HYDROLINK

The Hydrolink project proposes a holistic and original approach to produce simulations, predictions and projections of average and extreme groundwater and surface water levels in the entire Seine basin, in the short (a few days), medium (3 to 6 months) and long (2070 and 2100) terms, with a particular interest in the estuarine exutory. These predictions established on the land-sea continuum will be derived from the use of new generation machine learning algorithms whose performances will be compared with the results obtained with the physical-based models, CaWaQS and Telemac2D, which were built in the previous works on the underground reservoirs and the bed of the Seine on the whole and the downstream of the basin respectively. Indeed, the machine learning tools are free of any explicit integration of the physical mechanism, especially those on the aquifer/river interfaces. This particularity offers us an interesting alternative to produce predictions respecting the hydrological continuity between these two compartments while covering the gradients of forcing (anthropic, geological, climatic and oceanic) between the continental upstream and the estuarine downstream. To do so, we will use and develop Machine Learning algorithms to provide joint predictions of water levels and flows at piezometric and hydraulic stations by analyzing a large database of environmental data: geological, hydrological, hydrometeorological and climatic. At this geographical scale, a rich environmental database can be built from in situ monitoring and enriched by satellite images for some crucial variables controlling the hydrological cycle in the critical area, such as soil moisture and vegetation cover, which are essential for flood forecasting.