Predicting monsoons gets easier
Timing of Indian droughts yields to better climate analysis.
It is notoriously difficult to predict failures in the all-important Indian monsoon, on which the country's agriculture depends. Now researchers think they have made this task a little easier.
Drought years have been specifically linked to a warming of the western side of the Pacific during an El Nio event. The link makes prediction easier, but also brings bad news: climate trends suggest that this kind of warming will become more common in the future.
More than 130 years of Indian rainfall records show that droughts are always related to El Nio events, which feature a warm tropical Pacific. This causes changes in evaporation from the seas and wind patterns that generally result in reduced summer rainfall over India. So a measure of sea surface temperatures in the Pacific has been used to issue drought warnings.
But that no longer seems to be good enough. Over the past 30 years, say researchers publishing in Science1, a range of monsoon rains from strong to weak have accompanied sea temperature warnings. In 1997, the twentieth century's strongest El Niño didn't greatly affect Indian monsoon rainfall, whereas in 2002 a moderate El Nio resulted in a severe and unexpected Indian drought.
"The skill of the current monsoon forecast is low," says co-author Martin Hoerling of the US's National Oceanic and Atmospheric Administration in Boulder, Colorado. "It fails to explain more than 90% of monsoon variability".
What's your flavour?
The reason, Hoerling and his team say, is that El Nio events come in different 'flavours': some heat the water more on the western/central side of the Pacific, others more on the eastern side.
The team's analysis of recent monsoons shows that it is the western-shifted warming that predominantly causes the Indian drought. If researchers had looked to this variable in the past, they say, it would have been possible to predict major unexpected droughts that struck in 2002 and 2004.
This flavour of El Nio is becoming more common as mankind's activities cause the oceans to warm, they note. That could spell bad news for India, as droughts may become ever more frequent.
There is already a trend towards lessening monsoon rains in India, says Hoerling, perhaps because of this shift in El Nio flavours. But the future is uncertain. As sea temperatures change, dominant El Nio patterns, and their effects, may also alter.
The new way of predicting droughts certainly isn't perfect, however. "We're only looking for skill in predicting the devastating droughts, which may occur one year out of ten," Hoerling points out. The technique isn't good enough to predict smaller events with accuracy.
Part of the problem is that the Indian monsoon system is very noisy, with many variables causing rainfall averages to leap about from year to year. Aerosols including soot, for example, can block sunlight over industrially polluted areas such as Southeast Asia, which in turn cuts down on water evaporation and rainfall. And there is a long time lag between the El Nio, which peaks in December through February, and the Asian summer rains in May through August.
In the past, India's Meteorological Department has attempted to keep a firm line on monsoon forecasts in order to help coordinate responses to predicted weather conditions. Last year, they tried to prevent scientists from publishing their own monsoon predictions if they disagreed with the official line. One of those affected was Prashant Goswami, who works on monsoon prediction at the Centre for Mathematical Modelling and Computer Simulation (C-MMACS) in Bangalore, India.
Goswami points out that the scope of monsoon forecasts not just their accuracy is also important. At the moment, a monsoon prediction is made in terms of rainfall across the whole of India. "If there is severe flood in Gujarat, and severe drought in Andhra Pradesh, is it right and humane to average them to arrive at a comfortable figure?" Goswami asks.
Visit our newsblog to read and post comments about this story.
- Krishna Kumar K., et al. Science doi:10.1126/science.1131152 (2006)10.1126/science.1131152 (2006).