Using OpenMI to model surface and groundwater exchange

Jan Gregersen (
Jacob Gudbjerg (

The exchange of between surface- and ground water may contribute significantly to the overall hydrological mass balance. Consequently, quantitatively determination of this exchange is essential in many situations. However, effective and accurate methods of direct measurements of such exchange are currently not available.

One way to address this challenge is to use inverse modelling. In this approach the overall hydrological system, including lakes, streams and the groundwater is modelled in one fully coupled system. The system is calibrated against observed time series of e.g. groundwater heads and river discharge, and subsequently the groundwater exchange can be extracted from the model results.

In this study we have investigated the possibility utilizing measured chloride concentration to further improve the model results. The chloride concentration in rainwater is practically zero, whereas groundwater often contains chloride. Comparing the chloride concentration in e.g. a lake with chloride concentrations in water samples from a nearby borehole will give some indications of origin of the lake water. If the chloride concentration in the lake is close to the chloride concentration in the groundwater, most likely the lake water originates from groundwater. If the chloride concentration in the lake is close to zero, there is no or very little exchange with the groundwater. However, to get a quantitative measure for this exchange, and especially its dynamic nature, inverse modelling is needed.

The challenge is to set up a fairly complex model system, which simulates the dynamic and spatial nature of groundwater flow, solute transport, surface water flow, and the exchange of water and solutes between surface water and groundwater.

In order to investigate the possibilities of setting up such system, OpenMI (Gregersen et al., 2007) was used to link three components/models.

· The time series tool from HydroInform

· The conceptual HydroNet model from HydroInfom (surface water and solute transport)

· The fully distributed and physically based ground water model Mike She from DHI.

We used a “simple” hypothetical case.  The system consisted of two lakes connected by  a steam. The upper lake had a water table that was below the groundwater table whereas the lower lake was above the groundwater table. Both lakes and the stream exchanged water with the underlying groundwater. (See the figure 1).

Figure 1 hypothetical case to investigate integrated modelling of surface and groundwater interaction.

The actual groundwater interaction was handled by HydroNet using a simply Darcy calculation based on the water level in the surface water bodies and the groundwater level obtained from Mike She through OpenMI. The leakage (positive or negative) was provided to Mike She (also through OpenMI), and thus influencing the groundwater flow and groundwater head.

In OpenMI terms the HydroNet model consisted of four ElementSets: The perimeters of the upper lake, the upper stream, the lower stream, and the lower lake. The groundwater in Mike She was defined by single ElementSet covering the full grid. Eight OpenMI links were established between the two models. One for each HydroNet water body providing leakage to Mike She and one for each HydroNet water body retrieving groundwater heads from Mike She.

Figure 2 OpenMI ElementSets used for the linking of Mike She and HydroNet

In order to ensure that Mike She used the same precipitation timeseries as HydroNet, the timeseries tool was connected to Mike She.

Figure 4 Groundwater potentials and flows for the base scenario.

As expected, the groundwater flowed towards the upper lake and away from the lower lake.

The modelled chloride concentration in the two lakes over the simulation period is shown in the figure 5 below:

Figure 5 Modelled chloride concentrations in the two lakes


The upper lake, which received ground water, had a significant higher chloride concentration as compared to the lower lake, which was draining to the ground water. (The chloride in the lower lake originated from upper lake). The two concentrations varied over time due to changing precipitation. The precipitation impacted the concentrations in two ways. Direct impact from the rain on the surface of the lakes, and indirect, due to its influences on the groundwater potentials, which again impacts the rate of groundwater exchange.

A second scenario was made to se differences due to higher precipitation rates. As seen on the figure 6 below, the groundwater potentials were increased, which made the ground water flow to both lakes. The chloride concentrations got higher and the difference between the concentrations in the two lakes were smaller.

Figure 6 Groundwater potentials and flows for scenario with increased precipitation

Discussion and conclusion

This study has taught us, that setting up such a fairly complex system is relatively easily done with OpenMI.

Naturally, things get more difficult for real cases. However, the largest part of the work in setting up the system for a real case relates to setting up the groundwater model. Consequently, we see opportunities in extending existing groundwater models with HydroNet, and thus utilize information about chloride concentrations to improve the accuracy of the model.


This study was co-financed by the Danish Agency for Spatial and Environmental Planning, which is part of the Danish Ministry of the Environment.


Gregersen J.B. , P. J. A. Gijsbers, S. J. P. Westen (2007). OpenMI – Open Modelling Interface. Journal of Hydroinformatics, 09.3 2007.