To track rapid changes within our water sector, Global Water Models (GWMs) need to realistically represent hydrologic systems' response patterns - such as baseflow fraction - but are hindered by their limited ability to learn from data. Here we introduce a high-resolution physics-embedded, differentiable, big-data-trained model as a breakthrough in reliably capturing characteristic hydrologic response patterns ('signatures') and their shifts. Such models allow process-based equations to be seamlessly trained together with neural networks, https://t.co/qyuAzYPA6Y) and offer highly competitive performance, physical process clarity, and the ability to learn interpretable knowledge. By realistically representing the long-term water balance, the model revealed widespread shifts - up to ~20% over 20 years - in fundamental green-blue-water partitioning, baseflow ratios and streamflow elasticities worldwide. Shifts in these response patterns, previously considered static, contributed to increasing flood risks in northern mid-latitudes, heightening water supply stresses in southern subtropical regions, and declining freshwater inputs to many European estuaries, all with ecological implications. Furthermore, we show that AI and differentiable approaches are applicable to solving high-resolution environmental partial differential equations (PDEs), and our domain insights can bring unique contributions to the general AI community. We further consider training highly efficient surrogate models for these models (mostly parameterized ordinary or partial differential equations) with the correct sensitivity. AI-infused solutions to PDEs like the Fourier Neural Operator (FNO) are unbelievably efficient, but often suffer from incorrect learned relationships and sensitivities, resulting in large errors in inversion and optimization tasks. We propose Sensitivity-Constrained Fourier Neural Operators (SC-FNO, presented in the general AI Conference ICLR 2025), which shows robustness even under sparse training data or concept drift scenarios. Differentiable modeling together with sensitivity-constrained neural operators are posed to drastically improve our simulation and learning capabilities for a wide range of engineering and geoscientific problems.
Bio
Chaopeng Shen is Professor in Civil Engineering at The Pennsylvania State University. He received the Ph.D. degree in environmental engineering from Michigan State University, East Lansing, MI, USA, in 2009. His PhD research focused on computational hydrology and he developed the hydrologic model Process-based Adaptive Watershed Simulator(PAWS), which was later coupled to the community land model to study the interactions between hydrology and ecosystem. He was a Post-Doctoral Research Associate with the Lawrence Berkeley National Laboratory, Berkeley, CA, USA, from 2011 to 2012, working on high-performance computational geophysics. His recent efforts focused on harnessing the big data and machine learning (ML) and physics-informed ML opportunities in advancing hydrologic predictions and understanding. As an early advocate for ML in geosciences, he has written technical, editorial, review and collective opinion papers on hydrologic deep learning to call to attention the emerging opportunities for scientific advances. He currently promotes differentiable modeling which seamlessly integrates neural networks and physics for knowledge discovery. In addition, his research interests also include floodplain systems, scaling issues, process-based hydrologic modeling, and hydrologic data mining. He is currently Editor of Journal of Geophysical Research - Machine Learning & Computation, an Editor of the Water Resources Research, and Chief Special Editor for Frontiers in AI: Water and AI.
website: https://water.engr.psu.edu/shen/