Estimating the water retention shape parameter from sand and clay content

This study developed an alternative way of estimating the van Genuchten water retention shape parameter n from a soil's sand and clay content. This estimation can be used to complement an infiltration experiment called the Beetkan method, which has been proposed for estimating the E van Genuchten water retention function and Brooks-Corey hydraulic conductivity characteristic. To estimate the water retention shape parameter, the Beerkan method requires a distribution function fitted to particle-size distribution data (more than five fractions) and a measurement of bulk E density. Using three published databases, we were able to derive a neural network model that predicts the shape parameter and its uncertainty from sand and clay content. Its accuracy ranges from 0.2 to 0.4. This method is comparable to prediction using parameterized particle-size distribution data. The response surface of n as a function of sand and clay content shows an increasing value of n with increasing sand content in a nonlinear way. We also show that using simpler methods for predicting shape parameter n does not influence the accuracy of the Beerkan method in estimating the soil hydraulic properties.