A novel physically constrained deep learning framework that explicitly incorporates biogeochemical knowledge of the nitrogen cycle into model design to enhance robustness and applies a soft directional penalty during training, thereby encouraging physical consistency while preserving point-wise accuracy.
Abstract
Accurate forecasting of total nitrogen (TN), a key driver of eutrophication, is critical for effective water environment management. However, due to the dynamic nature of hydrological and meteorological conditions, water quality data commonly exhibit non-stationarity and distribution shift, leading to significantly reduced generalization performance of existing data-driven models during practical deployment. To address this challenge, this paper proposes a novel physically constrained deep learning framework that explicitly incorporates biogeochemical knowledge of the nitrogen cycle into model design to enhance robustness. The framework synergistically optimizes three key components: (1) Feature Engineering: A dual-criterion selection mechanism based on correlation and stability is developed to identify robust predictors of TN; (2) Model Structure: Reversible Instance Normalization (RevIN) is incorporated to mitigate distribution shift between training and testing phases; (3) Loss Function: A physically constrained joint loss function is proposed. By adaptively identifying the dominant biogeochemical driver among three candidate environmental proxies, it applies a soft directional penalty during training, thereby encouraging physical consistency while preserving point-wise accuracy. Extensive experiments using real-world datasets from a reservoir in North China demonstrate the superior performance of our framework (RMSE = 0.1172 mg/L, R2 = 0.9269). The proposed framework significantly outperforms the state-of-the-art DLinear, reducing prediction error RMSE by approximately 30.7%. Ablation experiments further quantify the contribution of each component, confirming the pivotal role of RevIN in handling non-stationarity. This study provides a systematic solution to address distribution shift in water quality time-series forecasting.
The key finding is that hybrid and attention-enhanced deep learning architectures are frequently reported to outperform single architecture baselines in the reviewed literature and offer the greatest promise for robust, interpretable, and scalable systems for predicting water quality.
W. W. Wan Ismail, Abdallah Abuishaq, Nor Azlina Bt Ab. Aziz et al.· Environments· 0 citations
Soil salinity poses a significant challenge to agricultural productivity and sustainable land management in coastal and semi-arid regions. Soil salinity in coastal agricultural regions is controlled by complex interactions among surface conditions, groundwater processes, soil properties, and terrain characteristics. Ac...
Ming Ye, Peng Chen, Jun-Chi Ma et al.· Italian National Conference...· 0 citations
Surface ozone (O3) forms via nonlinear chemistry and is highly non-stationary, complicating 168-h
forecasts. Many models omit key physical drivers, mix time scales, and degrade quickly with lead time.
We propose a decomposition-ensemble framework that fuses surface observations with NWP predictors
(e.g., solar radiatio...
Ju-Chen Li, Chuan-Yu Zhou, G. Shi et al.· Polish Journal of Environmen...· 0 citations
This paper systematically reviews deep learning advances in ENSO prediction, tracking the transition from purely data-driven black-box approaches to physically interpretable modeling, and establishes a three-tier technical framework for physically interpretable deep learning.
Hao-Lin Jiang, Wei Fang, Yi-Jing Li et al.· Earth Science Informatics· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.