Articles | Volume 26, issue 12
https://doi.org/10.5194/hess-26-3079-2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/hess-26-3079-2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Research article
| 
20 Jun 2022
Research article |
| 20 Jun 2022
| 20 Jun 2022
Hydrological concept formation inside long short-term memory (LSTM) networks
School of Geography and the Environment, University of Oxford, South Parks Road, Oxford, OX1 3QY, United Kingdom
International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria
Steven Reece
Department of Engineering, University of Oxford, Oxford, United Kingdom
Frederik Kratzert
Google Research, Vienna, Austria
Daniel Klotz
LIT AI Lab & Institute for Machine Learning, Johannes Kepler University Linz, Linz, Austria
Martin Gauch
LIT AI Lab & Institute for Machine Learning, Johannes Kepler University Linz, Linz, Austria
Jens De Bruijn
International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria
Institute for Environmental Studies, VU University, De Boelelaan 1087, 1081 HV Amsterdam, the Netherlands
Reetik Kumar Sahu
International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria
Peter Greve
International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria
Louise Slater
School of Geography and the Environment, University of Oxford, South Parks Road, Oxford, OX1 3QY, United Kingdom
Simon J. Dadson
School of Geography and the Environment, University of Oxford, South Parks Road, Oxford, OX1 3QY, United Kingdom
UK Centre for Ecology and Hydrology, Maclean Building, Crowmarsh Gifford, Wallingford, OX10 8BB, United Kingdom
Viewed
Total article views: 6,076 (including HTML, PDF, and XML)
Cumulative views and downloads
(calculated since 23 Nov 2021)
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 4,266 | 1,739 | 71 | 6,076 | 76 | 64 |
- HTML: 4,266
- PDF: 1,739
- XML: 71
- Total: 6,076
- BibTeX: 76
- EndNote: 64
Total article views: 4,075 (including HTML, PDF, and XML)
Cumulative views and downloads
(calculated since 20 Jun 2022)
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 2,963 | 1,072 | 40 | 4,075 | 58 | 50 |
- HTML: 2,963
- PDF: 1,072
- XML: 40
- Total: 4,075
- BibTeX: 58
- EndNote: 50
Total article views: 2,001 (including HTML, PDF, and XML)
Cumulative views and downloads
(calculated since 23 Nov 2021)
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,303 | 667 | 31 | 2,001 | 18 | 14 |
- HTML: 1,303
- PDF: 667
- XML: 31
- Total: 2,001
- BibTeX: 18
- EndNote: 14
Viewed (geographical distribution)
Total article views: 6,076 (including HTML, PDF, and XML)
Thereof 5,717 with geography defined
and 359 with unknown origin.
Total article views: 4,075 (including HTML, PDF, and XML)
Thereof 3,832 with geography defined
and 243 with unknown origin.
Total article views: 2,001 (including HTML, PDF, and XML)
Thereof 1,885 with geography defined
and 116 with unknown origin.
| Country | # | Views | % |
|---|
| Country | # | Views | % |
|---|
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
1
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
1
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
1
Cited
40 citations as recorded by crossref.
