Data gaps and opportunities to reduce predictive uncertainty

In previous Clarence-Moreton Bioregional Assessment companion products 1.2 (Raiber et al., 2014), 1.3 (Murray et al., 2015), 1.5 (McJannet et al., 2015), 2.1-2.2 (Raiber et al., 2016a) and 2.3 (Raiber et al., 2016b), data and knowledge gaps were highlighted. As discussed in the qualitative uncertainty analysis in Section, not all of these data gaps have the same effect on predictions.

The overarching issue that arises through that discussion is the considerable uncertainty in the understanding and conceptualisation of the deeper geological layers.

Seismic data would be of great value to reduce the uncertainty in geometry of the geological model and to identify large-scale fault features. While considerable effort has been invested in the stratigraphic interpretation of existing lithological bore logs to arrive at a unified aquifer assignment, a formal hydrostratigraphic interpretation of new bores would greatly aid in reducing uncertainty in aquifer assignment.

Any groundwater observations that contain information of the flow directions and hydraulic gradients would make the current conceptualisation more robust which would increase the confidence in the model results. Obvious candidate observations are groundwater level measurements of nested or multi-level monitoring bores and environmental tracers.

The knowledge base of hydraulic properties in the deeper layers is very limited in the Clarence-Moreton bioregion. The sensitivity analysis of model predictions highlighted that especially the vertical hydraulic conductivity and storage parameters have a high influence on the model predictions. Any information that can constrain the prior distributions of these parameters will increase the predictability of the groundwater model.

While estimates of recharge and discharge are essential for groundwater management in the bioregion, they are of lesser importance when assessing the change caused by coal seam gas (CSG) development. Additional information would undoubtedly make for a better conceptual model and a groundwater model that can reproduce historical observations more accurately, but would have a very limited potential to reduce the predictive uncertainty of drawdowns and fluxes.

Related to this is the quality of the current groundwater level observations. Analysis of the metadata of the observations highlighted that the horizontal and vertical accuracy of many observation locations is insufficient to be used in formal model evaluation. While this can be addressed by additional quality control in the database in combination with field verification of observation locations, the reduction in predictive uncertainty is limited as the groundwater level observations in the alluvial aquifers cannot constrain the parameters relevant to drawdown predictions.

Last updated:
24 October 2018
Thumbnail images of the Clarence-Moreton bioregion

Product Finalisation date