Surface water temperature prediction in large-deep reservoirs using a long short-term memory model
Surface water temperature (SWT) is a key indicator to characterize the ecological health of a reservoir. Many newly built large-deep reservoirs, however, lack enough SWT observation data and high-efficient SWT predicting methods for water ecosystem management. This paper proposed a Long Short-Term M...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-01-01
|
| Series: | Ecological Indicators |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1470160X21011560 |
| _version_ | 1828923553737605120 |
|---|---|
| author | Longfan Wang Bo Xu Chi Zhang Guangtao Fu Xiaoxian Chen Yi Zheng Jingjie Zhang |
| author_facet | Longfan Wang Bo Xu Chi Zhang Guangtao Fu Xiaoxian Chen Yi Zheng Jingjie Zhang |
| author_sort | Longfan Wang |
| collection | DOAJ |
| description | Surface water temperature (SWT) is a key indicator to characterize the ecological health of a reservoir. Many newly built large-deep reservoirs, however, lack enough SWT observation data and high-efficient SWT predicting methods for water ecosystem management. This paper proposed a Long Short-Term Memory (LSTM) based SWT predicting method by surrogating a Delft3D hydrodynamic model. The Delft3D model calibrated by a handful of measured data was used to generate 30-year daily SWT data for training the LSTM model. The LSTM model that uses the air temperature, relative humidity, radiation, and water level data as input variables, can significantly improve the efficiency of SWT prediction. The SWT predicting method was implemented in the Nuozhadu reservoir, a large deep reservoir located in southwest China. The results showed that the LSTM model could predict the SWT generated by Delft3D accurately with an R2 value of 0.99, and had a dramatic reduction in computational burden. Meanwhile, the R2 between the LSTM model results and measured data was also over 0.93. Based on the SWT predicting method, we analyzed the sensitivity of SWT to the air temperature and water level to reveal the impacts of climate change and reservoir operation policies on the SWT. The major contribution of this study is that we greatly improve the computational efficiency of the SWT predicting method so that it can easily be coupled with reservoir operation optimization models, thereby enabling reservoir managers to identify optimal operation rules simultaneously considering the water temperature targets and other targets such as hydropower generation and water supply, and to eventually support reservoir management strategies that aim to reduce the potential aquatic ecosystems risk from abnormal water temperature. |
| first_indexed | 2024-12-13T22:40:00Z |
| format | Article |
| id | doaj.art-9b463e6e67cd40e59e8fc2eba8f785d0 |
| institution | Directory Open Access Journal |
| issn | 1470-160X |
| language | English |
| last_indexed | 2024-12-13T22:40:00Z |
| publishDate | 2022-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Ecological Indicators |
| spelling | doaj.art-9b463e6e67cd40e59e8fc2eba8f785d02022-12-21T23:28:52ZengElsevierEcological Indicators1470-160X2022-01-01134108491Surface water temperature prediction in large-deep reservoirs using a long short-term memory modelLongfan Wang0Bo Xu1Chi Zhang2Guangtao Fu3Xiaoxian Chen4Yi Zheng5Jingjie Zhang6School of Hydraulic Engineering, Dalian University of Technology, Dalian 116024, Liaoning, ChinaSchool of Hydraulic Engineering, Dalian University of Technology, Dalian 116024, Liaoning, China; Corresponding author.School of Hydraulic Engineering, Dalian University of Technology, Dalian 116024, Liaoning, ChinaCentre for Water Systems, University of Exeter, Exeter EX4 4QF, UKSchool of Hydraulic Engineering, Dalian University of Technology, Dalian 116024, Liaoning, ChinaSchool of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, Guangdong, ChinaSchool of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, Guangdong, China; Department of Civil and Environmental Engineering, National University of Singapore, 117577, SingaporeSurface water temperature (SWT) is a key indicator to characterize the ecological health of a reservoir. Many newly built large-deep reservoirs, however, lack enough SWT observation data and high-efficient SWT predicting methods for water ecosystem management. This paper proposed a Long Short-Term Memory (LSTM) based SWT predicting method by surrogating a Delft3D hydrodynamic model. The Delft3D model calibrated by a handful of measured data was used to generate 30-year daily SWT data for training the LSTM model. The LSTM model that uses the air temperature, relative humidity, radiation, and water level data as input variables, can significantly improve the efficiency of SWT prediction. The SWT predicting method was implemented in the Nuozhadu reservoir, a large deep reservoir located in southwest China. The results showed that the LSTM model could predict the SWT generated by Delft3D accurately with an R2 value of 0.99, and had a dramatic reduction in computational burden. Meanwhile, the R2 between the LSTM model results and measured data was also over 0.93. Based on the SWT predicting method, we analyzed the sensitivity of SWT to the air temperature and water level to reveal the impacts of climate change and reservoir operation policies on the SWT. The major contribution of this study is that we greatly improve the computational efficiency of the SWT predicting method so that it can easily be coupled with reservoir operation optimization models, thereby enabling reservoir managers to identify optimal operation rules simultaneously considering the water temperature targets and other targets such as hydropower generation and water supply, and to eventually support reservoir management strategies that aim to reduce the potential aquatic ecosystems risk from abnormal water temperature.http://www.sciencedirect.com/science/article/pii/S1470160X21011560Surface water temperatureLarge-deep reservoirLSTMDelft3D |
| spellingShingle | Longfan Wang Bo Xu Chi Zhang Guangtao Fu Xiaoxian Chen Yi Zheng Jingjie Zhang Surface water temperature prediction in large-deep reservoirs using a long short-term memory model Ecological Indicators Surface water temperature Large-deep reservoir LSTM Delft3D |
| title | Surface water temperature prediction in large-deep reservoirs using a long short-term memory model |
| title_full | Surface water temperature prediction in large-deep reservoirs using a long short-term memory model |
| title_fullStr | Surface water temperature prediction in large-deep reservoirs using a long short-term memory model |
| title_full_unstemmed | Surface water temperature prediction in large-deep reservoirs using a long short-term memory model |
| title_short | Surface water temperature prediction in large-deep reservoirs using a long short-term memory model |
| title_sort | surface water temperature prediction in large deep reservoirs using a long short term memory model |
| topic | Surface water temperature Large-deep reservoir LSTM Delft3D |
| url | http://www.sciencedirect.com/science/article/pii/S1470160X21011560 |
| work_keys_str_mv | AT longfanwang surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel AT boxu surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel AT chizhang surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel AT guangtaofu surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel AT xiaoxianchen surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel AT yizheng surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel AT jingjiezhang surfacewatertemperaturepredictioninlargedeepreservoirsusingalongshorttermmemorymodel |