Improved Prediction of Unsaturated Hydraulic Conductivity with the Mualem‐van Genuchten Model

• DOI: 10.2136/sssaj2000.643843x
• Publication Date: 2010-07-27T23:12:05Z
• Authors: Marcel G. Schaap; Feike J. Leij
• Keywords: Tortuosity; Richards equation; Saturation (graph theory)

In many vadose zone hydrological studies, it is imperative that the soil's unsaturated hydraulic conductivity known. Frequently, Mualem–van Genuchten model (MVG) used for this purpose because allows prediction of from water retention parameters. For and similar equations, often assumed a measured saturated ( K s ) can be as matching point o while factor S L e to account pore connectivity tortuosity (where relative saturation ). We data set 235 soil samples with test improve predictions MVG equation. The standard practice using resulted in root mean square error log( (RMSE 1.31. Optimization yielded RMSE 0.41. fitted were, on average, about one order magnitude smaller than Furthermore, was predominantly negative, casting doubt interpreted physical way. Spearman rank correlations showed both were related van parameters neural network analyses confirmed could indeed predicted corresponding 0.84, which half an better traditional model. Bulk density textural poor predictors addition improved only marginally. Bootstrap analysis uncertainty near larger at lower contents.