Improving AI system awareness of geoscience knowledge: Symbiotic integration of physical approaches and deep learning

Department of Hydrology and Atmospheric Sciences

4 pm on Thursday, January 21, 2021
Contact the department for zoom details or to subscribe to the seminar email list

Yi Zheng
School of Environmental Science and Engineering
Southern University of Science and Technology, China


Recent advances in artificial intelligence (AI) provide unprecedented opportunities for data-driven hydrological modeling. With the amazing predicting power of deep learning (DL) demonstrated in the field of hydrology, whether and how theory can still play an important role in hydrological modeling is now questioned. This presentation introduces a recent study to fuse DL with hydrology theory in rainfall-runoff simulation. In this study, a novel DL framework was developed which contains a special recurrent neural layer to “memorize” physical rules behind system dynamics. Following this framework, a conceptual hydrologic model was encoded into a DL structure, leading to a hydrology-aware DL model. Model applications in 569 catchments across the conterminous United States show that this hybrid model has enhanced prediction accuracy, robust transferability, and good intelligence for inferring unobserved processes. This presentation stresses that physics-AI integration is a promising direction for hydrological modeling in the era of big data. This presentation also discusses some challenges faced by DL-based hydrological modeling, such as interpretability of DL models and influences of human activities, using ongoing studies in Dr. Zheng’s group as illustrative cases.


Prof. Yi Zheng received his Ph.D. from the University of California, Santa Barbara (2007). He is Associate Dean of the School of Environmental Science and Engineering at Southern University of Science and Technology (SUSTech), China. Before he joined SUSTech in 2016, he was an Associate Professor at Peking University. His current research interests include hydrology and water resources, environmental system modeling, and environmental big data. His major scientific contributions cover integrated ecohydrological modeling, uncertainty analysis for complex environmental models, human-water nexus, and artificial intelligence for hydrology. He has published over 90 peer-reviewed papers (including 7 ESI highly cited paper), nearly all in top-tier journals of Earth and environmental sciences, such as Environmental Science & Technology, Geophysical Research Letters, Water Resources Research, Remote Sensing of Environment, Water Research, and Environmental Modelling & Software. He currently serves as an Associate Editor for Water Resources Research and Journal of Hydrologic Engineering-ASCE. He received the Excellent Young Investigator Award from the National Natural Science Foundation of China in 2016 and the Outstanding Research Award from the China Society of Natural Resources in 2019.