[论文解读] Physics-aware Machine Learning Revolutionizes Scientific Paradigm for Machine Learning and Process-based Hydrology

Qingsong Xu, Yilei Shi+4 more|arXiv (Cornell University)|Oct 8, 2023

Hydrological Forecasting Using AI被引用 15

一句话总结

This paper conceptualizes and reviews physics-aware ML (PaML) for hydrology, categorizing methods (physical data-guided ML, physics-informed ML, physics-embedded ML, and physics-aware hybrid learning), and introducing HydroPML as an open-source platform to bridge ML with process-based hydrology.

ABSTRACT

Accurate hydrological understanding and water cycle prediction are crucial for addressing scientific and societal challenges associated with the management of water resources, particularly under the dynamic influence of anthropogenic climate change. Existing reviews predominantly concentrate on the development of machine learning (ML) in this field, yet there is a clear distinction between hydrology and ML as separate paradigms. Here, we introduce physics-aware ML as a transformative approach to overcome the perceived barrier and revolutionize both fields. Specifically, we present a comprehensive review of the physics-aware ML methods, building a structured community (PaML) of existing methodologies that integrate prior physical knowledge or physics-based modeling into ML. We systematically analyze these PaML methodologies with respect to four aspects: physical data-guided ML, physics-informed ML, physics-embedded ML, and physics-aware hybrid learning. PaML facilitates ML-aided hypotheses, accelerating insights from big data and fostering scientific discoveries. We first conduct a systematic review of hydrology in PaML, including rainfall-runoff hydrological processes and hydrodynamic processes, and highlight the most promising and challenging directions for different objectives and PaML methods. Finally, a new PaML-based hydrology platform, termed HydroPML, is released as a foundation for hydrological applications. HydroPML enhances the explainability and causality of ML and lays the groundwork for the digital water cycle's realization. The HydroPML platform is publicly available at https://hydropml.github.io/.

研究动机与目标

• 通过构建 PaML 方法学来缩小基于物理的水文学与机器学习之间的知识鸿沟。
• 在水文学情境中系统分析四类 PaML 方法。
• 评估 PaML 在水动力学和暴雨-径流过程中的应用。
• 提出 HydroPML 作为一个开源平台，以实现基于 PaML 的水文学应用。

提出的方法

• 定义 PaML，并将方法分为物理数据引导的 ML、物理信息 ML、物理嵌入 ML，以及物理感知混合学习。
• 提供 PaML 方法的系统性综述，示例涵盖神经网络、深算子网络和物理发现模型。
• 分析 PaML 在水动力学和暴雨径流过程中的应用，以按目标和方法进行分类。
• 突出面向过程性水文学的 PaML 的有前景方向与挑战。
• 将 HydroPML 作为PaML基础，支持基于 PaML 的水文应用和实时洪水预测。

实验结果

研究问题

• RQ1 用于将物理学与 ML 集成至水文学中的主要 PaML 范式有哪些？
• RQ2 PaML 方法在水动力学和暴雨径流过程中的表现如何？
• RQ3 面向过程性水文学的 PaML 的有前景方向与尚存挑战有哪些？
• RQ4 开源平台（HydroPML）如何促进 PaML 在水文学中的应用？

主要发现

• PaML 可分为四类：物理数据引导的 ML、物理信息 ML、物理嵌入 ML，以及物理感知混合学习。
• 对水文学中的 PaML 进行了系统性综述，聚焦于水动力学和暴雨-径流过程。
• HydroPML 作为开源水文学平台发布，旨在提升水系统的可解释性、因果性以及实时预测。
• 论文强调 PaML 能通过支持 ML辅助的假设和更充分地利用水文学中的大数据来加速科学洞察。
• 研究概述了 PaML 方法在不同应用目标中的有前景方向与挑战。