Data assimilation aims at quantifying and minimizing model uncertainty. In hydrological models, this uncertainty is mainly caused by the uncertain soil hydraulic parameters and their spatial variability. In this study, the impact of updating these parameters along with the model states on the estimated soil moisture is investigated. It is shown that parameter updates are beneficial and that it is advisable to resolve heterogeneous structures instead of applying a simplified soil structure.

Data assimilation methods are used throughout the geosciences to combine information from uncertain models and uncertain measurement data. In this study, we distinguish between the characteristics of geophysical systems, i.e., divergent systems (initially nearby states will drift apart) and convergent systems (initially nearby states will coalesce), and demonstrate the implications for sequential ensemble data assimilation methods, which require a sufficient divergent component.

Data assimilation methods like the ensemble Kalman filter (EnKF) can combine models and measurements to estimate states and parameters, but require a proper representation of uncertainties. In soil hydrology, model errors typically vary rapidly in space and time, which is difficult to represent. Inflation methods can account for unrepresented model errors. To improve estimations in soil hydrology, we designed a method that can adjust the inflation of states and parameters to fast varying errors.

Water movement through unsaturated soils generates streaming potential (SP). Reliability of SP for the determination of soil properties is investigated. First, influence of hydraulic and geophysical soil parameters on the SP signals is assessed using global sensitivity analysis. Then, a Bayesian approach is used to assess the identifiability of the parameters from SP data. The results of a synthetic drainage column experiment show that all parameters can be reasonably estimated from SP signals.