Operational Turbidity Forecast Using Both Recurrent and Feed-Forward Based Multilayer Perceptrons
Approximately 25% of the world population drinking water depends on karst aquifers. Nevertheless, due to their poor filtration properties, karst aquifers are very sensitive to pollutant transport and specifically to turbidity. As physical processes involved in solid transport (advection, diffusion, deposit…) are complicated and badly known in underground conditions, a black-box modelling approach using neural networks is promising. Despite the well-known ability of universal approximation of multilayer perceptron, it appears difficult to efficiently take into account hydrological conditions of the basin. Indeed these conditions depend both on the initial state of the basin (schematically wet or dry), and on the intensity of rainfalls. To this end, an original architecture has been proposed in previous works to take into account phenomenon at large temporal scale (moisture state), coupled with small temporal scale variations (rainfall). This architecture, called hereafter as “two-branches” multilayer perceptron is compared with the classical two layers perceptron for both kinds of modelling: recurrent and non-recurrent. Applied in this way to the Yport pumping well (Normandie, France) with 12 h lag time, it appears that both models proved crucial information: amplitude and synchronization are better with “two-branches” feed forward model when thresholds surpassing prediction is better using classical feed forward perceptron.
KeywordsNeural networks Recurrent Feed-forward Turbidity Karst
The authors would like to thank the CODAH for providing rainfall and turbidity data. The Normandie Region and Seine-Normandie Water Agency are thanked for the co-funding of the study. We are also very grateful to S. Lemarie and J. Ratiarson of for the very helpful discussions they helped organize. Our thanks are extended to D. Bertin for his extremely fruitful collaboration in the design and implementation of the Neural Network simulation tool: RnfPro.
- 3.Massei, N., Dupont, J.P., Mahler, B.J., Laignel, B., Fournier, M., Valdes, D., Ogier, S.: Investigating transport properties and turbidity dynamics of a karst aquifer using correlation, spectral, and wavelet analyses. J. Hydrol. 329, 1–2, 244–25 (2006)Google Scholar
- 9.Dreyfus, G.: Neural Networks: Methodology and Applications, p. 497. Springer Science & Business Media (2005)Google Scholar
- 11.Johannet, A., Vayssade, B., Bertin, D.: Neural networks: from black box towards transparent box. Application to evapotranspiration modeling. Int. J. Comput. Intell. 4(3), 163–170 (2008)Google Scholar
- 14.Stone, M.: Cross-validatory choice and assessment of statistical predictions. J. R. Stat. Soc. Ser. B (Methodological) 111–147 (1974)Google Scholar
- 17.Darras, T., Johannet, A., Vayssade, B., Kong-A-Siou, L., Pistre, S.: In: Garcia, G.R. (eds.) Influence of the Initialization of Multilayer Perceptron for Flash Floods Forecasting: How Designing a Robust Model, (ITISE 2014), pp. 687–698. Ruiz, IR (2014)Google Scholar
- 19.Moussa, R.: When monstrosity can be beautiful while normality can be ugly: assessing the performance of event-based flood models. Hydrol. Sci. J. 55(6) (2010). Special Issue: the court of miracles of hydrology, pp. 1074–1084Google Scholar