Building large-scale 3D coastal groundwater models with iMOD-WQ and global datasets

Large-scale coastal groundwater models (LCGMs) covering areas of several tens of thousands of km2 can provide valuable insights into (supra)regional coastal groundwater system dynamics over time. This is crucial in understanding the current and future state of transboundary groundwater resources as well as identifying potential hotspots for fresh groundwater shortages in coastal regions worldwide. Recent developments in code parallelization, namely the SEAWAT based iMOD-WQ code (Verkaik et al. 2021) and high performance computing open new possibilities in building LCGMs using open source tools like Python and available global datasets as input into the LCGMs. These LCGMs, despite simplifications and uncertainties related to hydrogeological data availability, yet provide first order approximations of groundwater conditions in data scarce large-scale regions.. In this research, we describe current and future developments of our tool and demonstrate potential opportunities in using these LCGMs as basis for further developments in (supra)regional water management in data-scarce regions. The recent development of parallel open-source iMOD-WQ signalled an important breakthrough in variable-density groundwater flow and salt transport modelling with regular (and irregular) grids options and geological data ensemble modules. Whereas in before only serial simulations using the regular SEAWAT code were possible, now these simulations can be split into multiple (practically at least tens of) partitions and executed in parallel, leading to significant reduction in computation time. This opens possibilities in terms of both the physical and temporal size of the LCGMs. The LCGMs developed in this research cover several tens of thousands km2 and simulate groundwater salinity dynamics over a full glacial-interglacial cycle (viz. approx. 125 kA). The HydroBASINS global-watershed-boundaries dataset is used to delineate the boundary of the LCGM in the inland part of the model domain. Our LCGMs also cover the offshore continental shelves; those are manually outlined and added to the selected HydroBASINS. The top elevation is derived from a global DEM dataset (GEBCO) while the bottom elevation is estimated by the bottom of the unconsolidated sediment formations as well as the sedimentary rock formations (limited to siliclastic lithology). Using this approach can lead to considerable uncertainties and therefore, whenever local hydrogeological input data is available (e.g. bore logs, groundwater salinity, extractions), we use tools like GEMPY to improve the hydrogeological model in our LCGM building tool. We believe that building a LCGM using global datasets is a necessary first step to provides valuable information for (supra)regional coastal groundwater management in data-scarce regions. Verkaik, J., J. Van Engelen, S. Huizer, M.F.P. Bierkens, H.X. Lin, and G.H.P. Oude Essink. 2021. Distributed Memory Parallel Computing of Three-Dimensional Variable-Density Groundwater Flow and Salt Transport. Advances in Water Resources 154 (August): 103976. https://doi.org/10.1016/j.advwatres.2021.103976.