Abstract's details
Yangtze intermediate basin lakes and sub-lakes monitoring based on SWOT products
Event: 2025 SWOT Science Team Meeting
Session: Hydrology: Open Science & Applications
Presentation type: Poster
Lakes in the basin of the Yangtze River, play a fundamental role in regional bio-geochemical cycles and provide major services to the communities, provisioning services (drinking water, fishing) and biodiversity keeping. However, the extreme temporal and spatial variability of these massive but extremely shallow ecosystems prevents a reliable quantification of their dynamics with respect to changes in climate and land use.
Three sets of water bodies presenting diverse hydrological context and all having rich biodiversity are taken into account:
- The Poyang lake and it’s interconnected sub-lakes of the Poyang (PLNR) and of Nanjishan Natural Reserves.
- The Xiaoxi, Daxi, Caisang sub-lakes located into the Western Dongting lake, Hunan Province.
- The disconnected lakes of the Anhui Province Caizi lake (~175 Km2), Baidang Lake (~73 Km2), Shengjin Lake (~ 120 Km2) and Wuchang Lake (~ 73 Km2).
A first step of the study carried out within the framework of the ESA MOST DRAGON program consisted in setting up a reference database in terms of water extent and height, over the different targets, in order to be able to characterize the dynamics of these areas, their interrelationship as well as the maximum extent of these water bodies.
Water extents were derived from Sentinel-2 time series, i.e. based on the analysis of about 150 images for each of the Anhui province lakes, as well as a few additional SAR datasets, Sentinel-1, IceEye, Radarsat-2, allowing to cover some critical periods such as the 2020 severe flood event giving access to the water maximum extent mask. Altitude information was derived from Sentinel-3 times series, as well as a few ICESat-2 data. It is noticeable that the spatial coverage of altimetry data is very weak and only a few lakes were covered by Sentinal-3 tracks. A particular attention was paid over the Baidang and Shengjin Lakes.
In a second step, the SWOT Orbit Science data were analyzed. Based on two SWOT products to retrieve water height levels and lake surface areas: L2_HR_PIXC and L2_HR_LakeSP during the scientific phase starting from July 2023. In this paper we present only the altitude assessment work during the filling period, some LakeSP observations tend to incorporate neighbouring bodies with the main water extraction of the lake, resulting in incorrect/spoiled surface areas. Water surface elevations were generated from the L2_HR_PIXC products by computing the mean height of PIXC class 4 points (Open Water) per acquisition over the polygons defining the 7 and 6 sub-lakes both for the Nanjishan Natural reserve and Poyang lake National reserve.
The analysis of SWOT products was carried out as follows:
• Evaluation of the accuracy of the PIXC classes (water only, land and water, dark water, etc).
• Outliers removal.
• Exploitation of Bright Land and Dark Water flags.
The first step was to check whether SWOT could be used to monitor the filling/emptying cycles of the sub-lakes, bearing in mind that during the dry season these sub-lakes are at different altitudes within the same sub-set constituting the Natural reserves.
At all the sites, the SWOT data clearly identifies the sub-assemblies during the dry season, highlighting the differences in altitude (between the different sub-lakes: i.e. 2 m of elevation difference for the Nanjishan case, 4 m in the case of PLNR sub-lakes. When the water level rises during the rainy season, there is a clear homogenization of altitudes, corresponding to the sub-lakes flooding, forming a single vast complex. In addition, in the Dongting lake sector, the sub-systems are clearly distinct, with Caising lake separated from Dongting lake and the other sub-lakes by a high dyke.
In terms of accuracy, the SWOT altitudes were compared with water elevation derived from ICESat-2 and Sentinel-3. Based on this, accuracy is better than 10 cm (at 1 sigma).
In addition it was possible to access to in-situ water level from DuChang station, corresponding to the central area of the lake. These data were referenced in Wusong Datum for which the transposition in WGS84 is not so simple and accurate. We can observe that the lake water elevation dynamics derived from the SWOT PIXC and the in situ data are very similar, presenting the same peaks and trends but with a mean residual bias of 1.48 m. When this bias is removed the two curves are merged.
This work presents the first exploitation of SWOT products over Chinese sensitive and threatened ecosystems. Over these complex hydrographical systems, SWOT allows a unique monitoring of LWE and LWL sub-lake by sub-lake with never unreached accuracy and revisit. Therefore SWOT LakeSP is very sensitive to the neighborhood of these wetlands and complicated lakes inducing incorrect/spoiled surface areas.
