Assessing surface water pollution in Hanoi, Vietnam, using remote sensing and machine learning algorithms

Area of Science:

• Environmental Science
• Remote Sensing
• Machine Learning

Background:

• Rapid urbanization worldwide, particularly in the Global South, has led to significant land-use changes and threats to surface water bodies.
• Hanoi, Vietnam, experiences chronic surface water pollution, necessitating advanced monitoring methods.
• Effective management of water resources requires improved tracking and analysis of pollutants.

Purpose of the Study:

• To introduce and evaluate a novel machine learning model (ML-CB) for estimating key surface water pollutants.
• To assess the efficacy of combining optical and RADAR satellite data for water quality monitoring.
• To provide an alternative method for water quality assessment for urban planners and managers.

Main Methods:

• Developed a machine learning model with the cubist algorithm (ML-CB).
• Integrated optical (Sentinel-2A) and RADAR (Sentinel-1A) satellite imagery for pollutant estimation.
• Validated model predictions against field survey data using regression analysis.

Main Results:

• The ML-CB model demonstrated significant accuracy in estimating surface water pollutants.
• The integration of optical and RADAR data proved effective for water quality assessment.
• Predictive estimates of total suspended sediments (TSS), chemical oxygen demand (COD), and biological oxygen demand (BOD) were achieved.

Conclusions:

• The ML-CB model offers a viable and effective alternative for surface water quality monitoring.
• This approach can support sustainable water resource management in urban areas, especially in the Global South.
• The study highlights the potential of combining machine learning and earth observation for environmental management.