Soil physical models are widely used to describe water flow and solute transport in the vadose zone, for example to estimate groundwater recharge. These models require accurate and precise information about the properties of the vadose zone that link fluxes with state variables, such as the soil water retention curve, the unsaturated soil hydraulic conductivity, and the solute dispersion coefficient. Typically, these properties are determined from laboratory experiments, but with low accuracy to describe field-scale processes due to spatial variability at the field scale. The inverse model approach allows optimizing model parameters by fitting simulations to observed data at the scale of interest. The majority of field-scale inverse modeling studies have used solely information about water content to determine the properties of the vadose zone. Recently, attention has been given to combining water content and water stable isotope ratio profiles to calibrate soil models based on one single field campaign.
The main objective of this study is to determine if the inclusion of pore water isotope data allows a realistic parameterization of soil physical models without the need of continuous monitoring data. The METIS code, a soil physical model including isotope transport and isotopic fractionation due to evaporation, was used. A sensitivity analysis based on Morris Sobol method was performed, and highlighted strong interactions between the parameters, reinforcing the need of combining different observation types to calibrate models. A synthetic case allowed to determine the performances of the calibration methods proposed here. Then, using soil moisture and isotope profiles of a first campaign in a multi-objective approach to optimize the model parameters, a best set of parameters was determined for a site in South of Quebec, Canada. The realism of the parameterization obtained was judged based on the ability to reproduce soil moisture and isotope profiles of a second campaign, one year later. The multi-objective function approach with calibration and validation was able to satisfactorily reproduce the two data. However, based on the result of a synthetic case, monitoring at various time water content and water stable isotope ratio, even if if it is not continuous, would allow a more realistic parameterization of the soil parameters. The development of new in situ methods to easily measure in time the water content and pore water isotope composition is thus needed.