Abstract & Bio
The hydrology community has made a conscious and well-intentioned attempt to distinguish hydrology as a science from hydrology as a branch of engineering. I will argue in this talk that the unprecedented power and versatility of modern machine learning (deep learning in particular) poses a threat to that distinction. It is uncontroversial that hydrologists have been unsuccessful at the most fundamental task of any branch of science: developing scale-relevant theories (in our case, of watersheds). My more hyperbolic argument is that due to this systemic failure, and also to the fact that deep learning models are so much more powerful than traditional hydrology models (both calibrated conceptual and process based), it is difficult to see a clear role for hydrological science to help inform the societally-relevant aspects of what hydrologists are tasked to do going forward into the future. I see it as our job as research hydrologists to clearly delineate where and when hydrological understanding is valuable in the age of machine learning.
Dr. Grey Nearing is Research Director at Upstream Tech, Public Benefit Corporation, and also an Assistant Professor at the University of Alabama in the Department of Geological Sciences. His PhD is in Hydrology and Water Resources from the University of Arizona (2013). Grey previously worked as a project scientist in the Hydrological Sciences Lab at NASA Goddard Space Flight Center, and at the US National Center for Atmospheric Research (NCAR). At NASA he was part of the Land Information System (LIS) team and the Soil Moisture Active Passive (SMAP) satellite mission team. Grey was also previously a Research Assistant Professor at the University of Maryland Department of Computer Science and Electrical Engineering, where he worked on applications of quantum computing to Earth science modeling and remote sensing applications. Grey’s PhD dissertation and postdoc research were on applications of information theory and machine learning for data assimilation and uncertainty quantification. His current research focuses on physics-informed machine learning for hydrological applications.