Researchers Develop Deep-Learning Model for Streamflow, Flood Forecasting
Streamflow and flood forecasting remains one of the long-standing challenges in hydrology. Traditional physically based models are hampered by sparse parameters and complex calibration procedures particularly in ungauged catchments, according to a recent research article published in the journal Innovation.
More than 95 percent of small and medium-sized water catchments in the world lack monitoring data, according to the Chinese Academy of Sciences (CAS).
Researchers from the Institute of Mountain Hazards and Environment of the CAS used the datasets of more than 2,000 catchments around the world to conduct model training in order to cope with streamflow forecasting at a global scale for all gauged and ungauged catchments.
The distribution of these catchments was significantly different, ensuring the diversity of data.
The results show that the forecasting accuracy of the model was higher than traditional hydrological models and other AI models.
The study demonstrated the potential of deep-learning methods to overcome the lack of hydrologic data and deficiencies in physical model structure and parameterization, the research article noted.
4155/v
