Uncertainty propagation in the modelling chain of the impact of climate change on a specific site on French subsurface drainage

Subsurface drainage is a hydraulic technique currently used on agricultural soils showing strong waterlogging issues. This technique is applied on near 10% of French arable soils. Consequently, assessing its future in a context of climate change is relevant since subsurface drainage may have a major influence on the environment, specifically on agricultural water pollution. Here, the purpose is to analyse the evolution of hydrological indicators of subsurface drainage under climate change by 2100, on the La Jaillière plot being representative of French drainage in the Great West region. We define seventeen hydrological indicators to best describe the main aspects of subsurface drainage: drained water balance, temporality of the drainage season, flood events… These indicators are simulated by modelling subsurface drained discharge on the projected future using three hydrological subsurface drainage models: DRAINMOD, MACRO and SIDRA-RU. These three models are fed by twelve climate projections CPs provided by the EURO-CORDEX project, being selected by Météo-France to ensure the representativeness of future conditions in France. We consider three climatic scenarios RCPs (Representative Concentration Pathways): the RCP 2.6 (optimistic), the RCP 4.5 (intermediate) and the RCP 8.5 (pessimistic). The propagation of uncertainties on the hydrological indicators from the modelling chain is assessed by variance decomposition using the QUALYPSO method, from the various potential sources of uncertainties (hydrological models, CPs and RCPs).  Preliminary results obtained with the SIDRA-RU model alone showed that subsurface drainage will be exposed to more extreme events such as more intense flood events over shorter periods by 2100. Depending on RCPs, the sustainability of the current network may no longer be ensured.