Drainage models to predict soil water regimes in drained soils: a UK perspective

Drainage is an intervention in the natural hydrology of the soil to alter the duration of adverse (waterlogged) soil conditions. The effects of drainage can be investigated by models that predict the position of the water table at a site in the presence of drainage. An inter-related series of models, which include the van Schilfgaarde non-steady state model, that have been used in the UK for the evaluation of drainage design options, are described. A simplified form of the van Schilfgaarde equation is presented, equivalent to a standard time series model, allowing both the efficient implementation of the model, and the inverse use of the model to derive performance parameters from observational data using statistical methods. A sensitivity analysis is used to investigate the relative importance of the two soil parameters, drainable porosity and soil hydraulic conductivity, on the performance of the model. This shows a far greater effect due to the variation of hydraulic conductivity.The use of a similar model to predict water tables in non-homogeneous soils has also been explored, including the investigation of a two-phase model to describe water movement in soils which are dominated by macropores. More useful, however, is the prediction of water table fluctuations in soils in which the soil hydraulic conductivity is a continuous function of soil depth, using the drainage theory of Youngs (1965). Solutions are presented for the logarithm of the hydraulic conductivity varying linearly with depth. The improvement in model performance is however gained at the expense of an additional parameter that describes the variation of hydraulic conductivity with depth. Some methods for deriving this parameter are discussed. Results from the use of this model are compared with those derived from the simple uniform conductivity model, showing superior performance.Lastly, the issue of soil lateral heterogeneity and the replicability of measurements is discussed. A detailed study of the variation of water table levels from a replicated drainage experiment indicates that in a practical situation very real limits exist on the accurate measurement of water tables, and that these present limits on our ability to verify models.