SYDNEY, Oct. 13 (Xinhua) -- An international research team has developed artificial intelligence (AI) models that could improve long-term flood forecasting, helping communities get warnings weeks ahead of floods.
Researchers used two decades of historical precipitation and streamflow data from the DuPage River basin in Illinois, the United States, to develop deep learning models incorporating attention mechanisms to identify patterns in rainfall and river flow, according to a statement released Tuesday by Australia's University of Technology Sydney (UTS).
By combining the models' predictions, the team improved forecasting accuracy, with the ensemble deep learning models reducing prediction errors by about 38 percent and the proposed fusion framework improving predictive performance by about 47 percent, according to the study published in the Journal of Hydrologic Engineering.
The approach remained highly accurate for forecasts up to 12 weeks ahead, said UTS Professor Amir H. Gandomi, the study's co-author, stressing that additional warning time could help communities prepare for floods.
The findings could have implications for flood risk management, including disaster preparedness, infrastructure resilience and water resource management, according to the team.
"Ultimately, the goal is not simply to predict floods more accurately, but to turn better predictions into better decisions," Gandomi said. ■
