A Data-facilitated Numerical Method for Richards Equation to Model Water Flow Dynamics in Soil

• DOI: 10.48550/arxiv.2310.02806
• Publication Date: 2023-01-01
• Authors: Zeyuan Song; Zheyu Jiang
• Keywords: Richards equation; Linearization; Solver; Benchmark (surveying)

Root-zone soil moisture monitoring is essential for precision agriculture, smart irrigation, and drought prevention. Modeling the spatiotemporal water flow dynamics in typically achieved by solving a hydrological model, such as Richards equation which highly nonlinear partial differential (PDE). In this paper, we present novel data-facilitated numerical method mixed-form equation. This method, call D-GRW (Data-facilitated global Random Walk) synergistically integrates adaptive linearization scheme, neural networks, random walk finite volume discretization framework to produce accurate solutions of with guaranteed convergence under reasonable assumptions. Through three illustrative examples, demonstrate discuss superior accuracy mass conservation performance our compare it benchmark methods commercial solver.