Unsaturated zone flow and solute transport modelling with MIKE SHE: model test and parameter sensitivity analysis using lysimeter data

Abstract

This study presents the test of the widely used model MIKE SHE at the lysimeter in Wagna in Southern Styria (Austria). At the lysimeter comprehensive data of vegetation, i.e., cultivation types, and soil hydraulic properties are available to parameterise the model. The aim of the study is to test the water movement and solute transport simulation in the unsaturated zone and the evapotranspiration module by a comparison with measured recharge rates at the lysimeter bottom and measured soil water contents and soil tensions in different depths. For the unsaturated zone calculation, the Richards equation and the Van Genuchten–Mualem approach is used. With the parameterised model the solute transport module is tested against data of a tracer experiment (Bromide). The water content simulation in the different soil depths yielded good results. However, particularly in the gravel zone (below 70 cm) the strong recharge dynamics and at the same time also strong water content dynamics can hardly be reproduced by one individual soil physical parameter set. Two different sets were identified and the corresponding soil water retention functions are located at the opposite ends of the relatively large range obtained by the measurements in this zone. Also, the saturated hydraulic conductivity had to be increased by an order of magnitude to accurately represent the recharge dynamics. The test of the solute transport simulation showed generally very good results. However, the early rapid rise of the measured concentrations could not be reproduced with either parameter set. The discrepancies found in the study indicate a complex pore system in the gravel zone consisting of different flow paths (e.g., system of matrix flow and macro-pore flow) activated at different soil moisture conditions, which cannot be fully represented by the used approach. However, the parameter study herein highlights the main parameters to be altered in a calibration procedure at point scale. The paper includes also a step from point to regional scale. For comparison test calculations with data, which are usually available on regional scale, showed good results for the 5-year average balance (−7 % compared to measurements) but rather large deviations from the measurements at shorter time scales.