Multiscale Hydrology, Processes and Intelligence Group, Penn State University
Redefining natural disaster and water predictions with AI.
Hydrologic time series forecasting is is a difficult problem but deep learning could help to make predictions more accurate, with less effort. The Multiscale Hydrology, Processes and Intelligence group, led by Dr. Chaopeng Shen, at Penn State Univ, is developing deep learning tools for hydrological variables like soil moisture and streamflow, with critical applications. Predictions from their tools have helped with locust plague control efforts in Eastern Africa and enable very efficient and accurate global flood forecasting. AI is not a black box: It can be used to reveal new ways of asking questions and understanding hydrology.
The group is also working on a landslide prediction tool called deepLDB, supported by the Google AI Impact Challenge, involving training an AI system to recognize and catalogue landslides around the globe. They need help with data for training and verifying the model – learn more at www.MHPI.info.