Soil moisture plays a crucial role in the hydrological cycle, but accurately predicting soil moisture presents challenges due to the nonlinearity of soil water transport and the variability of boundary conditions. Deep learning has emerged as a promising approach for simulating soil moisture dynamics. In this study, we explore 10 different network structures to uncover their data utilization mechanisms and to maximize the potential of deep learning for soil moisture prediction, including three basic feature extractors and seven diverse hybrid structures, six of which are applied to soil moisture prediction for the first time. We compare the predictive abilities and computational costs of the models across different soil textures and depths systematically. Furthermore, we exploit the interpretability of the models to gain insights into their workings and attempt to advance our understanding of deep learning in soil moisture dynamics. For soil moisture forecasting, our results demonstrate that the temporal modeling capability of long short-term memory (LSTM) is well suited. Furthermore, the improved accuracy achieved by feature attention LSTM (FA-LSTM) and the generative-adversarial-network-based LSTM (GAN-LSTM), along with the Shapley (SHAP) additive explanations analysis, help us discover the effectiveness of attention mechanisms and the benefits of adversarial training in feature extraction. These findings provide effective network design principles. The Shapley values also reveal varying data leveraging approaches among different models. The t-distributed stochastic neighbor embedding (t-SNE) visualization illustrates differences in encoded features across models. In summary, our comprehensive study provides insights into soil moisture prediction and highlights the importance of the appropriate model design for specific soil moisture prediction tasks. We also hope this work serves as a reference for deep learning studies in other hydrology problems. The codes of 3 machine learning and 10 deep learning models are open source.
Soil moisture plays a crucial role in the hydrological cycle, but accurately predicting soil moisture presents challenges due to the nonlinearity of soil water transport and variability of boundary conditions. Deep learning has emerged as a promising approach for simulating soil moisture dynamics. In this study, we explore ten different network structures to uncover their mechanisms of data utilization and maximize the potential of deep learning for soil moisture prediction, including three basic feature extractors and seven diverse hybrid structures, six of which are applied to soil moisture prediction for the first time. We compare the predictive abilities and computational costs of the models across different soil textures and depths systematically. Furthermore, we exploit the interpretability of the models to gain insights into their workings and attempt to advance our understanding of deep learning in soil moisture dynamics. For soil moisture forecasting, our results demonstrate that the temporal modeling capability of Long Short-Term Memory (LSTM) is well-suited. Besides, the improved accuracy achieved by feature attention LSTM (FA-LSTM) and the generative adversarial network-based LSTM (GAN-LSTM), along with the Shapley additive explanations (SHAP) analysis, help us discover the effectiveness of attention mechanisms and the benefits of adversarial training in feature extraction. These findings provide effective network design principles. The Shapley values also reveal varying data leveraging approaches among different models. The t-Distributed Stochastic Neighbor Embedding (t-SNE) visualization illustrates differences in encoded features across models. In summary, our comprehensive study provides insights into soil moisture prediction and highlights the importance of the appropriate model design for specific soil moisture prediction tasks. We also hope this work serves as a reference for deep learning studies in other hydrology problems. The codes of 3 machine learning and 10 deep learning models are open sourced.
