3.7.4 Gaps, limitations and opportunities

This impact and risk analysis allows governments, industry and the community to focus on areas that are potentially impacted when making regulatory, water management and planning decisions. Due to the conservative nature of the modelling, the greatest confidence in results is for those areas that are very unlikely to be impacted (that is, outside the zone of potential hydrological change). Where potential impacts have been identified, further work may be required to obtain better predictions of the potential magnitude of impacts to ecosystems and individual assets.

Key knowledge gaps have been identified in each of the companion products for the Hunter subregion. The next section is a summary of the key knowledge gaps where understanding the potential impacts of coal resource development can be improved through further work.

3.7.4.1 Overall

The probabilistic approach to modelling undertaken in the Assessment was specifically chosen to deal with data and knowledge gaps. The Assessment team focused on integrating data and information that was quality assured and relevant for this regional-scale analysis. However, this meant that some data and information about the Hunter subregion were not used to inform the modelling – for instance, because it was localised and ad hoc in its coverage; lacked reliable metadata to quality assure the data; not available to the Assessment team at the time of analysis; or because operational constraints prevented collating and scrutinising the data to the standards set out in the BA.

Note that the probabilistic approach to both hydrological and ecological modelling did not consider structural errors in the modelling. As a result, the calculated probabilities are not strict probability, but are rather constrained by the model structures chosen. This issue is dealt with in greater detail in Section 2.6.2.8 in companion product 2.6.2 for the Hunter subregion (Herron et al., 2018b).

The use of regional-scale models and the characterisation of uncertainty were done specifically to deal with shortcomings in data and conceptual understanding. By modelling a wide array of parameterisations to represent the possibility of highly conductive, highly connected landscapes through to low-conductivity, poorly connected landscapes, result sets were generated that effectively put an upper limit on the area of potentially significant hydrological change. Another strength of the approach is that the development of emulators of the groundwater model allowed more localised, rapid assessments of groundwater drawdown and changes in baseflow based on local information. In flagging gaps and identifying opportunities for improvement in the following sections, it is important to be aware that more and better data will not necessarily improve the model predictions from the regional-scale model, but could potentially contribute to constraining model results for more local-scale application using the model emulators.

3.7.4.2 Geology model

The geological model constructed to underpin the groundwater modelling, described in Section 2.1.2 of companion product 2.1-2.2 for the Hunter subregion (Herron et al., 2018a), did not make use of all the data that were available at the time. NSW Department of Industry was identified as having additional data and, at the time the Hunter geological model was being built, was in the process of building a geological model for the Hunter Coalfield within the geological Sydney Basin. The feasibility of incorporating their model into the geological model of the subregion was investigated, but it was agreed that in terms of timing it introduced too great a risk to delivery on the BA, and the collaboration was not progressed. The opportunity to improve the Hunter geological model remains.

Additional data that could improve the geological model include:

The wide range in groundwater model parameters explored in the groundwater modelling partly accommodates and compensates for uncertainties in the geological model. A more detailed geological model has the potential to reduce the degree of conservatism in the predicted range of hydrological change. Especially at a local scale, combining detailed geological information with the groundwater emulators has great potential to refine the predictions, as shown through the Wallarah example in companion product 2.6.2 for the Hunter subregion (Herron et al., 2018b).

Faults were not represented in the geological model and, hence, were also not represented in the groundwater model. The extent to which faults act as conduits for groundwater flow between strata over different depths in the Hunter subregion is not well understood. The range of hydraulic conductivity parameters used in the groundwater modelling varies from corresponding to a conceptualisation close to faults impermeable to flow, to corresponding to a conceptualisation in which aquitards are compromised by local flow through faults on a regional scale. Results from stochastic representations of fault distributions in the Gloucester subregion groundwater modelling found that the faults had minimal impact on changes in groundwater due to additional coal resource development in that subregion. However, the conclusions of this local study may not be transferrable to the Hunter subregion. There remains a knowledge gap in the Hunter geological model regarding the distribution of faults and their influence on inter-aquifer connectivity.

