2.6.2.9.1 Limitations

Quantification of the hydrogeological changes due to coal resource development in the Hunter subregion was undertaken using a regional-scale groundwater model. Having a large modelling domain and the requirement for computational efficiency to perform many simulations necessitated the implementation of a relatively coarse model resolution, albeit with areas of higher resolution around mines and along streams. Because the resolution of the model mesh (at best 500 m) defines the scale of information that can be represented in the model, local-scale information, even where it might have been available, was not incorporated into the parameterisation of the model. The probabilistic approach to quantifying the hydrological changes due to coal resource development in the bioregional assessments (BAs) was adopted to account for and be transparent about this uncertainty, but also to make best use of observation data to constrain model parameters.

A probabilistic framework, however, means there is not a single parameter combination that provides a ‘best fit’ to observations and a corresponding single set of predictions. In other words, the groundwater model is not a calibrated model designed to give a deterministic result to a model future. Any evaluation or further use of the parameter combinations used in the models or the predictions needs to take into account the full posterior distributions reported in Section 2.6.2.8 . Input data, model files (including pre- and post-processing scripts and executables) and results are available at www.bioregionalassessments.gov.au.

The Hunter subregion groundwater model is a regional-scale model covering 34,000 km² at a resolution of more than 500 m. This means a decision was made to not try to capture information at the mine scale or less in developing the model, even though more detailed information may be available locally, particularly in the vicinity of coal mines. The implications of this are that local variations in geology and hydrogeology, which can significantly influence the local effects of coal mining activity, are not represented, and the regional model predictions may overestimate the likely hydrological changes. The reliability and accuracy of any predictions made by this regional model will be less than the reliability and accuracy of predictions made by a local mine-scale groundwater model that fully accounts for this level of detail, and direct comparisons between results from these local-scale models and results from this groundwater model should be viewed with caution.

The qualitative uncertainty analysis (Section 2.6.2.8.2) lists the major assumptions and model choices that form the basis of the probabilistic assessment of the impacts of coal resource development on groundwater-related receptors in the Hunter subregion. The BA groundwater modelling team (and authors of this report) deem these assumptions to be valid and acceptable for the purposes of the BA. These assumptions are not necessarily valid or acceptable for addressing other water resource management questions. Therefore, the authors recommend against using the Hunter subregion groundwater model for other purposes without assessing its suitability for the new purpose, having regard to its conceptualisation, parameterisation and implementation, and in line with the Australian groundwater modelling guidelines (Barnett et al., 2012).

Some limitations of the model relate to features or processes it does not represent. The groundwater model does not include faults or other geological structures across the modelling domain. Fracturing of rocks associated with folding and faulting creates pathways for water movement and, if well connected, can enhance hydraulic conductivities laterally and/or vertically. The model does, however, include hydraulic enhancement of the goaf above longwall mines as part of its representation of mine impacts, which conceptually includes hydraulic conductivity increases from fracturing of rock.

The model also does not include changes in recharge at the surface of subsidence zones. If fracturing of rock above longwall mines extends to the land surface, this would be expected to have some impact on recharge, particularly if the subsidence creates closed basins that intercept runoff, or leads to fracturing of a streambed and loss of streamflow, as has occurred in the Hunter subregion and Sydney Basin bioregion (NSW Scientific Committee, 2005; Krogh, 2015).

As implemented for the BA, the model assumes that the hydraulic enhancement from underground mining occurs on the first day of mining across the maximum footprint area of the mine. This assumption often causes a relatively short-lived surge (increase) in baseflow which gradually decays, but is largely a model artefact. Phasing in the area impacted by longwall mine collapse over the life of the mine has been shown to typically reduce this baseflow surge. The surface water – groundwater flux changes incorporated into the river model were modified to eliminate baseflow increases and ensure that hydrological changes to water-dependent assets is focused on the impacts of baseflow decreases.