Estimating soil water retention function from its particle-size distribution

Abstract

Knowledge of the soil water retention curve (SWRC) is indispensable for characterizing and modeling water flow and solute transport in soils. However, since direct measurement of the SWRC in a laboratory is expensive, time-consuming, and laborious, the SWRC has been frequently estimated from more easily measurable soil properties. Previously, we formulated an asymmetry-based pore-solid fractal (PSF) model that addresses asymmetries in fractal dimensions between the distributions of particlesize (PSD) and pore-size (POD) to better estimate the SWRC from the PSD data of a soil. Despite this effort, however, the asymptotic problems of a single power-law function when soil water contents are close to saturation still remains unresolved. To overcome such drawbacks, we addressed functional similarities between the cumulative PSD (cPSD) and the SWRC to avoid the asymptotic problems and the concept of the slope of the SWRC at its inflection point to treat a bimodal POD, and evaluated the performance of both models using experimental PSD and SWRC data from the UNSODA database (103 soils for model calibration and 46 soils for model validation). Some limitations of the performance of the models were discussed by applying the models to various soils. The fit of the cPSD and the SWRC to the calibration dataset showed that both models performed well irrespective of soil textures and the square of the Pearson product-moment correlation coefficient was 0.987 and 0.965, respectively. The values of the inflection points of the cPSD (p_c) increased with increasing sand fraction, while those of the SWRC (h_c) decreased, indicating that coarse-textured soils had smaller particle-size and larger suction head at the inflection points than finer-textured soils. Quadratic regression relationship between the shape-related parameters (mp for the cPSD and m_h for the SWRC) showed that the magnitude of m_h changed little in the lower range of m_p but increased abruptly in the higher range of m_p, indicating that the slope of the SWRC becomes steeper as the soil texture becomes coarser. Overall, the cPSD model fitted the measured data reasonably well, indicating that the residual fraction (F_r) that accounts for the contribution of fine-size colloidal fraction to soil water retention was adequate. Validation results with various soils showed that the performance of the SWRC model was dependent on the accuracy of the estimated inflection point. Although the cPSD model performed better than the SWRC model, the fitting results indicated that the adoption of functional similarities between the cPSD and the SWRC and the concept of point of inflection was adequate.