📄 中文摘要
地下水是水循环的重要组成部分，但其复杂且依赖于环境的关系使得建模变得困难。传统的基于理论的模型虽然是科学理解的基石，但其计算需求高、简化假设多以及校准要求严格，限制了其应用。近年来，数据驱动模型作为强有力的替代方案逐渐崭露头角，尤其是深度学习因其设计灵活性和学习复杂关系的能力而成为领先的方法。研究提出了一种基于注意力机制的纯深度学习模型STAINet，旨在预测任意位置的每周地下水位，充分利用空间稀疏的观测数据。该模型能够处理变化的地点数量，展现了深度学习在地下水位预测中的潜力。
基于纯深度学习和物理引导的时空地下水位预测解决方案
📄 English Summary
Pure and Physics-Guided Deep Learning Solutions for Spatio-Temporal Groundwater Level Prediction at Arbitrary Locations
Groundwater is a crucial component of the water cycle, yet its intricate and context-dependent relationships pose significant challenges for modeling. Traditional theory-based models have served as the foundation of scientific understanding; however, their high computational demands, simplifying assumptions, and calibration requirements limit their applicability. Recently, data-driven models have emerged as powerful alternatives, with deep learning standing out due to its design flexibility and capability to learn complex relationships. This study proposes an attention-based pure deep learning model, named STAINet, for predicting weekly groundwater levels at arbitrary and variable locations, effectively leveraging spatially sparse observational data. The model demonstrates the potential of deep learning in groundwater level prediction.