Open research questions in Hydrological Forecasting Using AI

What the literature leaves open

• Traditional statistical methods have limitations in capturing nonlinear and dynamic nature of hydrological data. Advanced machine learning and deep learning techniques have not been fully explored for lake water level forecasting.

  Robust Monthly Lake Water Level Forecasting Using an Ensemble of Singular Spectrum Analysis Based Deep Learning and Categorical Boosting Models · 2026 · DOI

• the models remain structurally challenged under rare peak flow dynamics despite peak-weighted training, - MCD-based intervals primarily capture epistemic model/parameter uncertainty rather than total predictive uncertainty, - under-coverage can arise not only from intervals that are too narrow, but also from intervals that are centered around biased peak flow predictions

  Peak-oriented one-day-ahead streamflow forecasting using hybrid deep learning: uncertainty quantification, SHAP-based interpretability, and MCDA-based model selection · 2026 · DOI

• Traditional process-based hydrologic and hydraulic models have limitations, such as requiring extensive input data and careful calibration. Peak flow forecasting remains challenging due to its nonlinearity and rarity. There is a need for a peak-oriented deep learning framework for one-day-ahead daily streamflow forecasting.

  Peak-oriented one-day-ahead streamflow forecasting using hybrid deep learning: uncertainty quantification, SHAP-based interpretability, and MCDA-based model selection · 2026 · DOI

• Variations in discharge, flow direction, travel time, tributary contributions, reservoir regulation, and pollutant-specific transformation remain incompletely represented in current graph models. The development of more physically informed and dynamically adaptive graph representations is a significant challenge. Evaluating the performance of graph models in river water-quality forecasting is challenging due to the complexity of the underlying hydrological processes.

  Graph Learning for River Water Quality Forecasting: Advances in Hydrological Connectivity Representation, Dynamic Topology, and Physics-Informed Modeling · 2026 · DOI

• Dynamic edge updating, event- and pollutant-adaptive graphs, travel-time-aware message passing, mass-conserving architectures, and cross-basin representation learning should be explored. Graph uncertainty should be evaluated as a first-class output.

  Graph Learning for River Water Quality Forecasting: Advances in Hydrological Connectivity Representation, Dynamic Topology, and Physics-Informed Modeling · 2026 · DOI

• None of these studies evaluate machine learning weather forecasting models on real-time or near-real-time operational deployment; existing work uses historical datasets and offline validation, leaving a gap in understanding computational latency, data availability constraints, and model performance in live forecasting systems.

  Weather Analysis Using Machine Learning · 2026 · DOI

• Across these studies, interpretability and explainability of machine learning weather forecasting models remain unaddressed; while one study identifies interpretability as a limitation, none of the papers propose or evaluate methods to explain which meteorological features drive predictions or how models capture nonlinear weather relationships.

  Weather Analysis Using Machine Learning · 2026 · DOI

• The proposed framework is particularly promising for regions where hydrological observations are sparse.

  Enhanced early warning of extreme river discharge events in the Yangtze River Basin using atmospheric circulation signals · 2026 · DOI

• However, the pathways of Irrawaddy Diluted Water (IDW), an important source of freshwater, are still not well understood.

  Impact of the Irrawaddy Diluted Water on Upper‐Layer Salinity in the Bay of Bengal · 2025 · DOI

• This study highlights the efficacy of meta-learning techniques in overcoming the limitations posed by data scarcity and enhancing flood forecasting accuracy where historical data are limited.

  MetaTrans-FSTSF: A Transformer-Based Meta-Learning Framework for Few-Shot Time Series Forecasting in Flood Prediction · 2024 · DOI

• These datasets were preprocessed to align with the meta-learning approach, ensuring their suitability for tasks with limited data availability.

  MetaTrans-FSTSF: A Transformer-Based Meta-Learning Framework for Few-Shot Time Series Forecasting in Flood Prediction · 2024 · DOI

• There is a gap in comparative studies evaluating the performance of machine learning models in predicting both groundwater and surface water dynamics in semi-arid regions. The study addresses this gap by systematically comparing seven machine learning models.

