Comparing alternative conceptual models for tile drains and soil heterogeneity for the simulation of tile drainage in agricultural catchments

Abstract

• The seepage node concept was used to simulate tile drainage at the catchment scale. • Tile drainage was simulated without the specific location of tile drains. • Similar drainage discharge volumes were simulated with different conceptual models. • Drainage discharge was sensitive to soil heterogeneity. • Observed stream discharge and depth to water table were well simulated. Tile drains are important water flow paths in agricultural catchments and must be included in hydrological models. However, their locations are rarely known and the explicit incorporation of tile drains in hydrological models requires refined meshes around the drains, which can significantly increase computational times. Although seepage nodes have been used to represent tile drains with satisfactory performance, they have never been applied to represent all tile drainage systems in a catchment. The goal of this study is to compare different conceptual models for tile drains and soil heterogeneity for the numerical simulation of tile drainage in an agricultural catchment in Denmark. The first conceptual model for tile drains uses seepage nodes to represent only the main collector drains in the catchment and the second model uses seepage nodes distributed over all the agricultural areas, without considering the specific locations of tile drains. A third conceptual model, labelled the Benchmark Model, represent all tile drains at their known locations with seepage nodes and a fourth conceptual model implicitly represents tile drains as a high-permeability layer. The four models performed satisfactorily to simulate the observed outlet stream discharge and could be recommended almost interchangeably. The simulation of the water table depth was very satisfactory compared to modeling studies with similar mesh resolution (∼50 m). Results indicated that the three models using seepage nodes i) simulated similar monthly discharges and cumulative discharge volumes for most of the studied tile-drained areas, and ii) simulated surface water flow in tile-drained fields without runoff or ponding water. The shorter simulation times (around 35% faster) of the model representing the main drains and the distributed seepage node model suggest that they are suitable for model calibration, compared to the Benchmark Model. Whenever the location of tile drains is unavailable, using seepage node to represent drains in agricultural areas may satisfactorily simulate catchment-scale stream and drainage discharges. Four alternative soil models were developed to evaluate the effect of soil heterogeneity on the simulations. Our results suggest that at smaller scales (drainage area) soil heterogeneity is more relevant than the drainage conceptualization to improve model results. However, at the subcatchment scale, the opposite was observed and, at the catchment scale, both criteria had a comparable effect.