Machine learning (ML) models have achieved a remarkable success in hydrological and hydraulic (H&H) modeling and show increasing applications in the H&H communities. However, purely data-driven ML models have some limitations such as data hungry, data generation, lack of interpretability, deterministic predictions without uncertainty quantification (UQ), and extrapolation to different climate, land use land cover, and other hydrological conditions.
In this Research Topic, we invite manuscripts that use data-driven ML to advance hydrological and hydraulic predictability at all scales, where the data could include both measurements and model simulation data. The scope of this special issue includes both novel ML methodologies developed for advancing hydrological science (e.g., interpretability of ML algorithms, UQ for ML predictions, transfer learning, and data-efficient ML) and applications of ML to hydrological and hydraulic modeling (e.g., water availability, water quality, floods and droughts).
This research topic was curated in association with the HydroML conference of May 2022, held at Penn State University. The conference goals included building up the community around the field of Machine Learning, bolstering expertise and encouraging collaborative progression. This Article Collection was published in alignment with these goals.
Possible topics include but are not limited to:
? Application of data-driven ML models in hydrological and hydraulic modeling
? Data-efficient ML methods for small-sample datasets
? ML methods to prioritize and optimize data collection
? ML methods to improve predictions of extreme events and in ungauged basins
? Explainable and interpretable ML models for advancing hydrological science
? Uncertainty quantification of ML models in hydrological prediction
? Parameter estimation and inversion with ML-based surrogates
Machine learning (ML) models have achieved a remarkable success in hydrological and hydraulic (H&H) modeling and show increasing applications in the H&H communities. However, purely data-driven ML models have some limitations such as data hungry, data generation, lack of interpretability, deterministic predictions without uncertainty quantification (UQ), and extrapolation to different climate, land use land cover, and other hydrological conditions.
In this Research Topic, we invite manuscripts that use data-driven ML to advance hydrological and hydraulic predictability at all scales, where the data could include both measurements and model simulation data. The scope of this special issue includes both novel ML methodologies developed for advancing hydrological science (e.g., interpretability of ML algorithms, UQ for ML predictions, transfer learning, and data-efficient ML) and applications of ML to hydrological and hydraulic modeling (e.g., water availability, water quality, floods and droughts).
This research topic was curated in association with the HydroML conference of May 2022, held at Penn State University. The conference goals included building up the community around the field of Machine Learning, bolstering expertise and encouraging collaborative progression. This Article Collection was published in alignment with these goals.
Possible topics include but are not limited to:
? Application of data-driven ML models in hydrological and hydraulic modeling
? Data-efficient ML methods for small-sample datasets
? ML methods to prioritize and optimize data collection
? ML methods to improve predictions of extreme events and in ungauged basins
? Explainable and interpretable ML models for advancing hydrological science
? Uncertainty quantification of ML models in hydrological prediction
? Parameter estimation and inversion with ML-based surrogates
