Leveraging multi-source satellite imagery for improving daily reservoir water storage and inflow estimation in a catchment hydrological model
(2025) In Journal of Hydrology 661.- Abstract
Accurate reservoir dynamics, particularly water storage and inflow, are crucial for release decisions and operation rule assessments. However, the lack of sufficient monitoring data has hindered the accurate representation of reservoir behavior in hydrological models. While the increasing availability of satellite imagery provides unprecedented opportunities for dynamic reservoir monitoring, its potential to enhance hydrological models for simulating daily storage and inflow remains largely unexplored. This study aims to evaluate the applicability of optical (Landsat 8 and Sentinel-2) and Synthetic Aperture Radar (SAR, Sentinel-1) satellite imagery in extracting reservoir-specific Surface Water Extent (SWE), using these data as a... (More)
Accurate reservoir dynamics, particularly water storage and inflow, are crucial for release decisions and operation rule assessments. However, the lack of sufficient monitoring data has hindered the accurate representation of reservoir behavior in hydrological models. While the increasing availability of satellite imagery provides unprecedented opportunities for dynamic reservoir monitoring, its potential to enhance hydrological models for simulating daily storage and inflow remains largely unexplored. This study aims to evaluate the applicability of optical (Landsat 8 and Sentinel-2) and Synthetic Aperture Radar (SAR, Sentinel-1) satellite imagery in extracting reservoir-specific Surface Water Extent (SWE), using these data as a calibration reference to constrain the Xinanjiang (XAJ) hydrological model for estimating daily reservoir water storage and inflow. To improve the accuracy of SWE identification, we use the random forest classification (RFC) for optical imagery and the OTSU algorithm for SAR imagery to distinguish water from non-water areas. Our study focuses on a large, multi-year regulating reservoir located in the upper reaches of the Wujiang River in China, namely the Hongjiadu Reservoir. The results show that: (1) SWE of the Hongjiadu Reservoir can be accurately extracted from satellite imagery using the RFC and OTSU algorithms, particularly for Sentinel-1; (2) Using Sentinel-1-derived SWE as a calibration reference, the XAJ model performs excellently in simulating daily inflow, with a Nash-Sutcliffe Efficiency (NSE) of 0.92, and in reconstructing daily water storage volume with NSE of 0.98; and (3) A merged dataset of Sentinel-1, Sentinel-2, and Landsat 8 outperforms Sentinel-2 and Landsat 8 alone in reservoir area identification, water storage change estimation, and the XAJ simulation, likely due to the greater availability and accuracy of Sentinel-1. These findings highlight the potential of satellite imagery in constraining hydrological models to simulate daily reservoir behaviors in highly regulated basins. This approach provides a crucial supplement or alternative to traditional streamflow gauging stations and opens new avenues for improving water resources management in data-scarce regions.
(Less)
- author
- Ma, Qiumei ; Gui, Xu ; Duan, Zheng LU ; Li, Jiqing ; Li, Yuying and Xiong, Lihua
- organization
- publishing date
- 2025-11
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Hydrological modeling, Model calibration, Reservoir water storage, Satellite imagery, Surface water extent
- in
- Journal of Hydrology
- volume
- 661
- article number
- 133761
- publisher
- Elsevier
- external identifiers
-
- scopus:105009509675
- ISSN
- 0022-1694
- DOI
- 10.1016/j.jhydrol.2025.133761
- language
- English
- LU publication?
- yes
- id
- 8a53d8d9-89f2-46a3-b77b-ec82fe8e373c
- date added to LUP
- 2025-10-27 14:02:13
- date last changed
- 2025-10-27 15:56:53
@article{8a53d8d9-89f2-46a3-b77b-ec82fe8e373c,
abstract = {{<p>Accurate reservoir dynamics, particularly water storage and inflow, are crucial for release decisions and operation rule assessments. However, the lack of sufficient monitoring data has hindered the accurate representation of reservoir behavior in hydrological models. While the increasing availability of satellite imagery provides unprecedented opportunities for dynamic reservoir monitoring, its potential to enhance hydrological models for simulating daily storage and inflow remains largely unexplored. This study aims to evaluate the applicability of optical (Landsat 8 and Sentinel-2) and Synthetic Aperture Radar (SAR, Sentinel-1) satellite imagery in extracting reservoir-specific Surface Water Extent (SWE), using these data as a calibration reference to constrain the Xinanjiang (XAJ) hydrological model for estimating daily reservoir water storage and inflow. To improve the accuracy of SWE identification, we use the random forest classification (RFC) for optical imagery and the OTSU algorithm for SAR imagery to distinguish water from non-water areas. Our study focuses on a large, multi-year regulating reservoir located in the upper reaches of the Wujiang River in China, namely the Hongjiadu Reservoir. The results show that: (1) SWE of the Hongjiadu Reservoir can be accurately extracted from satellite imagery using the RFC and OTSU algorithms, particularly for Sentinel-1; (2) Using Sentinel-1-derived SWE as a calibration reference, the XAJ model performs excellently in simulating daily inflow, with a Nash-Sutcliffe Efficiency (NSE) of 0.92, and in reconstructing daily water storage volume with NSE of 0.98; and (3) A merged dataset of Sentinel-1, Sentinel-2, and Landsat 8 outperforms Sentinel-2 and Landsat 8 alone in reservoir area identification, water storage change estimation, and the XAJ simulation, likely due to the greater availability and accuracy of Sentinel-1. These findings highlight the potential of satellite imagery in constraining hydrological models to simulate daily reservoir behaviors in highly regulated basins. This approach provides a crucial supplement or alternative to traditional streamflow gauging stations and opens new avenues for improving water resources management in data-scarce regions.</p>}},
author = {{Ma, Qiumei and Gui, Xu and Duan, Zheng and Li, Jiqing and Li, Yuying and Xiong, Lihua}},
issn = {{0022-1694}},
keywords = {{Hydrological modeling; Model calibration; Reservoir water storage; Satellite imagery; Surface water extent}},
language = {{eng}},
publisher = {{Elsevier}},
series = {{Journal of Hydrology}},
title = {{Leveraging multi-source satellite imagery for improving daily reservoir water storage and inflow estimation in a catchment hydrological model}},
url = {{http://dx.doi.org/10.1016/j.jhydrol.2025.133761}},
doi = {{10.1016/j.jhydrol.2025.133761}},
volume = {{661}},
year = {{2025}},
}