Uncertainty in estimation of Soil Hydraulic properties and root zone state variables in inverse method

Soil hydraulic parameters such as residual water content, saturated water content, hydraulic conductivity, are key factors to be considered when assessing the soil capabilities to provide ecosystem services. Proper computation of fluxes from vadose zone using the hydrological models strongly depends on correct estimation of input parameters, process scale, boundary, and initial conditions. Estimation of soil parameters for many hydrological models is always an arduous task due to uncertainty bounded with parameters. Over the last few years many researchers have favoured to estimate the parameters using inversion approach due to increasing computing capabilities and easily measurable output variables. The current study deepens the understanding of the soil hydraulic parameter estimation using inversion approach. The inversion was conducted on synthetic data set using the SWAP (Soil water atmosphere and plant) model along with the GLUE (Generalized Likelihood Uncertainty Estimation) algorithm. Several constrain variables, able to be derived from remote sensing or in-situ measurements (Leaf Area index - LAI, Evapotranspiration – ET and Surface soil moisture – SSM), were used in the inversion process alone or in different combinations. The current study uses the two types of soil profile, homogenous soil system and two layered soil system. In this synthetic experiment, we compared the effect of different soil type, different surface conditions, different water conditions, and frequency of observed variables on parameter estimation. Effect of initial predefine range of the parameter space, on SHP estimation, were also investigated. Use of DSM data to define the initial range of parameter space were also investigated. We also simulated the state variables with uncertainty using the estimated parameters. Main outcomes could be reported when retrieving the SHPs, retrieval was significantly correlated with soil type and water stress condition, although overall retrieving performances were quite poor specially in layered soil system. We could identify some promising combinations of constrain variables for better estimation of parameter in different soil types. Our approach may further provide spatial sampling of DSM data components to improve the SHPs estimation, to be used as surrogate input for defining the initial range.