  A comparative analysis of machine learning models for predicting groundwater and surface water in a stressed semi-arid watershed: The Khanmirza case study · 2026 · DOI

• Existing DL models lack physical consistency and generalization. Prior work ignores inter-variable dependencies and lacks physical knowledge.

  Integrating multi-task learning with a differentiable physics constrained framework for hydrological forecasting · 2026 · DOI

• The physical constraint currently employed is primarily based on water balance alone, which is insufficient to capture the complex, nonlinear coupling of multiple factors including energy availability and vegetation physiological responses that govern high-ET events.

  Integrating multi-task learning with a differentiable physics constrained framework for hydrological forecasting · 2026 · DOI

• The lack of effective methods for detecting anomalies in water quality data. The need for a deep learning-based approach for unsupervised anomaly detection.

  Seasonal Anomaly Detection in the Halda River Using a Multivariate Deep Learning Framework · 2026 · DOI

• The complexity of the Madden-Julian Oscillation. The need for accurate and reliable predictive models. The challenge of integrating multiple datasets.

  Model Hybrid Artificial Neural Network and Reservoir Computing for Extreme Weather Predicting Based on Madden Julian Oscillation · 2026 · DOI

• The ANN-RC framework is proposed to enhance physical consistency by explicitly representing sub-seasonal MJO memory; however, the paper does not specify what additional datasets (e.g., upper-level wind fields, sea surface temperature patterns, convective cloud observations) or physical constraints should be incorporated to improve the weak rainfall prediction capability.

  Model Hybrid Artificial Neural Network and Reservoir Computing for Extreme Weather Predicting Based on Madden Julian Oscillation · 2026 · DOI

• Existing machine learning models struggle with imbalanced and small-sample datasets. There is a need for a novel approach to address these limitations.

  A new machine learning technique for predicting river water quality using AVOA-RNN · 2026 · DOI

• This study proposed a hybrid AVOA-RNN framework for predicting river water quality using the Cauvery River data- set. By integrating advanced imbalance-handling strategies (SMOTE, SMOGN) with African Vulture Optimization for hyper-parameter tuning, the model demonstrated supe- rior predictive capability in both regression and classifica- tion tasks. In classification, the proposed model achieved 97.3% accuracy, 99.52% sensitivity, 97.42% specificity, and a ROC-AUC of 0.97, outperforming comparative approaches such as CNN (93.85%), LSTM (94.92%), RNN (95.73%), GRU (96.38%), WOA-NN (0.82), GA-LSTM (0.80), EMD- (F1 = 0.86), WOA-LSTM (F1 = 0.84), AQPSO-SOFNN Y. et al. Water Science (2026) 40:16 Page 19 of 21 Fig. 11 SHAP-based feature importance (dissolved oxygen (DO ≈ 1.3), BOD (≈ 0.65), pH (≈ 0.45), and TDS (≈ 0.25) dominate, while NO₃, Cl, PO₄, SO₄, turbidity, and FC contribute marginally) Fig. 12 Comparative SHAP analysis for the existing models Y. et al. Water Science (2026) 40:16 Page 20 of 21 and SOFNN-HPS (Accuracy = 0.87). In regression tasks, AVOA-RNN obtained the lowest RMSE (8.3) compared to WOA-NN (9.0), GA-LSTM (9.6), EMD-WOA-LSTM (8.8), AQPSO-SOFNN (8.7), and SOFNN-HPS (8.6), along with a high coefficient of determination ( R2 =0.91). These results confirm that AVOA-RNN not only surpasses tradi- tional and hybrid baselines but also ensures greater ecolog- ical interpretability through SHAP-based feature analysis, highlighting DO, BOD, pH, and TDS as dominant param- eters. The framework thus provides a robust and interpret- able solution for sustainable water quality monitoring and management.

  A new machine learning technique for predicting river water quality using AVOA-RNN · 2026 · DOI

• The study focuses on a single river segment, which may not be representative of all fluvial landscapes. The framework is trained using a limited dataset, which may not capture all possible scenarios.

  An integrated deep learning framework enables rapid spatiotemporal morphodynamic predictions toward long-term simulations · 2026 · DOI

• Future research can focus on applying the framework to other fluvial landscapes, providing insights into morphodynamic processes. Future research can also focus on improving the accuracy of the framework, exploring new architectures and training datasets.

