Soil Water Retention: Uni-Modal Models of Pore-Size Distribution Neglect Impacts of Soil Management

Most models describing soil water retention imply a uni-modal pore-size distribution (PSD). The uni-modal model presented by van Genuchten (termed vanG) is widely used although double-exponential models (termed Dex) implying a bi-modal PSD may better reflect reality. We tested the ability of vanG and Dex models to represent water retention in sandy top- and subsoils with different texture, in soil with contrasting management (Highfield), and in soil exposed to different tillage (Flakkebjerg). Soils were subjected to matric potentials from –10 hPa to –1.5 MPa. For all soils, the bi-modal Dex model showed a better fit to water retention data than the uni-modal vanG model. Neither of the models worked well for highly sorted soils. The vanG model gave a poorer fit for topsoils than for subsoils because of a more pronounced bi-modality of the PSD in topsoils caused by larger soil organic carbon (SOC) content and tillage. For Highfield soils, the root mean squared error (RMSE) of the vanG fit increased from long-term bare fallow (low C content, intensive tillage) to permanent grass (high C content, no tillage) reflecting a more distinct bi-modality of the PSD for well-structured soils. We conclude that uni-modal models should be used with great caution when describing effects of texture and management on PSD and that bi-modal models may provide a better fit to PSD.