Inferences can be made about the hydrological function of faults by measuring stream salinity and chemistry with a run of river. Targeted monitoring would be required to obtain a set of measurements that are sensitive to changes in river chemistry due to discharges from a fault. This means selecting sites with the right flow conditions and unimpeded flow – for example, not influenced by dam operations or discharges under the Hunter River Salinity Trading Scheme.

Sensitivity analysis of the modelling results indicated that the greatest potential to reduce predictive uncertainty lies in improved characterisation of hydraulic properties of the sedimentary rocks, especially the porosity and storage parameters. In order to unlock the information of historical groundwater levels to constrain these properties, it is essential that uncertainty of boundary conditions in the regional model, such as the geomorphology of the river network or recharge, is reduced.

3.7.4.3 Groundwater data

Groundwater data are often only available at a very limited number of sites, and are often poorly documented, particularly for regional applications. This was an issue in the Hunter subregion, particularly for depth to watertable, recharge and contributions to baseflow from groundwater.

Groundwater data from state databases primarily include monitoring data for shallow groundwater systems and aquifers used for irrigation, stock and domestic purposes. These data are usually in the form of water level measurements and major ion analyses, which support knowledge of groundwater recharge processes and interactions between rivers and groundwater. However, it provides limited understanding of deeper groundwater systems. Local monitoring of the effect on groundwater levels by proponents of their mining activities is less relevant. Such data, rather than constraining the predictions, can bias the predictions if the historical stresses and local geological conditions of this monitoring data are not well understood and represented in the regional model. This has been factored into the Assessment’s uncertainty analysis and modelling. Future assessments would be assisted by improved information on deeper groundwater systems.

Improved mapping of depth to groundwater, and its spatial and temporal variation, not only have potential to constrain hydrological change predictions, they provide much needed context for the interpretation of the ecological impacts due to hydrological change. Interactions between changes in groundwater availability and the health and persistence of terrestrial groundwater-dependent vegetation remain uncertain due, in part, to sparse mapping of groundwater depths outside of alluvial layers.

3.7.4.4 Integrated hydrological modelling

More closely coupled surface water and groundwater models could incorporate feedback mechanisms not included in this Assessment. For example, a reduction in baseflow to a stream could induce small groundwater drawdowns in downstream regions where the stream is leaking to the groundwater system.

A choice was made to not build a detailed river management model; rather a few key river management rules were incorporated into AWRA-R. There is an opportunity for NSW Department of Primary Industries to integrate modelled runoff and baseflow changes into a river management model such as IQQM for water resource planning.

While not undertaken as part of this BA, the surface water and groundwater models can be run with different permutations of the additional coal resource developments to explore alternate development futures. An opportunity for further consideration is tailoring the models to receive the results of site-scale modelling undertaken by the mining companies to assess off-site, cumulative impacts.

In the Hunter modelled stream network, the distribution of model nodes was too sparse to enable a comprehensive extrapolation to network reaches, resulting in many ‘potentially impacted’ reaches, where hydrological changes could not be quantified with any accuracy. A higher density of model nodes, located immediately upstream of major stream confluences and upstream and downstream of mine operations, would allow the point-scale information to be extrapolated to a greater proportion of the stream network. A more extensive quantification of hydrological changes along the stream network would enable better spatialisation of the results of the receptor impact modelling.

There is an opportunity to consider the dependency between hydrological response variables, particularly between groundwater and surface water hydrological response variables, more fully in any further assessment. This may be important if additional receptor impact models are considered where that dependency is more important to address.

3.7.4.5 Water quality

Changes in water quality parameters that could occur due to reductions in surface runoff and groundwater to streamflow or due to enhanced connectivity between aquifers of differing water quality, for example, are not represented in the models. Modelling the changes in water quality was not part of the scope of the BAs. Potential changes in stream salinity were considered in Section 3.3.4, having regard to salinity hazard mapping, stream salinity data and existing management arrangements. Saline discharges from the mining and power generation sites along the Hunter Regulated River are managed through the Hunter River Salinity Trading Scheme. Salinity hazard mapping indicates extensive areas of risk associated with Permian coal measures, and changes in the relative contributions of surface runoff and groundwater could result in changes in stream salinity to smaller, unregulated streams near mines. The risk of adverse impacts will depend on local conditions, including current stream salinities and stream condition more generally.