- On how data are partitioned in model development and evaluation: Confronting the elephant in the room to enhance model generalization H. Maier et al. 10.1016/j.envsoft.2023.105779
- Hybrid forecasting: blending climate predictions with AI models L. Slater et al. 10.5194/hess-27-1865-2023
- Developing a Physics‐Informed Deep Learning Model to Simulate Runoff Response to Climate Change in Alpine Catchments L. Zhong et al. 10.1029/2022WR034118
- Continuous streamflow prediction in ungauged basins: long short-term memory neural networks clearly outperform traditional hydrological models R. Arsenault et al. 10.5194/hess-27-139-2023
- Short-term runoff forecasting in an alpine catchment with a long short-term memory neural network C. Frank et al. 10.3389/frwa.2023.1126310
- Pluvial flood susceptibility mapping for data-scarce urban areas using graph attention network and basic flood conditioning factors Z. Wang et al. 10.1080/10106049.2023.2275692
- Improving hydrologic models for predictions and process understanding using neural ODEs M. Höge et al. 10.5194/hess-26-5085-2022
- Investigation of hydrometeorological influences on reservoir releases using explainable machine learning methods M. Fan et al. 10.3389/frwa.2023.1112970
- On the need for physical constraints in deep learning rainfall–runoff projections under climate change: a sensitivity analysis to warming and shifts in potential evapotranspiration S. Wi & S. Steinschneider 10.5194/hess-28-479-2024
- Improving short-term streamflow forecasting by flow mode clustering S. Liu et al. 10.1007/s00477-022-02367-z
- Toward interpretable LSTM-based modeling of hydrological systems L. De la Fuente et al. 10.5194/hess-28-945-2024
- A hydrological process-based neural network model for hourly runoff forecasting S. Gao et al. 10.1016/j.envsoft.2024.106029
- Enhancing process-based hydrological models with embedded neural networks: A hybrid approach B. Li et al. 10.1016/j.jhydrol.2023.130107
- Information and disinformation in hydrological data across space: The case of streamflow predictions using machine learning A. Gupta 10.1016/j.ejrh.2023.101607
- Development of a Long-Range Hydrological Drought Prediction Framework Using Deep Learning M. Khan & R. Maity 10.1007/s11269-024-03735-w
- Application of Deep Learning in Drainage Systems Monitoring Data Repair—A Case Study Using Con-GRU Model L. He et al. 10.3390/w15081635
- Inverse Modeling for Subsurface Flow Based on Deep Learning Surrogates and Active Learning Strategies N. Wang et al. 10.1029/2022WR033644
- A Data-Driven Approach for Building the Profile of Water Storage Capacity of Soils J. Zhou et al. 10.3390/s23125599
- DeepGR4J: A deep learning hybridization approach for conceptual rainfall-runoff modelling A. Kapoor et al. 10.1016/j.envsoft.2023.105831
- Ensemble and stochastic conceptual data-driven approaches for improving streamflow simulations: Exploring different hydrological and data-driven models and a diagnostic tool D. Hah et al. 10.1016/j.envsoft.2022.105474
- Revisit hydrological modeling in ungauged catchments comparing regionalization, satellite observations, and machine learning approaches R. Dasgupta et al. 10.1016/j.hydres.2023.11.001
- Using Deep Learning Algorithms for Intermittent Streamflow Prediction in the Headwaters of the Colorado River, Texas F. Forghanparast & G. Mohammadi 10.3390/w14192972
- Geospatial Artificial Intelligence (GeoAI) in the Integrated Hydrological and Fluvial Systems Modeling: Review of Current Applications and Trends C. Gonzales-Inca et al. 10.3390/w14142211
- On the use of convolutional deep learning to predict shoreline change E. Gomez-de la Peña et al. 10.5194/esurf-11-1145-2023
- The Great Lakes Runoff Intercomparison Project Phase 4: the Great Lakes (GRIP-GL) J. Mai et al. 10.5194/hess-26-3537-2022
- A Machine Learning-Based Probabilistic Approach for Irrigation Scheduling S. Srivastava et al. 10.1007/s11269-024-03746-7
- Understanding the Way Machines Simulate Hydrological Processes—A Case Study of Predicting Fine-Scale Watershed Response on a Distributed Framework D. Kim et al. 10.1109/TGRS.2023.3285540
- Near‐term forecasts of stream temperature using deep learning and data assimilation in support of management decisions J. Zwart et al. 10.1111/1752-1688.13093
