Prediction of a pore distribution factor from soil textural and mechanicalparameters

Soil-water retention properties (WRC) are required for modeling purposes, b ut data availability is restricted by the high cost of measurements. Predic tion of WRC from particle size distribution (PSD) is a useful approach that could be improved by accounting for soil structure. Mechanical parameters can characterize soil structure in situ, Our objective was to use mechanica l and PSD parameters to estimate a pore distribution factor, lambda, from a power-law model of WRC, Samples for WRC and PSD determinations were taken in pre-wetted horizons after characterization of soil structure with multip le measurements of a single-vane shear test (SB) and penetration resistance (PR), Mechanical parameters were the mean, M, and standard deviation, a, o f SB and PR. Parameters characterizing a PSD were the power exponent of a c umulative exponential function, beta; the geometric mean, mu ((r)) = e(mu 1 n(r)), and standard deviation, a of a lognormal distribution; and clay cont ent. Models of lambda were built with the Group Method of Data Handling (GM DH) first using textural and mechanical parameters separately and then usin g each textural variable with all mechanical parameters. Both mu ((r)) and sigma (1n(r)) were consistently selected as the best textural estimators of lambda. The best mechanical estimators were log(M-SB) and sigma (SB). In m odels that included textural and mechanical parameters, sigma (PR) was sele cted consistently regardless of the textural parameter used. Textural param eters were better predictors than mechanical parameters, even though the la tter alone provided a reasonable estimate of lambda. Mechanical parameters improved textural estimates of lambda only when clay content was used to ch aracterize a PSD.