Background Soil-water characteristic curve and hydraulic conductivity of forest soil are important mathematical models, which play conversion roles in coupling analysis of water force coupling in porous media, but it is not deeply understood yet when the existence of plant roots was taken into account.

Methods The soil water characteristics and hydraulic conductivity based on VG model of undisturbed soil for four stands (coniferous and broad-leaved mixed forest, evergreen broad-leaved forest, Phyllostachys pubescens forest, and shrub forest) were studied by centrifuge test and variable head permeability test. Meanwhile, the normalized mathematical model of unsaturated hydraulic conductivity versus soil root was established.

Results 1) P. pubescens forest has the most root amount especially in the aspect of coarse roots; the broad-leaved mixed forest has the widest distribution range of each diameter class, but the number of roots in each layer is not much, the number of roots in each layer of shrubbery is large, and the shrubbery has considerable. 2) The saturated conductivities of four forests surface soil are 7.00 μm/s (coniferous and broad-leaved mixed forest), 7.04 μm/s (evergreen broad-leaved forest), 0.69 μm/s (P. pubescens forest) and 3.70 μm/s (shrub forest), respectively. And it is a decreasing tendency with soil depth, except for the second layer in P. pubescens forest. 3) The air entry parameter of four forest stands increased firstly and then decreased with the soil depth, while the pore diameter parameter n decreased with the soil depth. The saturated conductivity of forest soil versus the number of roots can be regressed as power law function, R²≥0.959 76.

Conclusions The saturated conductivity k_s of soil in mixed coniferous and broad-leaved forests, evergreen broad-leaved forests, P. pubescens forests and shrub forests are closely related to the root distribution, and the saturated conductivity and root number are in the form of power function. The unsaturated conductivity of forest soil can be described and predicted by normalized model when plant roots are taken into account.