| 研究実績の概要 |
This international collaborative project aimed to develop a data-driven inference framework to predict flood and drought events at lead times ranging up to 6 months to the present. The project involved a partnership between the University of Tokyo (U-Tokyo) in Japan and the National Aeronautics and Space Administration (NASA) in the United States. The NASA team contributed in-depth knowledge and expertise on satellite observations, specifically elucidating the memory impact of local water storage, such as terrestrial water storage and river water height as a delayed local response. The U-Tokyo team focused on elucidating the teleconnection mechanisms between global-scale forcings. They simulated these global and local lagged relations using a physics-informed deep learning approach developed by the U-Tokyo team. Newly proposed deep learning approach proved effective in simulating the complex relationships between global forcings and local hydrology. During the project period, the team faced two unexpected and severe situations: 1) the COVID-19 pandemic, which disrupted international collaboration, and 2) the relocation of the international counterpart from NASA to the University of Saskatchewan, which required adjustments in communication and coordination. Nenvertheless, the project successfully developed a data-driven framework to predict flood and drought. The framework leverages satellite observations, advanced modeling techniques, and international collaboration to provide valuable insights for disaster risk management and water resource planning.
|
