Abstract
Although decades of effort have been devoted to building Physical-Conceptual (PC) models for predicting the time-series evolution of geoscientific systems, recent work shows that Machine Learning (ML) based Gated Recurrent Neural Network technology can be used to develop models that are much more accurate. However, the difficulty of extracting physical understanding from ML-based models complicates their utility for enhancing scientific knowledge regarding system structure and function. Here, we propose a physically interpretable Mass-Conserving-Perceptron (MCP) as a way to bridge the gap between PC-based and ML-based modeling approaches. The MCP exploits the inherent isomorphism between the directed graph structures underlying both PC models and GRNNs to explicitly represent the mass-conserving nature of physical processes while enabling the functional nature of such processes to be directly learned (in an interpretable manner) from available data using off-the-shelf ML technology. As a proof of concept, we investigate the functional expressivity (capacity) of the MCP, explore its ability to parsimoniously represent the rainfall-runoff (RR) dynamics of the Leaf River Basin, and demonstrate its utility for scientific hypothesis testing. To conclude, we discuss extensions of the concept to enable ML-based physical-conceptual representation of the coupled nature of mass-energy-information flows through geoscientific systems.
| Original language | English |
|---|---|
| Article number | e2023WR036461 |
| Journal | Water Resources Research |
| Volume | 60 |
| Issue number | 4 |
| DOIs | |
| State | Published - Apr 2024 |
Funding
The authors would like to thank the WRR editorial team including Marc Bierkens (Editor), Juliane Mai (Associate Editor), and reviewers Dr. Yang Yang and Dr. Erwin Zehe plus one anonymous reviewer for taking the time to provide incredibly constructive comments. The first author (YHW) would like to thank the late Thomas Meixner, as well as Jennifer Mcintosh, Martha Whitaker, Eyad Atallah, Dale Ward, Ty Ferré, Jim Yeh, and Chris Castro, for their support, and acknowledge the TA and outreach assistantship support provided by the Department of Hydrology and Atmospheric Sciences and the University of Arizona Data Science Institute during the final 2 years of his Ph.D. study, which made the finalization of this work possible. We also thank University of Arizona Data Science Institute for providing HPC computation resources. YHW is an employee of UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US DOE. Accordingly, the US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so, for US Government purposes. The second author (HVG) acknowledges partial support by the Australian Centre of Excellence for Climate System Science (CE110001028), the inspiration and encouragement provided by members of the Information Theory in the Geosciences group (geoinfotheory.org), and support for a 4-month research visit to the Karlsruhe Institute of Technology, Germany provided by the KIT International Excellence Fellowship Award program. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US Government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so, for the US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan). The authors would like to thank the WRR editorial team including Marc Bierkens (Editor), Juliane Mai (Associate Editor), and reviewers and plus one anonymous reviewer for taking the time to provide incredibly constructive comments. The first author (YHW) would like to thank the late , as well as , and , for their support, and acknowledge the TA and outreach assistantship support provided by and the during the final 2 years of his Ph.D. study, which made the finalization of this work possible. We also thank for providing HPC computation resources. YHW is an employee of UT‐Battelle, LLC, under contract DE‐AC05‐00OR22725 with the US DOE. Accordingly, the US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid‐up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so, for US Government purposes. The second author (HVG) acknowledges partial support by the (CE110001028), the inspiration and encouragement provided by members of the group ( geoinfotheory.org ), and support for a 4‐month research visit to the Germany provided by the program. This manuscript has been authored by UT‐Battelle, LLC, under contract DE‐AC05‐00OR22725 with the US Department of Energy (DOE). The US Government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid‐up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so, for the US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( http://energy.gov/downloads/doe‐public‐access‐plan ). Dr. Yang Yang Dr. Erwin Zehe Thomas Meixner Jennifer Mcintosh, Martha Whitaker, Eyad Atallah, Dale Ward Ty Ferré, Jim Yeh, Chris Castro the Department of Hydrology and Atmospheric Sciences University of Arizona Data Science Institute University of Arizona Data Science Institute Australian Centre of Excellence for Climate System Science Information Theory in the Geosciences Karlsruhe Institute of Technology, KIT International Excellence Fellowship Award
Keywords
- catchment-scale rainfall-runoff (catchment-scale RR)
- gated recurrent neural network (gated RNN)
- information flow
- mass-conserving-perceptron (MCP)
- physically-interpretable