Work

Monitoring bacterial contamination of West African surface waters using Earth observation data and machine learning methods

Year: 2025

Type: article

Abstract: In 2021, diarrheal diseases caused approximately 434,000 deaths in sub-Saharan Africa, often linked to the consumption of water contaminated with fecal pathogens such as Escherichia coli. Ongoing environmental and anthropogenic changes are likely to exacerbate bacterial contamination of surface waters, increasing public health risks. In this context, the ability to monitor contamination without relying solely on field-based analyses is of growing interest. This study explores the potential of Earth observation data to monitor bacterial contamination in surface waters in West Africa. It investigates the relationship between E. coli concentrations and various water quality parameters, some measured in situ (biochemical parameters, suspended particulate matter, etc.), and others derived from Earth observation data (rainfall, specific humidity, etc.). These variables are integrated into eight machine learning models simulating the temporal dynamics of E. coli concentrations in two sites: the Bagré reservoir (Burkina Faso) and Kongou Lake (Niger). The models are first trained using the full set of variables, and then using only routinely available satellite-derived data. The best-performing model for both sites is the Extremely Randomized Trees algorithm, achieving an R² of 0.78 for Bagré and 0.645 for Kongou when using all available data. When limited to satellite-derived variables, the model still performed well, with R² values of 0.667 for Bagré and 0.642 for Kongou. These results demonstrate the feasibility of using Earth observation data alone to monitor E. coli contamination in West African surface waters. This approach can enable remote monitoring of microbial water quality.

Source: Environmental Advances

DOI: 10.1016/j.envadv.2025.100678