Oral Presentation Australasian Groundwater Conference 2017

Peering into groundwater models - a transparency wish-list (#213)

Pete Dupen 1
  1. WaterNSW, Penrith, NSW, Australia

The modern numerical model has become a paradox; the omnipotent hydrogeological future-teller for large in-ground developments such as mining is also the single biggest impediment to those wishing to know to what degree impact predictions can be trusted.

The most important shortcomings in the current system of reporting model outcomes to stakeholders are:

  • Specific questions that stakeholders want answered are rarely sought or addressed.
  • Key conceptualisations, assumptions and the detailed rationale for assigning parameters are rarely revealed.
  • The real uncertainties associated with the predictions are rarely assessed by the modellers, far less transparently presented to the stakeholders.

Even where the models provide good approximations of groundwater behaviour in historical and predicted future states, these shortcomings lead to a highly unsatisfactory lack of confidence on the part of agencies (and the community behind them) and an alternately baffled and strident call to “trust me, I’ve got a model” on the company and consultant’s part. The following important steps are suggested to move us on from this situation, to justify the miners’ development applications and to provide confidence to the community:

  1. Start the assessment by defining, in consultation with key stakeholders, prediction failure and clarify how a Type II error will be avoided to an acceptable (e.g. 95%) degree of certainty.
  2. Make modelling only as complex as is necessary, iteratively and only proceeding to greater complexity if it reduces uncertainty. In some cases this will mean using multiple simpler, possibly uncalibrated sub-models or other analysis to check or inform critical components of the larger models.
  3. Present key layers and parameters to reveal model conceptualisation for all models (list of suggested layers to be presented).
  4. Present uncertainties objectively and transparently, including conceptualisation and assumptions.
  5. Use the model(s) to inform future investigations and monitoring, identifying data with maximum “worth” that will plug gaps most effectively.
  • We are offering awards for Career and Early Career presentations and posters. Please indicate length of time since highest degree completed.: 5 Years or more
  • We are offering awards for Career and Early Career presentations and posters. Please indicate length of time since highest degree completed.: 5 Years or more