Uncertainties around modelling of steady-state phreatic evaporation with field soil profiles of δ18O and chloride

The application of the general advection–diffusion equation to model phreatic evaporation flux rates from solute concentration soil profiles has been applied to a variety of arid zone locations but the effects of heterogeneous and uncertain field data on modelling performance are rarely described. We revisit and extend the model by accounting for the effects of variable water content and sediment type on the impedance factor, and in turn, effective diffusion coefficient. We explore the local sensitivity of the extended model to its major assumptions and parameter values. The mean diffusion coefficient of the soil profile data is most sensitive to the volumetric water content close to the evaporation front and so modelled evaporation results can be very sensitive to uncertainty around the position of the evaporation front. The uncertainty in the position of the groundwater reservoir and its concentration has little effect where the lower part of the profile shows an asymptotic trend towards a representative value of groundwater concentration but can vary substantially with different values of groundwater depth and concentration if this is not the case. Field data from the southern margin of the Great Artesian Basin are shown to exhibit considerable uncertainty around the boundary conditions of the advection–diffusion model (e.g. position of evaporation front, depth and concentration of lower groundwater boundary condition). Resulting phreatic evaporation flux estimates show a wide range that needs to be considered when using these data in water balance studies of groundwater systems.