A Comparative Study of Three Forms of the Richard Equation Used for Predicting One‐Dimensional Infiltration in Unsaturated Soil

The performance of three finite difference models, each based on a different way of expressing the Richard transient water transfer equation are compared with regard to their precision, time consumption, and economical feasibility. The models are tested on both a quasi‐analytical solution and experimental results for two different types of soil and for two upper boundary conditions (Dirichlet and Neumann). Two test criteria are chosen to cope with the individual experimental requirements. For both the maximum error criterion and the overall error criterion the model based on the Kirchhoff integral transformation leads to more accurate results than that founded on the local balance form of the Richard equation. The latter model is slightly influenced by the generation of weighting errors; on the other hand, the former model is more time consuming. The use of the model based on the decomposed form of the Richard equation is not recommended because this nonconservative model is strongly affected by weighting errors, and its range of operation is limited insofar as the choice of mesh ratio is concerned. From the results obtained, nomograms of considerable decision‐making potential in helping to select the model most appropriate to a particular set of constraints in terms of accuracy and computer time can be constructed.