Three sets of water bodies presenting diverse hydrological context and all having rich biodiversity are taken into account:
- The Poyang lake and it’s interconnected sub-lakes of the Poyang (PLNR) and of Nanjishan Natural Reserves.
- The Xiaoxi, Daxi, Caisang sub-lakes located into the Western Dongting lake, Hunan Province.
- The disconnected lakes of the Anhui Province Caizi lake (~175 Km2), Baidang Lake (~73 Km2), Shengjin Lake (~ 120 Km2) and Wuchang Lake (~ 73 Km2).
A first step of the study carried out within the framework of the ESA MOST DRAGON program consisted in setting up a reference database in terms of water extent and height, over the different targets, in order to be able to characterize the dynamics of these areas, their interrelationship as well as the maximum extent of these water bodies.
Water extents were derived from Sentinel-2 time series, i.e. based on the analysis of about 150 images for each of the Anhui province lakes, as well as a few additional SAR datasets, Sentinel-1, IceEye, Radarsat-2, allowing to cover some critical periods such as the 2020 severe flood event giving access to the water maximum extent mask. Altitude information was derived from Sentinel-3 times series, as well as a few ICESat-2 data. It is noticeable that the spatial coverage of altimetry data is very weak and only a few lakes were covered by Sentinal-3 tracks. A particular attention was paid over the Baidang and Shengjin Lakes.
In a second step, the SWOT Orbit Science data were analyzed. Based on two SWOT products to retrieve water height levels and lake surface areas: L2_HR_PIXC and L2_HR_LakeSP during the scientific phase starting from July 2023. In this paper we present only the altitude assessment work during the filling period, some LakeSP observations tend to incorporate neighbouring bodies with the main water extraction of the lake, resulting in incorrect/spoiled surface areas. Water surface elevations were generated from the L2_HR_PIXC products by computing the mean height of PIXC class 4 points (Open Water) per acquisition over the polygons defining the 7 and 6 sub-lakes both for the Nanjishan Natural reserve and Poyang lake National reserve.
The analysis of SWOT products was carried out as follows:
• Evaluation of the accuracy of the PIXC classes (water only, land and water, dark water, etc).
• Outliers removal.
• Exploitation of Bright Land and Dark Water flags.
The first step was to check whether SWOT could be used to monitor the filling/emptying cycles of the sub-lakes, bearing in mind that during the dry season these sub-lakes are at different altitudes within the same sub-set constituting the Natural reserves.
At all the sites, the SWOT data clearly identifies the sub-assemblies during the dry season, highlighting the differences in altitude (between the different sub-lakes: i.e. 2 m of elevation difference for the Nanjishan case, 4 m in the case of PLNR sub-lakes. When the water level rises during the rainy season, there is a clear homogenization of altitudes, corresponding to the sub-lakes flooding, forming a single vast complex. In addition, in the Dongting lake sector, the sub-systems are clearly distinct, with Caising lake separated from Dongting lake and the other sub-lakes by a high dyke.
In terms of accuracy, the SWOT altitudes were compared with water elevation derived from ICESat-2 and Sentinel-3. Based on this, accuracy is better than 10 cm (at 1 sigma).
In addition it was possible to access to in-situ water level from DuChang station, corresponding to the central area of the lake. These data were referenced in Wusong Datum for which the transposition in WGS84 is not so simple and accurate. We can observe that the lake water elevation dynamics derived from the SWOT PIXC and the in situ data are very similar, presenting the same peaks and trends but with a mean residual bias of 1.48 m. When this bias is removed the two curves are merged.
This work presents the first exploitation of SWOT products over Chinese sensitive and threatened ecosystems. Over these complex hydrographical systems, SWOT allows a unique monitoring of LWE and LWL sub-lake by sub-lake with never unreached accuracy and revisit. Therefore SWOT LakeSP is very sensitive to the neighborhood of these wetlands and complicated lakes inducing incorrect/spoiled surface areas.
Contribution: ST2025HS6-Yangtze_intermediate_basin_lakes_and_sub-lakes_monitoring_based_on_SWOT_products.pdf (pdf, 1945 ko)
Back to the list of abstract