Karst-aquifer operational turbidity forecasting by neural networks and the role of complexity in designing the model: a case study of the Yport basin in Normandy (France)

• Published in: Hydrogeology journal 2021-02, Vol.29 (1), p.281-295
• Main Authors: Savary, Michaël, Johannet, Anne, Massei, Nicolas, Dupont, Jean-Paul, Hauchard, Emmanuel
• Format: Article
• Language: English
• Subjects: Aquatic Pollution; Aquifers; Chalk; Complexity; Earth and Environmental Science; Earth Sciences; Environmental Sciences; Gauges; Geology; Geophysics/Geodesy; Hydrogeology; Hydrology/Water Resources; Karst; Mathematical models; Multilayer perceptrons; Neural networks; Parameters; Pollution dispersion; Pollution transport; Quality assessment; Quality control; Rain; Rain gauges; Rainfall; Turbidity; Variance analysis; Waste Water Technology; Water Management; Water pollution; Water Pollution Control; Water purification; Water Quality/Water Pollution; Water supply
• ISSN: 1431-2174; 1435-0157
• DOI: 10.1007/s10040-020-02277-w

Karst aquifers are highly susceptible to pollution transport, particularly turbidity, because these aquifers do not filter water to any significant degree. This may occasionally induce sanitary issues for the population and, therefore, it is operationally important to predict the mobility of water with high turbidity. In this study, deep specific architectures, recurrent and feed-forward, inspired from a multilayer perceptron, are used to predict the turbidity peak and the 100 NTU threshold exceedance at the Yport pumping well in Normandy (France), used by the Le Havre Seine Métropole urban conglomeration to supply water to 240,000 inhabitants. This abstraction well pumps water from the chalk karst aquifer. For this purpose, 32 architectures were designed, with different sampling rates and numbers of used rain gauges, which, using a cross-test procedure, provided 288 different models. This arrangement represented directly the rainfall–turbidity relationship. The study also assessed model quality using original criteria defined specifically for this study. Rigorous model selection makes it possible to design models that can, with limited uncertainty, predict the threshold exceedance and the maximum turbidity peak up to 24 h in advance. More interestingly, the main outcome associated with this methodology is that the complexity of the models (i.e. the number of free parameters) behaved like a high-level parameter, controlling the quality of the results independently from field considerations. This result is consistent with the well-known bias-variance tradeoff.