Boston, Mass. -- The ability to predict drought one to several seasons in advance, may save water resource planners and farmers billions of dollars, according to a team of Penn State researchers.
These researchers developed a computer model that can predict reliably the severity and timing of drought episodes six months in the future.
"Farmers would find it useful to have drought predictions in the spring for three or six months later," says Kelly Brennan, recent recipient of a Penn State master's degree in civil engineering. "Farmers could then change the crop they plant for something that is more late-summer drought tolerant."
One suggestion would be to plant crops with deeper root systems.
"Another important area, especially in the Ohio River Basin where we worked, is the need to maintain adequate river levels for barge traffic," Brennan told attendees at the spring meeting of the American Geophysical Union today (May 29) in Boston.
The normal practice is to lower reservoirs in winter and early spring to create enough storage capacity to retain excessive spring runoff that would otherwise cause flooding downstream. This practice can have very harmful consequences in years of drought, especially during the summer.
"A grounded barge can cost as much as $10,000 a day," says Brennan. "Low water levels may also affect water purity and supply and recreation."
Working with Dr. Ana Barros, assistant professor of civil engineering, Brennan developed a model that can predict extreme drought episodes three, six and nine months in advance.
"The model uses a measure of the temporal evolution of the spatial variability of precipitation over a period of time in the past," says Brennan.
The researchers combine this measure with the accepted measure of current drought, the Palmer Drought Severity Index. The PDSI uses soil moisture, temperature and precipitation along with specific parameters for the region to detect if there is a current drought. The PDSI is used to issue drought emergencies and to rescind them as well.
The model was tested and validated on historic data through 1996.
"We can predict drought three to six months in advance pretty well," says Brennan. "At the nine-month scale we can capture the timing of extreme droughts, but the severity of the drought is not as accurate."
The researchers would like to see their model running on current data so that it could be used as a real-time forecast model.
Lisa Mead, a biology undergraduate and a participant in the Women in Science and Engineering Internship program, also participated in this research.
EDITORS: Ms. Brennan may be reached at (814) 865-2342 or kelly@essc.psu.edu by email. Dr. Barros may be reached at (814) 863-8609 or barros@essc.psu.edu by email.