Risk Predicting of Macropore Flow using Pedotransfer Functions, Textural Maps, and Modeling

Abstract

Leaching of P to tile drains by colloid‐facilitated transport in macropores is an important pathway for P loss in structured soils. Macropore flow is the main transport route for many organic and inorganic pollutants, but macropore flow cannot currently be reliably predicted. The objectives of this study were first to develop pedotransfer functions (PTFs) predicting near‐saturated [k(−1)] and saturated (Ks) hydraulic conductivity using simple soil parameters as predictors and second to use this information and a newly developed raster‐based soil property map of Denmark to identify risk areas for macropore flow. The data set was based on hydraulic measurements on large, undisturbed soil columns from different localities in Denmark. Unsaturated [k(h)] and saturated hydraulic conductivity were measured in the laboratory; k(−1) representing the hydraulic conductivity of the soil matrix and the difference between log(Ks) and log[k(−1)] that expresses the possible degree of preferential transport through macropores were considered crucial for the flow of water through macropores. Pedotransfer functions predicting the two key parameters (log(Ks) and log[k(−1)]) based on neural networks were derived using combinations of different soil physical parameters. The neural network was able to develop reasonably accurate PTFs predicting log[k(−1)], whereas the prediction of log(Ks) was less accurate. Using the distributed data of the hydraulic properties derived from the PTFs, outputs of water flow modeling were used to construct a new map dividing Denmark into risk categories for macropore flow. This map can be combined with other tools to identify areas where there is a high risk of contaminants leaching out of the root zone.