Improved prediction of water retention curves for fine texture soils using an intergranular mixing particle size distribution model

Abstract

Laboratory measurements to derive the soil water retention curve, , are time consuming and expensive. We present a cost-effective alternative using particle size distribution (PSD) and saturated water content. We propose a novel physical conceptual intergranular mixing PSD model (IMP model) which derives from PSD, exploiting the relation between particle size and pore size distributions and the intergranular arrangement of the soil particles. The IMP model successfully predicts for fine texture soil, which is the most challenging soil texture to be modelled. With our novel model, reliable can be obtained using only three general fitting parameters without needing to assume any particular type of soil particle packing, with mean Nash–Sutcliffe efficiency coefficient of 0.92 for 259 soils. The IMP model can accurately predict for fine texture soils because: a) it implements an intergranular mixing function that accounts for soil pores not all being perfectly spherical and takes into consideration the intergranular rearrangement (mixing) of the particles, which allows neighbouring particles to have different sizes resulting in variations in pore radius and pore shape of the corresponding pore fraction; b) it overcomes the absence of PSD data for sizes smaller than the clay fraction by developing a normalised form of the Young–Laplace capillary equation; and c) the residual pore volume accounting for water strongly bound to solid particles or in very small pores is incorporated as a function of the clay fraction.