  An integrated deep learning framework enables rapid spatiotemporal morphodynamic predictions toward long-term simulations · 2026 · DOI

• In summary, the combination of ML and DL models improved streamflow prediction in basins with limited data and successfully restored missing hydrometeorological data.

  Stream flow prediction utilizing deep learning models in the Lake Abaya-Chamo sub-basin, South Ethiopia · 2026 · DOI

• There has been no research done on the soil infiltration process employing machine learning methodologies. The study aims to fill this gap by demonstrating the potential of machine learning models for precise infiltration modeling.

  Approximation of cumulative infiltration of soils in arid and semi-arid regions · 2026 · DOI

• Developing optimized hybrid models for other environmental forecasting applications. Evaluating the performance of the XGBoost-FDA hybrid model using different datasets and case studies.

  Enhancing suspended sediment load forecasting using a physics-based optimized hybrid model: a regional study in the Mazandaran basins · 2026 · DOI

• The lack of accurate suspended sediment load forecasting models. The need for optimized hybrid models that combine machine learning and optimization algorithms.

  Enhancing suspended sediment load forecasting using a physics-based optimized hybrid model: a regional study in the Mazandaran basins · 2026 · DOI

Most-cited papers in Hydrological Forecasting Using AI

• Machine learning and deep learning—A review for ecologists · Methods in Ecology and Evolution · 2023 · 419 citations
• Groundwater level prediction using machine learning models: A comprehensive review · Neurocomputing · 2022 · 398 citations
• Research on particle swarm optimization in LSTM neural networks for rainfall-runoff simulation · Journal of Hydrology · 2022 · 331 citations
• Deep learning for water quality · Nature Water · 2024 · 259 citations
• Applications of machine learning to water resources management: A review of present status and future opportunities · Journal of Cleaner Production · 2024 · 252 citations
• Hybrid forecasting: blending climate predictions with AI models · Hydrology and earth system sciences · 2023 · 221 citations
• Evaluating Different Machine Learning Methods for Upscaling Evapotranspiration from Flux Towers to the Regional Scale · Journal of Geophysical Research Atmospheres · 2018 · 215 citations
• Continuous streamflow prediction in ungauged basins: long short-term memory neural networks clearly outperform traditional hydrological models · Hydrology and earth system sciences · 2023 · 213 citations
• Groundwater level prediction using machine learning algorithms in a drought-prone area · Neural Computing and Applications · 2022 · 199 citations
• Adoption of Machine Learning Techniques in Ecology and Earth Science · One Ecosystem · 2016 · 193 citations

Most recent work

• A physics informed deep learning framework for rainfall forecasting in diverse climatic regions · Discover Artificial Intelligence · 2026
• Machine learning-driven rainfall forecasting model for sustainable and adaptive infrastructure planning · Discover Sustainability · 2026
• MSTRFormer: A multi-factor daily runoff forecasting model integrating adaptive graph convolution and linear attention · Applied Soft Computing · 2026
• A comparative analysis of machine learning models for predicting groundwater and surface water in a stressed semi-arid watershed: The Khanmirza case study · Environmental Earth Sciences · 2026
• Modeling CO2 fluxes in coastal wetlands of China using explainable sequence-based deep learning · Ecological Informatics · 2026
• Estimation of Water Quality in Lakes and Rivers Using Remote Sensing and Artificial Intelligence: A Review of Image Processing and Validation Strategies · Limnological Review · 2026
• Correction: Machine learning-based correlation analysis of conventional water quality parameters and composite pollution index in the Luoqing river of the South China Sea Coastal Zone · Frontiers in Marine Science · 2026
• Robust Monthly Lake Water Level Forecasting Using an Ensemble of Singular Spectrum Analysis Based Deep Learning and Categorical Boosting Models · Water Resources Management · 2026
• A two-phase physics-informed hybrid deep learning model for flood forecasting: improving process learning and interpretability · Advances in Water Resources · 2026
• Convolutional autoencoder latent-space modeling with climate-index–conditioned recurrent networks for assessing subseasonal potential forecast skill of the Baltic Sea heat content · Engineering Applications of Artificial Intelligence · 2026

Related topics in Environmental Science