A mine can have years of groundwater data, a network of monitoring boreholes and water-quality records, yet still face the basic question of whether that network could detect a developing impact early enough for someone to act.
Monitoring gives us a record of what has happened in the groundwater system. It describes changes in water levels, chemistry, flows, and water balances. What it cannot always tell us on its own is why those changes are occurring, where an impact may move next, or whether the boreholes are positioned to detect it.
This is where numerical modelling can add another layer of understanding. Used alongside monitoring, it can help us test groundwater-flow paths and potential contaminant movement, identify weaknesses in the monitoring network, and focus attention where an intervention could make a meaningful difference.
Monitoring tells us what has already happened
Groundwater monitoring remains the empirical foundation of any assessment. Without reliable field data, there is nothing credible against which a model can be calibrated or tested.
In a recent case study at a mining operation in Mpumalanga, monitoring information was combined with a calibrated numerical model to understand the groundwater system and assess contamination risk. The work drew on groundwater levels and chemistry as well as water balances and flow information.
The quality of that record is crucial. If coverage is sparse in part of the site, if a borehole is screened in the wrong hydrostratigraphic unit, or if the available record is too short to establish a trend, we can only come to limited conclusions.
Numerical modelling does not solve weak data. What it can do is help expose those weaknesses and show where better information would improve the assessment.
The model tests whether we are looking in the right place
A monitoring borehole only tells us about the groundwater it actually intercepts. That sounds obvious, but mining environments can be hydrogeologically complex. Groundwater may move through weathered material and deeper fractured rock, while structures such as faults or lower-permeability features can influence the direction and extent of flow.
This means the location of a borehole on a map is only part of the question. Its depth and the unit it monitors are just as important. So is its position relative to the groundwater pathway between a potential source and a receptor.
A numerical model allows us to test that conceptual understanding against observed groundwater conditions. It can help us evaluate where groundwater is likely to move and whether the existing network is positioned to detect an impact along that pathway.
The model does not replace the field evidence. It gives us a disciplined way to interrogate what that evidence is telling us.
More boreholes do not necessarily mean better monitoring
When confidence in a monitoring network is low, the instinctive response can be to add more boreholes. While that might be necessary sometimes, the real problem may be poor positioning, unsuitable screening depths, gaps in the monitoring record, or a network that was designed around a source without enough consideration of the pathway groundwater is likely to follow.
Optimising a network should therefore be about improving detection confidence rather than simply increasing the number of monitoring points.
In our case study, modelling was used to compare observed areas of water-quality concern with simulated contaminant movement. Where the two aligned, confidence that the network was detecting the behaviour of interest was greater. Where they did not, it pointed to something requiring further attention in either the monitoring network or the model itself.
Scenario testing can also help assess potential mitigation measures and show where additional monitoring may be needed to verify whether an intervention is performing as expected.
The feedback loop is where the value lies
The strongest groundwater-management approach is not a monitoring programme on one side and a numerical model on the other. Each should continuously test and improve the other.
New monitoring data should feed back into the conceptual and numerical model. The updated model can be used to test potential source behaviour and mitigation scenarios, identify uncertainty and examine whether the monitoring network remains fit for purpose. Those findings can then inform changes to borehole location, depth, monitoring frequency or the parameters being analysed.
This becomes increasingly important as a mine changes through its operational life and moves towards closure. The groundwater system being monitored is not static, and the monitoring strategy should not be treated as static either.
There is also an important distinction between scenario testing and prediction. A model cannot tell us with certainty exactly what will happen decades from now. Its value lies in showing how the groundwater system could respond under defined assumptions, where the greatest uncertainties sit and where monitoring must test whether those assumptions remain valid.
Good groundwater management therefore depends on keeping that feedback loop alive. Monitoring shows what the system is doing. Modelling helps us understand what those observations could mean and where to look next.
When the two are used together, groundwater monitoring becomes something far more useful than a historical record. It becomes an early-warning tool that can help a mine identify emerging risk while there is still time to respond.
Erin Haricombe presented her paper Integrating Groundwater Monitoring and Numerical Modelling for Early Detection and Monitoring Network Optimisation in South African Mining Environments at the IAH-SA Symposium 2026.
For more information please visit: www.wsp.com
