RS monitoring of drainage outlet anomalies - Suspected Sewage Discharge

Home > Application > Suspected Sewage Discharge > RS monitoring of drainage outlet anomalies

Research Background

The coastal outfalls, as the entrance to directly discharge inland water into the coastal waters through sluices, pipelines, ditches, and other ways, have the characteristics of small area and wide distribution. It is scattered in various areas of the coastal area and is a typical source of terrestrial pollutants into the sea. Traditional on-site investigation and limited online automatic monitoring equipment are incapable of fully and effectively supervising a large number of coastal outfalls, which is a "pain point" in current marine ecological environment protection. Therefore, the development of an efficient technology that can quickly monitor sewage discharge from coastal outfalls over a wide range is expected.

Satellite remote sensing techniques can provide comprehensive and rapid dynamic monitoring on a large scale for coastal outfalls with the advantage of high spatial and temporal resolution. The water color anomalies result in abnormal remote sensing reflectance which can be captured by the satellite sensors. On-site verification based on remote sensing identification results can significantly improve problem detection efficiency and promote terrestrial pollution control and supervision. Meanwhile, long-time series data is useful for tracing water color anomaly events and determining whether they are a one-time occurrence or a long-term sewage phenomenon.

The application of this subject is to automatically find and monitor suspected sewage outfalls that are recognizable from satellite data in nearly real-time. The method can greatly contribute to the management and supervision of the discharge of coastal outfalls and protect the coastal ecological environment through a remote-sensing-based screening and warning system.

unfold

Figures

Fig.4 Typical results of the detection model: suspected highly polluted water from coastal outfalls.
Fig.3 Typical results of the detection model of suspected sewage discharge from coastal outfalls: eutrophic water from aquaculture ponds.
Fig.2 Typical results of the detection model of suspected sewage discharge from coastal outfalls and the on-site investigation photos. (1) Model result with the highlighted green regions on June 9, 2022, which means eutrophic water. (2) On-site investigation photos of the suspected sewage event on June 12, 2022.
Fig. 1 Flowchart of the development of the detection model of suspected sewage discharge from coastal outfalls and demonstrations of model evaluation, application, and on-site verification.

Scientific Progress

The module of automated identification and monitoring of suspected sewage of coastal outfalls by remote sensing techniques produces the optical water type classification product, which assigns a value ranging from 1 to 14 to different water types. In satellite imagery around coastal outfalls, water types 11, 12, and 14 are commonly referred to as suspected water color anomalies. A detailed description of water types from 1 to 14 can be found in Table 1. The spatial resolution of the product is 10 meters, and it covers nearshore areas since the launch of Sentinel-2 satellites in June 2015 with a temporal resolution of 5 days. To assess the accuracy of the model, a confusion matrix was used, revealing only one misclassification out of 9103 spectra. By comparing historical pollution events with corresponding model results, this module has demonstrated the ability to effectively identify suspected sewage discharge from coastal outfalls.

The input data of the suspected sewage discharge product consists of the remote sensing reflectance and multiple spectral indices after atmospheric correction. The model was initially developed based on sewage events in coastal outfalls, utilizing spectral data from the Sentinel-2 imagery captured during different seasons and tidal levels. Additionally, spectra from high turbidity water, clean waters far from the shore, reservoirs, rivers, aquaculture ponds, and adjacent shore water were extracted from multi-temporal images to establish a spectral library for the optical water type classification model. The unsupervised classification was then performed based on the spectral library, determining the optical water types corresponding to each cluster according to their distinct spectral characteristics. Subsequently, an optical water type classification model based on a random forest algorithm was established. Based on the characteristics of the spectral peak, valley, or variation trend of each type, the spectral features with high sensitivity for model input were determined. Through multiple iterative optimizations of different parameters, we finally determined an optical classification model with high accuracy and produced the optical water type classification map with 10 m resolution to identify coastal outfalls with anomaly water color. Compared to general optical water type classification products, this module focuses specifically on coastal waters and takes the spectral characteristics of various small-scale water bodies into account, including polluted water. As a result, it could achieve higher precision. The module can be applied to marine ecological environment monitoring and supervision, providing technical support for departments involved in the management and oversight of widespread coastal outfalls.

Table 1 Description of each optical water type

Optical water type	1	2	3	4	5	6	7	8	9	10	11	12	13	14
Water types	Clean water				Turbid water						Eutrophic water		High turbidity water, or tidal flats	Suspected polluted water (sewage)

unfold

References

Wang, Yuxin, Xianqiang He, Yan Bai, Yingyu Tan, Bozhong Zhu, Difeng Wang, Mengyuan Ou, Fang Gong, Qiankun Zhu, and Haiqing Huang. 2022. “Automatic Detection of Suspected Sewage Discharge from Coastal Outfalls Based on Sentinel-2 Imagery.” Science of The Total Environment 853. https://doi.org/10.1016/j.scitotenv.2022.158374.

unfold

Research Background

Figures

Scientific Progress

References

Links