Abstract: Soil-water characteristic curves (SWCCs) can be used to predict the properties of unsaturated soils, including unsaturated permeability coefficient, shear strength and thermal property. However, the measurement of SWCCs is time-consuming and expensive. To solve this problem, considerable research efforts have been made to developing the procedures to directly estimate the SWCCs from basic geotechnical properties. Two types of MK models (fitted equations and prediction equations) are used for the SWCCs models, and the cftool in the scientific programming language Matlab is applied to fit the MK models, based on which four types of soils are analyzed, i.e. Xining loess, silty sand, red silty clay, moraine soil. The effects and differences of fitting equations and prediction equations are compared with regard to their capability of describing SWCCs of fine-grained soils. A calculation formulation is developed for parametric analysis based on MK model. It is shown that both the fitted equations and predicted equations of MK model perform well in describing SWCCs of the 4 types of fine-grained soils, though the former is better than the latter; it is also shown that soil texture and clay content influence . The sensitivity analysis shows that the degree of saturation is more sensitive to parameter a than to parameter m.

Key words: soil-water characteristic curves, MK model, fitting, sensitivity analysis