- Improving LSTM hydrological modeling with spatiotemporal deep learning and multi-task learning: A case study of three mountainous areas on the Tibetan Plateau B. Li et al. 10.1016/j.jhydrol.2023.129401
- PatagoniaMet: A multi-source hydrometeorological dataset for Western Patagonia R. Aguayo et al. 10.1038/s41597-023-02828-2
- River flooding mechanisms and their changes in Europe revealed by explainable machine learning S. Jiang et al. 10.5194/hess-26-6339-2022
- Physics-informed machine learning method for modelling transport of a conservative pollutant in surface water systems D. Bertels & P. Willems 10.1016/j.jhydrol.2023.129354
- On strictly enforced mass conservation constraints for modelling the Rainfall‐Runoff process J. Frame et al. 10.1002/hyp.14847
- Observation‐Constrained Projection of Flood Risks and Socioeconomic Exposure in China S. Kang et al. 10.1029/2022EF003308
- Can machine learning accelerate process understanding and decision‐relevant predictions of river water quality? C. Varadharajan et al. 10.1002/hyp.14565
- Assessing the Physical Realism of Deep Learning Hydrologic Model Projections Under Climate Change S. Wi & S. Steinschneider 10.1029/2022WR032123
- Enhancing streamflow simulation in large and human-regulated basins: Long short-term memory with multiscale attributes A. Tursun et al. 10.1016/j.jhydrol.2024.130771
- Differentiable, Learnable, Regionalized Process‐Based Models With Multiphysical Outputs can Approach State‐Of‐The‐Art Hydrologic Prediction Accuracy D. Feng et al. 10.1029/2022WR032404
- Hydrological concept formation inside long short-term memory (LSTM) networks T. Lees et al. 10.5194/hess-26-3079-2022
- A Deep Learning Approach Based on Physical Constraints for Predicting Soil Moisture in Unsaturated Zones Y. Wang et al. 10.1029/2023WR035194
34 citations as recorded by crossref.
- On how data are partitioned in model development and evaluation: Confronting the elephant in the room to enhance model generalization H. Maier et al. 10.1016/j.envsoft.2023.105779
- Hybrid forecasting: blending climate predictions with AI models L. Slater et al. 10.5194/hess-27-1865-2023
- Developing a Physics‐Informed Deep Learning Model to Simulate Runoff Response to Climate Change in Alpine Catchments L. Zhong et al. 10.1029/2022WR034118
- Continuous streamflow prediction in ungauged basins: long short-term memory neural networks clearly outperform traditional hydrological models R. Arsenault et al. 10.5194/hess-27-139-2023
- Short-term runoff forecasting in an alpine catchment with a long short-term memory neural network C. Frank et al. 10.3389/frwa.2023.1126310
- Pluvial flood susceptibility mapping for data-scarce urban areas using graph attention network and basic flood conditioning factors Z. Wang et al. 10.1080/10106049.2023.2275692
- Improving hydrologic models for predictions and process understanding using neural ODEs M. Höge et al. 10.5194/hess-26-5085-2022
- Investigation of hydrometeorological influences on reservoir releases using explainable machine learning methods M. Fan et al. 10.3389/frwa.2023.1112970
- On the need for physical constraints in deep learning rainfall–runoff projections under climate change: a sensitivity analysis to warming and shifts in potential evapotranspiration S. Wi & S. Steinschneider 10.5194/hess-28-479-2024
- Improving short-term streamflow forecasting by flow mode clustering S. Liu et al. 10.1007/s00477-022-02367-z
- Toward interpretable LSTM-based modeling of hydrological systems L. De la Fuente et al. 10.5194/hess-28-945-2024
- A hydrological process-based neural network model for hourly runoff forecasting S. Gao et al. 10.1016/j.envsoft.2024.106029
- Enhancing process-based hydrological models with embedded neural networks: A hybrid approach B. Li et al. 10.1016/j.jhydrol.2023.130107
- Information and disinformation in hydrological data across space: The case of streamflow predictions using machine learning A. Gupta 10.1016/j.ejrh.2023.101607
- Development of a Long-Range Hydrological Drought Prediction Framework Using Deep Learning M. Khan & R. Maity 10.1007/s11269-024-03735-w
- Application of Deep Learning in Drainage Systems Monitoring Data Repair—A Case Study Using Con-GRU Model L. He et al. 10.3390/w15081635