19 July 2023 in Hydrology and Earth System Sciences
Conventional data worth (DW) analysis for soil water problems depends on physical dynamic models. The widespread occurrence of model structural errors and the strong nonlinearity of soil water flow may lead to biased or wrong worth assessment. By introducing the nonparametric data worth analysis (NP-DWA) framework coupled with the ensemble Kalman filter (EnKF), this real-world case study attempts to assess the worth of potential soil moisture observations regarding the reconstruction of fully data-driven soil water flow models prior to data gathering. The DW of real-time soil moisture observations after Gaussian process training and Kalman update was quantified with three representative information metrics, including the trace, Shannon entropy difference and relative entropy. The sequential NP-DWA framework was examined by a number of cases in terms of the variable of interest, spatial location, observation error, and prior data content. Our results indicated that, similarly to the traditional DW analysis based on physical models, the overall increasing trend of the DW from the sequential augmentation of additional observations within the NP-DWA framework was also susceptible to interruptions by localized surges due to never-experienced atmospheric conditions (i.e., rainfall events). The difference is that this biased DW in the former is caused by model structural errors triggered by contrasting scenarios, which is difficult to be compensated for by assimilating more prior data, while this performance degradation in the NP-DWA can be effectively alleviated by enriching training scenarios or the appropriate amplification of observational noise under extreme meteorological conditions. Nevertheless, a substantial expansion of the prior data content may cause an unexpected increase in the DW of future potential observations due to the possible introduction of ensuing observation noises. Hence, high-quality and representative small data may be a better choice than unfiltered big data. Compared with the observations in the surface layer with the strongest time variability, the soil water content in the middle layer robustly exhibited remarkable superiority in the construction of model-free soil moisture models. We also demonstrated that the DW assessment performance was jointly determined by 3C, i.e., the capacity of potential observation realizations to capture actual observations, the correlation of potential observations with the variables of interest and the choice of DW indicators. Direct mapping from regular meteorological data to soil water content within the NP-DWA mitigated the adverse effects of nonlinearity-related interference, which thus facilitated the identification of the soil moisture covariance matrix, especially the cross-covariance.
Soil moisture is a critical component of the hydrological cycle, but accurately predicting it remains challenging due to the nonlinearity of soil water transport, variability in boundary conditions, and the intricate nature of soil properties. Recently, deep learning has shown promise in this domain, typically by modeling temporal dependencies for soil moisture predictions. In this study, we propose non-local neural networks (NLNN) to convert this problem into a single-time-step, simultaneous multi-depth soil moisture forecasting. By facilitating mutual compensation among different depths, this method enables a representation of vertical heterogeneity and inter-layer connectivity without physical assumptions, leading to precise and efficient predictions in diverse scenarios. Our non-local operation design includes the embedded Gaussian operations and disentangled physics-guided operations, resulting in two variants: the self-attention non-local neural network (SA-NLNN) and the physics-guided non-local neural network (PG-NLNN). The models offer visual interpretability, providing insights into intricate mechanisms of soil moisture dynamics. Notably, the model guided by physics yields more stable and reasonable qualitative interpretations. With in-situ observations, we demonstrate that our proposed models perform satisfactorily. The physics-guided non-local operations significantly enhance accuracy and reliability. Additionally, our models adapt to diverse time-scale situations while maintaining high computational efficiency. Both models exhibit robust noise resistance, with physics guidance enhancing PG-NLNN’s noise resistance. In summary, our work addresses the soil moisture prediction challenge in a novel way, highlighting the potential of NLNN and the importance of incorporating physic guidance in data-driven models.
19 May 2026 in Hydrology and Earth System Sciences
Soil moisture (SM) is a critical component of the hydrological cycle, but accurately predicting it remains challenging due to the nonlinearity of soil water transport, variability in boundary conditions, and the intricate nature of soil properties. Recently, deep learning has shown promise in this domain, typically by modeling temporal dependencies for soil moisture predictions. In this study, we propose non-local neural networks (NLNNs) to convert this problem into a single-time-step, simultaneous multi-depth soil moisture forecasting. The non-local operation design includes embedded Gaussian operations and disentangled knowledge-guided operations, resulting in two variants: the self-attention non-local neural network (SA-NLNN) and the knowledge-guided non-local neural network (KG-NLNN). The knowledge-guided non-local operation is designed to capture vertical soil moisture relationships by decomposing the influences on soil moisture at a given depth into four components, each governed by distinct physical processes. The models offer visual interpretability through learned non-local weights, which reveal interactions among soil moisture across different depths, thereby enabling a qualitative representation of inter-layer connectivity. Notably, the model guided by soil moisture transport knowledge yields more stable and reasonable interpretations. With in-situ observations, we demonstrate that our proposed models perform satisfactorily. The knowledge-guided non-local operations significantly enhance accuracy and reliability. Additionally, our models adapt to diverse time-scale situations while maintaining high computational efficiency. Both models exhibit robust noise resistance, with knowledge guidance enhancing KG-NLNN's noise resistance. In summary, our work addresses the soil moisture prediction challenge in a novel way, highlighting the potential of NLNN and the importance of incorporating physic guidance in data-driven models.