Recently, the research team led by Researcher He Xianqiang from our laboratory, together with collaborators, has published a research paper entitled Satellite water quality monitoring of China’s coastal inland waters based on Google Earth Engine in the Journal of Hydrology. The co-first authors of this paper are Zhao Yaqi and Yan Yujia, joint PhD candidates cultivated by our laboratory and Zhejiang University; the corresponding author is Researcher He Xianqiang. The co-authors include Researcher Bai Yan, Associate Researchers Li Teng and Jin Xuchen, Senior Engineer Gong Fang from our laboratory, Researcher Jiang Fajun from Guangxi Academy of Sciences, PhD candidate Chen Zhiyan jointly trained by our laboratory and Shanghai Jiao Tong University, and PhD candidate Zhao Liao from the University of Tasmania.
Coastal water systems serve as a critical linkage for land-ocean interactions, and their water quality directly determines the ecological security of estuarine and nearshore zones, the assessment of land-sourced pollutant transport, as well as the conservation and management of water resources. China’s coastal regions feature dense populations and intensive human economic activities, alongside complex inland water bodies such as rivers, lakes, reservoirs and aquaculture ponds, accompanied by prominent spatiotemporal variations in water quality. Traditional automatic monitoring stations can provide high-frequency and accurate in-situ point observations, yet they fail to achieve full spatial coverage over numerous small-and-medium-sized rivers, lakes and reservoirs. Satellite remote sensing, characterized by spatially continuous coverage and periodic observations, acts as a vital technical support for large-scale water quality monitoring. Nevertheless, coastal water bodies feature complex optical properties and diverse pollution sources, making it difficult to directly retrieve non-optically active water quality parameters including permanganate index, total nitrogen (TN) and total phosphorus (TP) from remote sensing spectra. In addition, processing large-scale high-resolution remote sensing datasets is plagued by heavy computational loads and obstacles to operational application. Therefore, developing water quality remote sensing monitoring approaches with satisfactory accuracy, strong transferability and large-scale applicability has become a key challenge for current aquatic environment remote sensing research and its operational deployment.
Targeting the demand for full-coverage and high-frequency water quality monitoring of inland waters across China’s coastal zones, this study constructs a multi-parameter water quality remote sensing inversion framework based on Google Earth Engine (GEE). Relying on Sentinel-2 satellite remote sensing reflectance and in-situ measurements from national automatic surface water quality monitoring stations, the research establishes a large-scale matched satellite-ground observation dataset. Specifically, 12,650 matched samples for permanganate index, 12,610 for total nitrogen, 12,223 for total phosphorus, and 12,528 for turbidity are obtained, providing sufficient data foundation for developing a unified multi-parameter remote sensing inversion model for coastal water quality nationwide (Figure 1).
On this basis, the team develops lightweight multi-modal remote sensing inversion models for four core water quality indicators: turbidity, permanganate index, total nitrogen and total phosphorus (Figure 2). Balancing prediction accuracy and deployment efficiency on the GEE cloud platform, the proposed models realize collaborative modeling integrating optical information, geospatial features and environmental driving factors. Model validation demonstrates that remote sensing retrievals are generally consistent with in-situ measurements in both spatial distribution and temporal evolution, enabling effective characterization of regional disparities and seasonal variations in water quality within China’s coastal inland waters (Figures 3, 4, 5, 6).
This study represents a new advancement building upon the team’s previous breakthroughs in provincial-scale water quality remote sensing technology across Zhejiang Province. Compared with prior regional-scale research, this work further targets extensive and complex inland water bodies along China’s coastline, establishing an efficiently operable multi-parameter water quality remote sensing monitoring framework deployable on cloud platforms. It provides generalizable and transferable model support for the operational application of high-resolution satellite data in aquatic environment monitoring and integrated land-ocean management.
This research was financially supported by the National Key R&D Program of China (Grant No. 2023YFC3108101), the Interdisciplinary Innovation Cultivation Fund for PhD Students under the Basic Scientific Research Fund of Central Public Welfare Research Institutes, Second Institute of Oceanography, Ministry of Natural Resources (BS2601), and the Qiushi Rising Star Training Program for PhD Candidates of Zhejiang University.
Figure 1 Concentration distributions of individual water quality parameters within the matched dataset: (a) permanganate index; (b) total nitrogen; (c) total phosphorus; (d) turbidity.
Figure 2 Technical flowchart of this study.
Figure 3 Scatter plots comparing satellite-retrieved values versus in-situ measured values for each water quality parameter:
(a, b) training set and validation set of permanganate index;
(c, d) training set and validation set of total nitrogen;
(e, f) training set and validation set of total phosphorus;
(g, h) training set and validation set of turbidity.
Figure 4 Variations of satellite retrievals and in-situ measurements along longitude (left column) and latitude (right column). Solid lines denote the mean value within each 0.5° spatial grid, and shaded regions represent ±1 standard deviation:
(a) permanganate index; (b) total nitrogen; (c) total phosphorus; (d) turbidity.
Figure 5 Temporal variations of satellite retrievals and in-situ measurements at monitoring stations. Solid lines denote monthly mean values, and shaded regions represent ±1 standard deviation:
(a) permanganate index; (b) total nitrogen; (c) total phosphorus; (d) turbidity.
Figure 6 Spatial distribution maps of retrieved water quality parameter concentrations for inland waters in China’s coastal provinces (results are averaged within 0.5° grids):
(a) permanganate index; (b) total nitrogen; (c) total phosphorus; (d) turbidity.
Reference List
Primary Article
- Zhao, Y.#, Yan, Y.#, He, X.*, Chen, Z., Li, T., Jin, X., Zhao, L., Gong, F., Jiang, F., Bai, Y. Satellite water quality monitoring of China's coastal inland waters based on Google Earth Engine[J]. Journal of Hydrology, 2026. https://doi.org/10.1016/j.jhydrol.2026.135882
Note: # denotes co-first author; * denotes corresponding author
Related Publications
-
Zhao, Y., He, X.*, Pan, S., Bai, Y., Wang, D., Li, T., Gong, F., Zhang, X. Satellite retrievals of water quality for diverse inland waters from Sentinel-2 images: An example from Zhejiang Province, China[J]. International Journal of Applied Earth Observation and Geoinformation, 2024.
https://doi.org/10.1016/j.jag.2024.104048
-
Yan, Y., He, X.*, Bai, Y., Liu, J., Shanmugame, P., Zhao, Y., Zhang, X., Wang, Z., Zhang, Y., Gong, F. Monitoring Dissolved Organic Carbon Concentration and Flux in the Qiantang Riverine System Using Sentinel-2 Satellite Images[J]. Remote Sensing, 2024.
https://doi.org/10.3390/rs16224254
