Validation of DIVDRA for upscaling solute transport
Summary
To make an adequate estimation for travel time distribution for the solute transport occurring within the dimensions of the NHI-WQ grid cells, upscaling of the flow and transport processes at this scale is needed. Implementing the DIVDRA (DIVide DRAinage) upscaling concept, developed by Alterra, into MODFLOW could be a solution to solve this problem. This comes down to adjusting the way drainage features in MODFLOW are schematized and parameterized. When done correctly, this would give a proper travel time distribution for large cell sizes as in the NHI model.
This research will focus on the validation of DIVDRA that could be used for implementing DIVDRA (quasi-2D) into MODFLOW to represent the sub grid drainage fluxes. Before DIVDRA can be implemented the behaviour of the analytical concept has to be validated by comparing it with numerical alternatives. DIVDRA will be compared with MODFLOW and MT3DMS reference models to validate the concept. It is also interesting to compare the DIVDRA concept with other upscaling methods, to review the benefits and added value of the DIVDRA approach. The need for validation of DIVDRA is confirmed in Bakker and Schaars review report (Bakker & Schaars, 2015) Their report describes the findings and recommendations for validating the use of the DIVDRA concept within NHI for salt transport modelling. The recommendations on how to perform such a validation are divided into four steps: The first step is analysing the current analytical formula with relevant input of LHM, time depending water balances and time depending TRANSOL/DIVDRA simulations. For the second step a 2D or 3D reference model should be built with a high resolution of 1 meter in horizontal and vertical, with which groundwater flow and solute transport could be calculated with MODFLOW and MT3DMS. As a third step the simulations should be executed with fixed fluxes for the different drainage systems. And the final verification should be done with a simulation of fixed stage for each drainage system in MODFLOW.
Essential problem for upscaling of the NHI-WQ resolution to 250 by 250-meter cell size is that the horizontal discretization changes. Therefore, the travel time distribution of solute transport is changed and also not representative for solute transport. Figure 3 shows the different between a high resolution model and a scaling approach without horizontal discretization within the 250 by 250 meter cell size. This scaling approach of the 2D models gives unrealistic travel time distributions and therefore wrong solute transport representation. During this research different approaches of drainage distance would be compared with the 2D reference model. Important indicators for travel time distribution are the flow lines through the 2D reference model. An overestimation and underestimations due to upscaling, affects the solute transport and therefore the output concentrations.
To improve the scaling concept the travel time distribution has to be optimised. The travel time depends on the flow lines through the system. To optimize the flow line the occurrence of sub region under stream. This important process is underestimated in all of the scaling approaches. Therefore the decay is over or underestimated and the outcomes could not be used for regulation or policy makers. Improving the sub region under stream makes the DIVDRA scaling approach more reliable and to use as a replacement for the high resolution model.