Reduction of feasible parameter space of the inverted soil hydraulic parameters sets for Kosugi model.

Effective soil hydraulic parameters of Soil Vegetation Atmosphere Transfer (SVAT) models can be derived in a cost-efficient way by inverse modeling. Nevertheless, a serious drawback of SVAT models based on Richards??? equation is that they require as many as five unexploited correlated hydraulic parameters. To reduce the feasible parameter space, we propose a method to prevent nonphysical combinations of soil hydraulic parameter sets obtained by optimization. We adopt Kosugi???s (1994) soil hydraulic analytical model because it enables the feasible parameter space to be reduced by predicting parameter ?? from hm , which are the variance and mean of the log-transformed soil pore radius, respectively. To further decrease the parameter space we derive two models to predict saturated hydraulic conductivity, Ks, from three or four Kosugi's soil water retention parameters, respectively. These two models are based on the combination of the Hagen-Poiseuille and Darcy equations which employ three semi-empirical parameters (???1 , ???2 and ???3) calibrated on large UNSODA and HYPRES databases. Our derived models are compared to a version of the Mishra and Parker (1990) Ks model being modified to account for the parameters of Kosugi???s relationships. The results show that the uncertainties of the developed Ks model are comparable to the uncertainties of Ks measurements. Moreover, the developed Ks model outperforms the Mishra and Parker model. Therefore, the developed method will enable to substantially reduce the feasible range of the inverted Kosugi???s hydraulic parameters.