There are opportunities to develop water quality modules to accompany the regional hydrological models, although results from this BA identify areas where more local investigation of the potential for adverse water quality changes might be warranted.

3.7.4.6 Assessing ecological impacts

The approach adopted does not consider how different ecosystems within a given landscape class might respond differently to the same hydrological change. It also does not consider ecological interactions between landscape classes. It also does not consider how response models for one functional group might underestimate complex ecosystem functions, interactions and cascade effects.

In the Hunter subregion, the ‘Forested wetland’ landscape class includes two distinct types of forested wetlands: Eastern Riverine forested wetlands, which predominate in the Hunter river basin; and Coastal Swamp forested wetlands, which predominate in the Macquarie-Tuggerah lakes basin. The receptor impact modelling was premised on the Eastern Riverine forested wetlands, which represent about half of the forested wetlands in the zone of potential hydrological change. Thus, the potential impacts of hydrological changes on the coastal swamp communities represents a knowledge gap in this Assessment.

Similarly, the qualitative model developed for the ‘Rainforest’ landscape class was premised on rainforests that occupy low-order stream and gully habitats, and that were considered to use groundwater from perched watertables and local hillslope aquifers opportunistically. The mapped distribution of rainforests (NSW Office of Water, Dataset 1) showed that within the ‘Rainforest’ landscape class, there is also a riparian rainforest associated with the alluvium of the perennial Wyong River. The qualitative model for the ‘Rainforest’ landscape class is not considered appropriate for these riparian rainforests. Potential impacts of changes in flow regime along the Wyong River is a gap in this Assessment.

Experts at the qualitative modelling workshop were uncertain about aspects of the hydrology of the ‘Freshwater wetland’ landscape class. The extent to which these features interact with a regional watertable was unknown. The dependence of these systems on overbank flows from rivers was also uncertain. Thus the risk to freshwater wetlands from the Wallarah 2 and Mandalong Southern Extension projects is a knowledge gap. The assumptions and limitations of the receptor impact modelling are described in Table 4 in companion product 2.7 for the Hunter subregion (Hosack et al., 2018a).

Additional vegetation mapping and ongoing research to identify GDEs in the subregion would improve assessment of impacts on water-dependent assets. Additionally, tracking the biophysical processes, such as rate of actual evapotranspiration and vegetation growth rates, of the GDEs and interpreting these in an ecohydrological framework would improve understanding of the interactions between changes in groundwater availability and the health of terrestrial vegetation that relies on groundwater. This can be performed by field measurement and/or use of time series remote sensing.

As actual water requirements of different plant communities are only approximately known, future assessments would be assisted by more work to identify suitable bio-indicators of ecosystem condition, or alternative methods of assessing the condition of water-dependent ecosystems. Again, this is likely best performed using field measurement and/or time series remote sensing.

3.7.4.7 Sociocultural assets

Many sociocultural assets in the Hunter subregion from the Register of the National Estate are built infrastructure, such as historic buildings or bridges. The Bioregional Assessment Programme does not have the expertise to comment on potential impacts of changes in hydrological regimes to built infrastructure.

Although the NSW Department of Primary Industries Water were commissioned to engage with Indigenous communities to collect information on Indigenous water assets for several bioregions and subregions in BAs, this did not occur for the Hunter subregion. Thus the water-dependent asset register does not include sociocultural assets that have been specifically identified by the local Indigenous communities. However, many of the assets that are in the asset register are likely to be associated with Indigenous assets. Cultural sensitivities often attach to Indigenous assets, and the Indigenous communities may prefer that details of their location and value are retained with their Elders or within their communities. Thus, it is not clear what opportunity there might be to address this gap in the water-dependent asset register.

3.7.4.8 Climate change and land use

The implications of climate change and changes in land use were not considered in the modelling. A more complete picture of the potential impacts due to additional coal resource development could be obtained by considering these changes in the context of a warming climate and changing demands for water. This Assessment identified the risk to water-dependent landscape classes and assets from additional coal resource development, but how this information is used and the decisions made could differ if for example, coal-fired power stations were closed down in the subregion, or if more land were set aside for strategic agricultural uses, or if the water demands of coastal populations changed.