Local knowledge: a valuable tool for microbial water quality monitoring and risk assessment?

Drinking water is a fundamental necessity for people worldwide. However, surface water pollution poses challenges for water supply management and highlights the importance of water quality assessments. Common water quality monitoring parameters typically include turbidity, suspended and dissolved solids, conductivity, alkalinity, temperature, dissolved organic carbon, nitrogen and phosphorus, dissolved oxygen, pH, and microbiological water quality parameters such as chlorophyll-a and Escherichia coli. Machine learning methods (e.g., decision trees, random forests, support vector machines, k-nearest neighbors, and gradient boosting) or deep learning algorithms (e.g., multi-layer perceptron, convolutional neural networks, recurrent neural networks, and transformer models) have been used to create predictive models from water quality data. All these methods result in either statistical or physical scientific quantitative values. However, another valuable source of data remains barely explored in the field of water quality. Local knowledge from populations near water bodies or who are frequent users of rivers or lakes for recreation is a source of data that, in combination with scientific knowledge, could increase the accuracy of prediction or monitoring models. Local knowledge offers valuable insights into historical environmental changes over decades or generations.

This study focuses on developing a causal top-down Bayesian network to integrate scientific and non-scientific datasets of microbial surface water quality conditions and use the output for risk assessment. To this end, meteorological and water quality data, as well as data driven from municipal infrastructures against combined sewer overflows (CSOs), will be used as scientific indicators, while survey results from local people familiar with the river will be used as non-scientific data to conduct a microbial water quality risk assessment in the Greater Montreal Area. The study aims to investigate the capability of non-scientific data to assess microbial water quality risks in surface waters and assess its value compared to scientific data. To the best of our knowledge, this study is the first research focusing on the integration of local knowledge into microbial surface water risk assessment.