982 and Pinatubo in 1991 can throw up so much dust and sulphur that they cool the entire planet. The average effect of eruptions can be included in forecasts looking several decades ahead, but not in 10-year ones. "The forecast will be wrong when one occurs," says Smith.

There is no doubt that enormous progress has been made in understanding and predicting some of the factors responsible for the tremendous variability in surface air temperatures and rainfall from year to year. But the field is in its infancy. It is still far from clear to what extent natural variability can be predicted years in advance -- and it's going to take decades to find out for sure.

In the meantime, 10-year forecasts should come with a very clear health warning. "There is a danger that if we make a forecast and it's wrong that people will lose faith," Smith acknowledges. "But I don't agree we shouldn't make them."

Climate Change – Want to know more about global warming: the science, impacts and political debate? Visit our continually updated special report.