Using Physics-informed Neural Networks to Model the Hydro-morphodynamics of Mangrove Environments
Abstract
Modelling the hydro-morphodynamics of mangrove environments is key for implementing successful protection and restoration projects in a climatically vulnerable region. Nevertheless, simulating such dynamics is faced with computational and time complexities, given the nonlinear and complex nature of the problem, which could become a bottleneck for large-scale applications. The recent advances in machine learning, specifically, in physics-informed neural networks have gained much attention in the climate modelling field due to its ability in providing fast and accurate results, while preserving the binding physics laws and requiring small amounts of data from complex mathematical evaluations. This study investigates the application of physics-informed neural networks to quantify the ability of mangrove environments in attenuating waves and preventing erosion. Navier Stokes, the broadly used mathematical equation to solve for fluid dynamics, is used as the governing equation that constrains the neural network to respect the conservation of mass, energy, and momentum. The Sundarbans, the largest mangrove forest in the world located between India and Bangladesh, is considered as a case study. The Sundarbans is regularly subjected to tropical cyclones, the impacts of which endanger the lives of the region's four million people. This work represents the first application of physics-informed neural networks to model vegetation, and for a large-scale geographical application. The results demonstrate that the developed model is superior when compared to a conventional numerical model, in terms of time and data efficiency, yet produces equally strong overall results.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMNG23A..07F
- Keywords:
-
- Climate modelling;
- Hydro-morphodynamic Modelling;
- Machine Learning;
- Physics-informed neural networks;
- Mangrove environments.