Abstract:Large-scale river models are being refined over coastal regions to improve the scientific understanding of coastal processes, hazards and responses to climate change. However, coarse mesh resolutions and approximations in physical representations of tidal rivers limit the performance of such models at resolving the complex flow dynamics especially near the river-ocean interface, resulting in inaccurate simulations of flood inundation. In this research, we propose a machine learning (ML) framework based on the state-of-the-art physics-informed neural network (PINN) to simulate the downscaled flow at the subgrid scale. First, we demonstrate that PINN is able to assimilate observations of various types and solve the one-dimensional (1-D) Saint-Venant equations (SVE) directly. We perform the flow simulations over a floodplain and along an open channel in several synthetic case studies. The PINN performance is evaluated against analytical solutions and numerical models. Our results indicate that the PINN solutions of water depth have satisfactory accuracy with limited observations assimilated. In the case of flood wave propagation induced by storm surge and tide, a new neural network architecture is proposed based on Fourier feature embeddings that seamlessly encodes the periodic tidal boundary condition in the PINN's formulation. Furthermore, we show that the PINN-based downscaling can produce more reasonable subgrid solutions of the along-channel water depth by assimilating observational data. The PINN solution outperforms the simple linear interpolation in resolving the topography and dynamic flow regimes at the subgrid scale. This study provides a promising path towards improving emulation capabilities in large-scale models to characterize fine-scale coastal processes.

What problem does this paper attempt to address?

This paper aims to address the deficiencies of large - scale river models in the refined simulation of coastal areas. Specifically, due to the limitations of coarse grid resolution and approximate treatment of physical processes, large - scale river models perform poorly in simulating complex hydrodynamic flows, especially flood inundation at the river - sea junction. This leads to inaccurate simulations of flood wave propagation. To solve these problems, the author proposes a machine - learning framework based on the latest Physics - Informed Neural Networks (PINN) for simulating downscaled flows at the sub - grid scale. This method can directly solve the one - dimensional (1 - D) Saint - Venant equations (SVE) and improve the simulation accuracy by fusing multiple types of observational data. ### Main Objectives 1. **Improve the Simulation Accuracy of Large - scale River Models**: By introducing the PINN framework, more reasonable water depth solutions can be generated at the sub - grid scale, especially in the complex flows at the river - sea junction. 2. **Fuse Multi - source Observational Data**: Utilize in - situ measurement data and remote sensing data to improve model performance through data assimilation methods. 3. **Handle Periodic Tidal Boundary Conditions**: Propose a new neural network architecture to seamlessly encode periodic tidal boundary conditions through Fourier feature embedding, thereby better simulating the propagation of tidal waves. ### Specific Problems - **Coarse Grid Resolution**: Large - scale river models usually use low - resolution grids and cannot capture local flow details, especially in areas with large topographic changes. - **Approximate Treatment of Physical Processes**: To reduce computational costs, large - scale models usually simplify the momentum equation, which leads to inaccurate simulations of complex flow phenomena. - **Data Assimilation**: How to effectively fuse in - situ observational data and remote sensing data to improve the model's predictive ability. ### Solutions - **PINN Framework**: Use PINN to directly solve the Saint - Venant equations, calculate partial derivatives through automatic differentiation techniques, and train the model by minimizing the loss function. - **Multi - scale Fourier Feature Embedding**: For periodic tidal boundary conditions, design a new neural network architecture to handle time inputs through Fourier feature embedding, thereby better simulating the propagation of tidal waves. - **Data Assimilation**: Use in - situ observational data and remote sensing data as part of the training data, and adjust the weight coefficients to balance different loss terms, improving the robustness and accuracy of the model. ### Results and Significance - **Verification and Evaluation**: Through synthetic case studies, the effectiveness and accuracy of the PINN method in simulating flood wave propagation are verified. - **Potential for Practical Applications**: This method provides a new path for improving the simulation ability of large - scale models at fine scales, which is helpful for better understanding and predicting flood risks in coastal areas. In conclusion, this paper solves the key problems of large - scale river models in simulating complex flow phenomena by introducing the PINN framework, providing new ideas for improving the model's predictive ability and practical applications.

Physics-informed neural networks of the Saint-Venant equations for downscaling a large-scale river model

Enhanced physics‐informed neural networks for efficient modelling of hydrodynamics in river networks

Physics-Informed Neural Networks for Modeling Water Flows in a River Channel

Physical information-fused deep learning model ensembled with a subregion-specific sampling method for predicting flood dynamics

A Physics-Informed Neural Network to Model Port Channels

An Improved PINN Algorithm for Shallow Water Equations Driven by Deep Learning

Investigating Steady Unconfined Groundwater Flow using Physics Informed Neural Networks

A novel discretized physics-informed neural network model applied to the Navier-Stokes equations

Solving the one dimensional vertical suspended sediment mixing equation with arbitrary eddy diffusivity profiles using temporal normalized physics-informed neural networks

A novel solution for seepage problems using physics-informed neural networks

Inference of water waves surface elevation from horizontal velocity components using physics informed neural networks (PINN)

A multiscale stabilized physics informed neural networks with weakly imposed boundary conditions transfer learning method for modeling advection dominated flow

Physics-informed neural network combined with characteristic-based split for solving forward and inverse problems involving Navier–Stokes equations

Physics‐Informed Neural Networks for the Augmented System of Shallow Water Equations With Topography

MRF-PINN: A Multi-Receptive-Field convolutional physics-informed neural network for solving partial differential equations

Modeling water flow in unsaturated soils through physics-informed neural network with principled loss function

Application of Physics-Informed Neural Network on the Example of Modeling Hydrodynamic Processes that Allow an Analytical Solution

Higher-order error estimates for physics-informed neural networks approximating the primitive equations

Physics-Aware Downsampling with Deep Learning for Scalable Flood Modeling

Modeling multisource multifrequency acoustic wavefields by a multiscale Fourier feature physics-informed neural network with adaptive activation functions

Investigating neural networks with groundwater flow equation loss