How Complex Should a Groundwater Model Be?

by Catherine More and John Doherty

Because environmental systems are complex, models must be complex too. Otherwise, how can they simulate the impact of human management on natural systems? This applies especially to groundwater systems. After all, water flows through heterogeneous media and interacts with surface water bodies in complex ways.

This line of logic supports public expectations of groundwater modelling. “Seeing is believing” – right? If stakeholders can see a 3D moving picture of the subsurface that shows what groundwater will do when managed in a certain way, then their fears that something may go wrong will be allayed.

Unfortunately, the situation is more subtle than this.

Decision support modelling is not environmental animation. Modellers are scientists – not movie directors. Our job is to process data – mostly far fewer data than we would like. Therefore, any actual or notional pictures that emerge from our work will not be comprised of sharply focussed images of subsurface conditions; they will be hopelessly blurred by the infinite possibilities that remain after all data have been processed. Instead, our job as scientists is to answer the question “if a groundwater system is managed in a certain way, can processing of available data assure us that a bad thing will not happen; or if it does happen, can it be contained?”

Once we have our job description right, we can start thinking about appropriate model design in light of the data assimilation and uncertainty quantification tasks that modelling must perform.

Decisions on the best way to model will always be subjective. This is how it should be, for there are no prescriptions. However it should be modellers, and not bureaucrats or stakeholders, who decide how a model should look. Furthermore, specifications of appropriate model complexity should be based on clear and logical arguments on what is the best way to implement the scientific method in a particular decision-support context.

While stakeholders are not necessarily entitled to “modelling – the movie”, they are nevertheless entitled to clear and logical explanations of the concepts which underpin the design of a particular decision-support modelling process. At the same time, society at large is entitled to guidance documents that help stakeholder groups understand these arguments.