The new technique produced 20- to 25-day forecasts of rainfall in the 1-million-square-kilometer Ganges Valley of Bangladesh during the summer of 2002. The forecast closely mirrored actual precipitation for the season, according to U.S. State Department-funded research led by Professor Peter Webster and his students in Georgia Tech's School of Earth and Atmospheric Sciences.
In the future, such forecasts could guide farmers in choosing optimal planting times and making other decisions, such as better water management, that affect crop production, Webster said.
He presented his findings February 17 at the 2003 annual meeting of the American Association for the Advancement of Science (AAAS) in Denver.
"Forecasting weather a few days in advance is not particularly useful for agriculture," Webster said. "What is needed is a 20- to 25-day forecast...We are able to do that with our new method. We could have predicted the month-long break in the monsoon rains that lasted from the end of June to early July, and which caused a $6 billion loss in crops in the Ganges Valley. If farmers had this forecast last spring, they could have changed agricultural practices, such as delaying planting."
Webster's forecasting method is applicable to the rainy seasons of any monsoon region and adjusts for precipitation changes related to temporary climatic events such as El Nino and La Nina. Last year was an El Nino year, and, as expected, it resulted in decreased rainfall on the Indian subcontinent.