The first high-resolution time-dependent groundwater model of the world

Groundwater scientists and model developers from Utrecht University and Deltares joined forces to create the very first 1 km global groundwater model. This constitutes an important step toward better representation of groundwater in global water and climate models.

Although mostly hidden from sight, more than 97% of all non-frozen fresh water on earth is stored as groundwater in aquifers below ground. Groundwater plays an essential role in supporting the flow of rivers during periods of droughts and maintaining the integrity of wetland ecosystems. It also provides almost half of all the water that is applied for irrigation worldwide, especially in semi-arid regions, where groundwater extraction leads to falling water tables and aquifer depletion.

Impacts on water resources

Despite the importance of groundwater, it is not very well represented in global water and climate models that are used to assess the impacts of climate and socioeconomic change on water resources. In these models, groundwater is either not considered at all, or treated simplistically. Only recently, the first global groundwater models were made, but their resolutions were still too coarse to correctly include smaller aquifer systems and the exchange of water between groundwater and streams. The newly developed model GLOBGM is an important step forward because it aims to represent global groundwater systems with two layers at 30 arcsecond (~ 1 km) spatial resolution.

Model code

To be able to run such a very large model a computer cluster is needed, but the cluster itself does not have to be very large. Jarno Verkaik, a PhD candidate at Utrecht University and groundwater software expert at Deltares, was responsible for making this possible. Says Jarno: “I have changed the underlying model code (MODFLOW 6) in such a way that it runs efficiently in parallel on multiple processors of a computer cluster. I have applied unstructured grids to minimize the number of computational cells and associated runtime and storage. I have also applied parallel data processing which was required to handle the many terabytes of input and output data. This way we can run 60 years over night with as little computational resources as possible and makes the use of this model feasible for users that do not have access to very large computers.”

Forefront

The results of the global runs were compared with observations with satisfactory results, though still without any calibration, and also found to be better than a previous coarser (~ 10 km) version of the model. “This development brings us at the forefront of global groundwater modelling.”, says co-author Marc Bierkens, professor of Hydrology at Utrecht University. “It nicely combines the strength of Deltares in developing modelling tools with that of our group’s global modelling expertise and global data sets”.

Relevance for regional water managers

“It will not stop there”, adds Gu Oude Essink, groundwater expert at Deltares and guest associate professor at Utrecht University. “The next steps are to add more geological detail, as much local data as possible, including groundwater extractions, and to include the impact of salinity for coastal aquifers. This way, we set the stage for future global water assessments that are good enough to be of relevance for regional water managers”.