We sampled the soil horizons of 40 important Belgian soil series, with textures ranging from sand to heavy clay, and measured their moisture retention characteristic (MRC). Of these horizons, the particle-size distribution in nine fractions, the dry bulk density, and the carbon content were determined as well. The Van Genuchten equation (1980), reduced to four parameters on the basis of a model comparison technique, was fitted through the measured MRCs using Marquardt's algorithm (1967) for nonlinear parameter estimation. We performed a one-dimensional sensitivity analysis on the optimized parameters of different MRCs to check their relative importance and stability. Principal factor analysis was used to reveal the structure in the data and to examine the relation between MRC parameters and measured soil properties. Regression equations were established between these MRC parameters and measured soil properties. To test the need for detailed information regarding the textural composition of the soil, two information levels were discussed. The first level contains the textural composition in three fractions (sand, silt, and clay) while the second level takes into account the textural composition in nine fractions and two distributional parameters (GMPS, GSD).
We concluded that the MRC can be estimated at a reasonable level of accuracy from such simple soil properties as particle-size distribution, dry bulk density, and carbon content. Detailed information on the particle-size distribution results in a better estimation of the parameters describing the shape of the MRC (a, n). The study further shows that for the prospected horizons the Van Genuchten model gives a good description over the entire range of the MRC. In addition it has been shown that the approach is flexible enough to model the MRC for a wide range of textures.