Observations Analytic element model

The regional analytic element groundwater model (GW AEM) is designed to simulate the change in head due to coal resource development, not to simulate historical, pre-development groundwater flow dynamics. From companion product 2.1-2.2 for the Gloucester subregion (Frery et al., 2018) it is apparent that there is very limited historical groundwater flow or head data available that are suited to constrain the groundwater model predictions at a regional scale, outside the alluvium. Therefore, in this model the prior distributions of the parameters will not be constrained by observations of the state variables (see Section

Figure 18

Figure 18 Predicted extracted water flow profile over life of Gloucester Gas Project

Source: AGL Energy Limited (2015). This figure is not covered by a Creative Commons Attribution licence. It has been reproduced with the permission of AGL Energy Limited.

AGL Energy Limited (AGL) has provided estimates of the expected water production for its Stage 1 gas field development area (Figure 18; see also Section in companion product 2.1.-2.2 for the Gloucester subregion (Frery et al., 2018)). Figure 18 shows the 10th, 50th and 90th percentiles of predicted water extraction rates. These estimates are based on detailed reservoir modelling incorporating local information on the hydrogeology and coal seam gas (CSG) exploitation that is beyond the resolution of the regional-scale GW AEM. The water production rate increases rapidly in the early stage of CSG development and decreases slowly after this peak. The maximum predicted water extraction rate is approximately 0.32 ML/day, 0.6 ML/day and 0.9 ML/day for the 10th, 50th and 90th percentiles, respectively.

The CSG wells are implemented as a head-dependent boundary in the model and the simulated water production rates are stored for each simulation. While the water production cannot be considered as hard data to constrain the model, it provides a reality check of the model as it allows verification as to whether the simulated stress to the system is of the same order of magnitude. Based on Figure 18, any parameter combination that results in a maximum water extraction rate in excess of 1.1 ML/day is deemed to be unrealistic. This value is well above the 90th percentile predicted by AGL Energy Limited. Section details how this information is integrated in the uncertainty analysis. Alluvial MODFLOW models

Groundwater levels in the alluvium are highly influenced by local conditions, such as local heterogeneity in aquifer properties and temporal dynamics of surface water – groundwater interaction as illustrated in Figure 13. From Section in companion product 2.1-2.2 for the Gloucester subregion (Frery et al., 2018) it is apparent that the groundwater level monitoring data density, both spatially and temporally, is very low. There is a high risk that by constraining regional-scale parameters with head observations that are dominated by local factors, parameters will be biased and not representative (Doherty and Welter, 2010).

The choice is therefore made to constrain the model with estimates of the water balance, more specifically the upward influx into the alluvium from the weathered zone and the discharge to the river system. These estimates of the water balance integrate spatial heterogeneity and temporal fluctuations in boundary conditions and are therefore more representative to constrain regional-scale parameters.

The observed values are determined on a long-term basis, assuming that measured salinity values in aquifers and streams were a good indicator of water mixing ratios (see companion product 2.1-2.2 for the Gloucester subregion (Frery et al., 2018)). For the Avon model, the long-term estimate over the 1982 to 2011 period of discharge to the river system is 7 GL/year ± 2.0 GL/year. The estimated upward flux over the same period is assumed to be 0.5 GL/year ± 0.5 GL/year. For the Karuah model the estimates are 6.5 GL/year ± 2.5 GL/year and 0.9 GL/year ± 0.9 GL/year for drainage flux and upward flux, respectively (Table 7). The ranges in these fluxes reflect the natural variability and observation uncertainty in the water balance estimates.

These estimates are comparable to the water balance estimates for the entire Gloucester Basin by Parsons Brinckerhoff (2015) of 9.3 GL/year baseflow and 1.5 GL/year upward flow.

Table 7 Estimated mean and standard deviation of the long-term groundwater fluxes used to constrain the Avon and Karuah models for the Gloucester subregion


Observation data






Discharge to drain



Upward flow



Karuah-Mammy Johnson

Discharge to drain



Upward flow



Last updated:
5 November 2018
Thumbnail of the Gloucester subregion

Product Finalisation date