Coastal Water Quality Modelling Using E. coli, Meteorological Parameters and Machine Learning Algorithms.

Area of Science:

• Environmental Science
• Machine Learning
• Water Quality Management

Background:

• Coastal water quality monitoring is crucial for public health and ecosystem integrity.
• Directive 2006/7/EC mandates regular assessment of bathing water quality.
• Escherichia coli (E. coli) is a key indicator of fecal contamination in marine environments.

Purpose of the Study:

• To implement and evaluate machine learning models for predicting coastal water classification based on E. coli concentration.
• To assess the influence of meteorological variables on E. coli levels in coastal waters.
• To classify the water quality of Eastern Macedonia and Thrace (EMT) coastal areas according to EU standards.

Main Methods:

• Collected 1039 water samples from six sampling stations in EMT between 2009-2021 (May-September).
• Analyzed E. coli using ISO 9308-1 standard.
• Acquired meteorological data from nearby stations.
• Applied machine learning classifiers including Decision Forest, Decision Jungle, and Boosted Decision Tree.

Main Results:

• The vast majority of samples were classified as Category 1 (Excellent).
• Decision Forest, Decision Jungle, and Boosted Decision Tree classifiers achieved accuracy scores exceeding 99%.
• Comparison with other studies shows diverse machine learning algorithms (Decision Tree, Artificial Neural Networks, Bayesian Belief Networks) yield satisfactory results for water quality prediction.

Conclusions:

• Machine learning models effectively predict coastal water quality and E. coli contamination dynamics.
• Meteorological parameters can be integrated into water quality classification models.
• The coastal waters of EMT demonstrate excellent quality, with high predictability using advanced computational methods.