2.6.2.9.3 Conclusions

A thorough assessment of the existing groundwater models that cover the Richmond river basin revealed that they lacked advanced features such as transient boundaries and formal uncertainty analyses, which are crucial for fulfilling the purposes of the Clarence-Moreton Bioregional Assessment. Hence, they were deemed not to be fit-for-purpose.

The numerical MODFLOW groundwater model that was developed for this Assessment was the first regional groundwater model that covered the Richmond river basin. The conceptualisation of this model was underpinned by a three-dimensional geological model that was purpose-built for the current Assessment. The numerical groundwater model comprised six layers that represent the major types of hydrostratigraphic units prevailing in the Richmond river basin. Transient recharge and river boundaries were implemented in the model over a simulation period of 120 years spanning 1983 to 2102. Available observations were used to constrain the model parameters during the historical period from 1983 to 2012. The MODFLOW Multiple Node Well (MNW) package was adopted to simulate the 2454 non-CSG active bores within the model domain. A total number of 95 CSG extraction wells were simulated in the groundwater model using the MODFLOW Drain package, with a variable conductance that aimed at matching the predicted volumes of extracted water and achieved the target pressure heads.

Although 10,000 parameter combinations were generated for the entire parameter space of the model sequence, only 3,877 were successfully evaluated within the operational constraints of the Assessment. Recharge, drainage conductance and riverbed conductance were found to be the three most sensitive parameters for predicting pressure heads in layer 1. The drain conductance of the CSG wells, the storage coefficient and hydraulic conductivity of layer 6 controlled the amount of water that could potentially be produced from the depressurisation process. As a result, these parameters had a large impact on the drawdown predictions as they greatly influence the stress imposed on the system. Among the 3877 successful model runs, realisation 842 was deemed to be the most representative, with a drawdown prediction that compared well with the Metgasco modelling results. The predicted median change in groundwater level due to the additional coal resource development was less than 0.01 m, with a 95th percentile that did not exceed 1 m. The current model has shown that the overall potential impact of the additional coal resource development can be deemed to be very minor with a very low probability of exceeding trigger thresholds specified in the NSW Aquifer Interference Policy.

Predictions of drawdown can inform the assessment of direct impacts on groundwater-dependent assets, such as groundwater-dependent ecosystems (ecological asset), or groundwater bores used for stock, irrigation and domestic purposes (economic asset). The outputs of the numerical groundwater model can produce contours of the probability for exceeding certain drawdown thresholds. On the other hand, predictions of the exchange fluxes between the surface water and groundwater systems can inform the assessment of indirect impacts on surface water-dependent assets. This flux, which directly affects the baseflow component in surface water features, is input into the Australian Water Resources Assessment (AWRA) landscape model (AWRA-L) to quantify changes in relevant hydrological response variables at model nodes associated with surface-water-dependent assets.

The outcomes of the modelling exercise described in this product enables one to assess the level of risk associated with each asset that is potentially impacted either directly or indirectly by the CSG development. These could take the following forms: direct impact on an economic asset is: ‘there is a X% chance that Y% of domestic bores in a certain area will have a drawdown of Z m or more’; an indirect impact on a surface-water-dependent asset is: ‘there is a X% chance that the number of no-flow days will increase from Y days/year to Z days/year’.