Faster, higher-resolution modelling of fresh groundwater resources

Published: 29 November 2021

To make predictions more accurate, high-resolution groundwater models are needed that can simulate reality better. Worldwide, billions of people depend on fresh groundwater resources for their domestic, agricultural and industrial water consumption.

It is becoming increasingly important for water management authorities to know how the available stocks are affected by extreme drought, salinisation, sea level rise and land subsidence due to the excessive extraction of groundwater, and how to compensate for negative effects. They depend on groundwater models for that information.

Transport over a century calculated within a few days

Forecasting models of this kind are made, for example, for Dutch water management authorities with the Netherlands Hydrological Instrument (NHI), which provide users with software and data. These models generally make calculations covering longer periods of time and so they need long computation times and use a lot of memory. These limitations have a major effect on practicality when the models are calibrated and numerous scenario calculations have to be made. Recently, however, major progress has been made towards reducing these long computing times with code parallelisation.

In the first step, the current national application of NHI, the Integrated National Hydrological Model, has now been significantly accelerated by parallelising the MODFLOW code (iMODFLOW-MetaSWAP). The second step – the parallelisation of the SEAWAT code (iMOD-WQ) – made it possible to develop the national three-dimensional NHI fresh-salt model. It is now possible to calculate, within a few days, variable-density groundwater and salt transport for a period of, for example, a century. This is much faster than previously and so it is a genuine game-changer.

Jarno Verkaik, groundwater modelling expert: “As a solution for the run time and memory capacity of groundwater models, I used parallelisation, distributing memory to multiple computers. In my first two papers, I have shown that it is possible to achieve an acceleration of more than two orders of magnitude.”

Many more applications suddenly possible

Because of the significantly shorter computing times, many more applications have suddenly become possible, says Gualbert Oude Essink. “So we can now simulate local groundwater processes with a higher resolution much more accurately and, as a result, understand them better. Take, for example, rising salt groundwater below a groundwater extraction well, drought in a section of agricultural land or a small nature area, or fresh rainwater lenses floating on brackish groundwater. At the other end of the spectrum of possible applications, we see 3D fresh-salt groundwater models on a global scale. We can use them, for example, to calculate the effect of human activity and sea level rise along all the world’s coastal areas. We can also run multiple scenarios for compensation measures quickly. In that way, we can provide action perspectives for our clients, from the World Bank and the FAO to national and regional governments. Recently, the University of Oldenburg used our code to determine the effects of climate change on the groundwater system in Northwest Germany.”

Tested together with the U.S. Geological Survey

The research involves close cooperation with the U.S. Geological Survey (USGS), the producer of MODFLOW. MODFLOW is open source and it is the most widely used code in the world for groundwater. It is used by a large community of users working in government, consultancy firms and research institutes. A working parallel version of MODFLOW 6 has now been developed with the USGS to run a PCR-GLOBWB global groundwater model at a resolution of one kilometre for approximately 60 years.

The final experiments for this work are now being completed and the results will be written up in a new peer-reviewed paper. The plan is for this code to serve as the basis for a release to third parties in 2022 in collaboration with the USGS.