River network water quality and pollution tracing - River Water Quality&Pollution

Home > Application > River Water Quality&Pollution > River network water quality and pollution tracing

Research Background

The impacts of urbanization and human activities on the deterioration of river water quality, which leads to a series of ecological problems, have prompted the government to take measures to set up many monitoring stations in the corresponding sections of rivers. These methods can obtain water quality information on surroundings and obtain the pollution sources area around. Due to the limited measurements obtained for a few river sections, the requirements of spatially traceable water pollutants on long-term and large-area scale can’t meet just through monitoring stations.

Through the in-orbit operation of optical sensors that collect data at a high spatial resolution, high temporal resolution and high signal-to-noise ratio, Sentinel-2 satellite images make it possible to conduct large-scale water quality monitoring of small to large rivers. Usually, the monitoring of river water quality mainly focuses on the three most important parameters, namely, chemical oxygen demand (COD), total phosphorus (TP), and total nitrogen (TN). However, very few methods have been developed for detecting non-optically active water quality parameters (e.g., nutrients) because these parameters have no direct relationship with satellite-measured water spectral signals.

This topic established specific models to retrieve water quality parameters (COD, TP, and TN) in Zhejiang Province’s seven rivers from the Sentinel-2 satellite images. Based on the satellite-derived products on water quality parameters, we then studied the annual-average and seasonal-average spatial distributions of COD, TP, and TN. By drawing the spatial traceability map of various water quality parameters from downstream to upstream, we can get the local high-value area along the mainstream, and determine where the high-value area is located, which should be helpful for taking effective pollution prevention measures for the local high-value section.

unfold

Figures

Fig1 Multi-year average spatial distribution results of chemical oxygen demand (COD), total phosphorus, and total nitrogen in certain rivers. (a)~(c) represent the multi-year average spatial distribution results of chemical oxygen demand (COD), total phosphorus, and total nitrogen in the Qiantang River. (d)~(f) represent the multi-year average spatial distribution results of chemical oxygen demand (COD), total phosphorus, and total nitrogen in the Cao’e River. (g)~(f) represent the multi-year average spatial distribution results of chemical oxygen demand (COD), total phosphorus, and total nitrogen in the Yong River.
Fig2 Tracing Map along the Cao`e River

Scientific Progress

The dataset of river water quality parameters includes the permanganate index (COD_Mn), the total phosphorus (TP) and the total nitrogen (TN), with the unit of mg / L. The dataset has a high spatial resolution of 10 m × 10 m, covering main rivers in the Zhejiang Province (the Qiantang River, the Cao 'e River, the Yongjiang River, the Jiaojiang River, the Oujiang River, the Feiyunjiu River and the Aojiang River) from January 2016 to July 2023. The daily average monitoring water quality parameters of automatic monitoring stations and the model retrieval water quality parameters were verified. Correlation coefficients of the permanganate index, the total phosphorus and the total nitrogen were 0.68, 0.82, and 0.7 respectively.

The dataset uses the remote sensing reflectance of Sentinel-2 band obtained after atmospheric correction as input. Spectral bands B2 - B8a are regarded as candidates for the model construction since each band has different spatial resolutions. Rivers with similar characteristics ---similar characteristics of the river basin and similar water quality parameters are classified into the same model. The input parameters derived for each model were obtained by calculating the optimal correlation coefficient between the single band and the band ratio and the in-situ value. Machine learning algorithms such as Gaussian process regression model, support vector machine regression model, linear regression model, along with the five-fold cross-validation method were used to construct this model. Using the machine learning models constructed by optimal input parameters, the daily average permanganate index, total phosphorus and total nitrogen products of main rivers in the Zhejiang Province have been produced during 2016-2023 (every 5 days). Regionally, it can provide high spatial resolution river water quality parameter products, which can be used to trace the spatial source of water pollution in main rivers in the Zhejiang Province and can be helpful to the precise control of river water quality.

unfold

References

Zhang, Yingyin, Xianqiang He, Gang Lian, Yan Bai, Ying Yang, Fang Gong, Difeng Wang, Zili Zhang, Teng Li, and Xuchen Jin. 2023. “Monitoring and Spatial Traceability of River Water Quality Using Sentinel-2 Satellite Images.” Science of The Total Environment 894: 164862.https://doi.org/10.1016/j.scitotenv.2023.164862.

unfold

Research Background

Figures

Scientific Progress

References

Links