Advanced Metaheuristic and Deep Learning Models in Forecasting Hydrological Variables under Climate Uncertainty

The ability to accurately predict hydrological variables, including precipitation, streamflow, evapotranspiration, soil moisture, groundwater levels, evaporation, and infiltration is essential for the future of sustainable water resource management, disaster risk reduction, agricultural planning, and climate change adaptation. However, the non-linear, dynamic, and interconnected relationship of hydrological processes poses major problems to conventional modelling techniques, particularly under non-stationary climate circumstances.Recent advancements in metaheuristic optimization algorithms, coupled with deep learning, and AI, have revolutionised hydrological variables forecasting capabilities. While deep learning (DL) architectures (such as LSTM, GRU, and Transformers) have shown great promise in capturing intricate spatiotemporal patterns in hydrological data, metaheuristic algorithms such as Particle Swarm Optimisation, Genetic Algorithms, Grey Wolf Optimiser, Whale Optimisation Algorithm, and other nature-inspired algorithms have improved model calibration, feature selection, and parameter optimization. The integration of these methods has resulted in the development of hybrid models that are both computationally efficient and accurate in their forecasting capabilities. However, significant obstacles persist with respect to the interpretability of models, the quantification of uncertainty, the transferability of models across hydroclimatic regions, the integration of physical process comprehension, and the operational deployment of models in the face of evolving climatic conditions. There is an increasing demand for multidisciplinary research that blends advanced AI approaches with hydrological sciences to build reliable, explainable, and climate-resilient forecasting frameworks.This Research Topic focuses on development and implementation of hybrid metaheuristic-neural frameworks for predicting hydrological variables under climate uncertainty. It aims to promote collaboration among hydrologists, data scientists, climate researchers, and water resource managers to enhance predictive capabilities and facilitate evidence-based decision-making. Contributions on methodological improvements, operational applications, uncertainty assessment, explainable AI, and climate adaptation methods are especially welcome. To better understand this fast-growing field, contributions are encouraged on, but not limited to, the following themes:o Advanced Metaheuristic Integration: Novel applications of nature-inspired algorithms (e.g., PSO, GA, GWO, WOA, and Harris Hawks Optimization) for hyperparameter tuning and weight optimization in deep learning frameworks.o Hybrid Evolutionary-Neural Architectures: Development of coupled metaheuristic-DL models (e.g., PSO-LSTM, GA-GRU, Transformer-evolutionary hybrids) for robust multi-step-ahead hydrological forecasting.o Optimized Feature Selection and Dimensionality Reduction: Evolutionary and swarm intelligence approaches for identifying critical input features from large-scale geospatial, remote sensing, and IoT datasets.o Calibration Under Climate Non-Stationarity: Metaheuristic strategies for dynamic parameter recalibration of forecasting models in response to changing hydrological regimes and extreme weather events.o Uncertainty Quantification in Hybrid Frameworks: Evolutionary and probabilistic approaches to quantify structural, parameter, and input data uncertainty in AI-driven hydrological predictions.o Comparative Benchmarking of Metaheuristics: Systematic, peer-to-peer performance benchmarking of different metaheuristic optimization techniques (ensemble, bagging, boosting, stacking, voting) applied to complex hydrological simulators and predictors.o Optimization for Physics-Informed ML: Metaheuristic calibration of loss-function weights in physics-informed and process-guided neural networks.o Optimized Decision Support Systems: Hybrid models designed specifically to optimize water allocation, drought warning systems, flood mitigation, and reservoir operations. Topic editors: Anurag Malik, Parveen Sihag, Jitendra Rajput, Sani I. Abba Source: https://www.frontiersin.org/research-topics/83316/advanced-metaheuristic-and-deep-learning-models-in-forecasting-hydrological-variables-under-climate-uncertainty