- Inverse Modeling for Subsurface Flow Based on Deep Learning Surrogates and Active Learning Strategies N. Wang et al. 10.1029/2022WR033644
- A Data-Driven Approach for Building the Profile of Water Storage Capacity of Soils J. Zhou et al. 10.3390/s23125599
- DeepGR4J: A deep learning hybridization approach for conceptual rainfall-runoff modelling A. Kapoor et al. 10.1016/j.envsoft.2023.105831
- Ensemble and stochastic conceptual data-driven approaches for improving streamflow simulations: Exploring different hydrological and data-driven models and a diagnostic tool D. Hah et al. 10.1016/j.envsoft.2022.105474
- Revisit hydrological modeling in ungauged catchments comparing regionalization, satellite observations, and machine learning approaches R. Dasgupta et al. 10.1016/j.hydres.2023.11.001
- Using Deep Learning Algorithms for Intermittent Streamflow Prediction in the Headwaters of the Colorado River, Texas F. Forghanparast & G. Mohammadi 10.3390/w14192972
- Geospatial Artificial Intelligence (GeoAI) in the Integrated Hydrological and Fluvial Systems Modeling: Review of Current Applications and Trends C. Gonzales-Inca et al. 10.3390/w14142211
- On the use of convolutional deep learning to predict shoreline change E. Gomez-de la Peña et al. 10.5194/esurf-11-1145-2023
- The Great Lakes Runoff Intercomparison Project Phase 4: the Great Lakes (GRIP-GL) J. Mai et al. 10.5194/hess-26-3537-2022
- A Machine Learning-Based Probabilistic Approach for Irrigation Scheduling S. Srivastava et al. 10.1007/s11269-024-03746-7
- Understanding the Way Machines Simulate Hydrological Processes—A Case Study of Predicting Fine-Scale Watershed Response on a Distributed Framework D. Kim et al. 10.1109/TGRS.2023.3285540
- Near‐term forecasts of stream temperature using deep learning and data assimilation in support of management decisions J. Zwart et al. 10.1111/1752-1688.13093
- Improving LSTM hydrological modeling with spatiotemporal deep learning and multi-task learning: A case study of three mountainous areas on the Tibetan Plateau B. Li et al. 10.1016/j.jhydrol.2023.129401
- PatagoniaMet: A multi-source hydrometeorological dataset for Western Patagonia R. Aguayo et al. 10.1038/s41597-023-02828-2
- River flooding mechanisms and their changes in Europe revealed by explainable machine learning S. Jiang et al. 10.5194/hess-26-6339-2022
- Physics-informed machine learning method for modelling transport of a conservative pollutant in surface water systems D. Bertels & P. Willems 10.1016/j.jhydrol.2023.129354
- On strictly enforced mass conservation constraints for modelling the Rainfall‐Runoff process J. Frame et al. 10.1002/hyp.14847
- Observation‐Constrained Projection of Flood Risks and Socioeconomic Exposure in China S. Kang et al. 10.1029/2022EF003308
6 citations as recorded by crossref.
- Can machine learning accelerate process understanding and decision‐relevant predictions of river water quality? C. Varadharajan et al. 10.1002/hyp.14565
- Assessing the Physical Realism of Deep Learning Hydrologic Model Projections Under Climate Change S. Wi & S. Steinschneider 10.1029/2022WR032123
- Enhancing streamflow simulation in large and human-regulated basins: Long short-term memory with multiscale attributes A. Tursun et al. 10.1016/j.jhydrol.2024.130771
- Differentiable, Learnable, Regionalized Process‐Based Models With Multiphysical Outputs can Approach State‐Of‐The‐Art Hydrologic Prediction Accuracy D. Feng et al. 10.1029/2022WR032404
- Hydrological concept formation inside long short-term memory (LSTM) networks T. Lees et al. 10.5194/hess-26-3079-2022
- A Deep Learning Approach Based on Physical Constraints for Predicting Soil Moisture in Unsaturated Zones Y. Wang et al. 10.1029/2023WR035194
Latest update: 08 May 2024
Short summary
Despite the accuracy of deep learning rainfall-runoff models, we are currently uncertain of what these models have learned. In this study we explore the internals of one deep learning architecture and demonstrate that the model learns about intermediate hydrological stores of soil moisture and snow water, despite never having seen data about these processes during training. Therefore, we find evidence that the deep learning approach learns a physically realistic mapping from inputs to outputs.
Despite the accuracy of deep learning rainfall-runoff models, we are currently uncertain of what...
