2.6.2.7 Observations and predictions

Summary

Historical groundwater level and streamflow data are used to constrain groundwater model predictions. Through a rigorous quality control process, 134 groundwater monitoring sites are selected to constrain the numerical modelling. In the absence of reliable, regional estimates of surface water – groundwater flux, observed streamflow data are used to impose upper bounds on the modelled surface water – groundwater flux. The water production by coal resource developments, estimated by the proponents, is compared to the simulated water production rates to further constrain the model.

At each model node in the model domain, the model simulates the time series of groundwater level under the baseline and under the coal resource development pathway (CRDP). The maximum difference in drawdown (dmax) between the modelled CRDP and baseline, due to additional coal resource development, and the year of maximum change (tmax), are calculated as the difference between the two time series. At points along the prescribed stream network, the model also generates the change in surface water – groundwater flux. The resulting changes in flux time series are inputs to the river model and are encapsulated in the streamflow hydrological response variables reported in companion product 2.6.1 for the Namoi subregion (Aryal et al., 2018b).

A subset of 37 groundwater model parameters is allowed to vary stochastically to form the basis for the sensitivity and uncertainty analyses. The groundwater levels and surface water – groundwater flux are most sensitive to depth of incision of the streambed (dh), the scaler on the diffuse recharge (Scale_r_dr) and the hydraulic properties of the alluvium (al1_kh, al1_SY and al2_kh).

The main prediction, the drawdown due to additional coal resource development, is sensitive to the hydraulic properties of the interburden (IB1_k0, IB1_kv, IB1_ka) and the hydraulic properties of the coal-bearing units (hos_k0, hos_ka). As the groundwater level and streamflow observations are not sensitive to these parameters, these parameters will not be constrained greatly in the uncertainty analysis.