It is challenging for rainfall-runoff modelling to simulate low-streamflow metrics right (Nicolle et al., 2014). Firstly, since the daily flow rate at P01 can be many orders of magnitude smaller than the high daily flow rate at P99, it makes the AWRA-L model almost always over estimate the low-streamflow hydrological response variable P01 when it is calibrated using the Nash–Sutcliffe efficiency based objective function using normal streamflow or using the Box-Cox transformed streamflow. As a result of over estimation of low flows the duration hydrological response variables such as ZFD, LFD, LFS and LLFS are under estimated. There are no widely accepted rules to determine the degree of transformation needed to help eliminate this problem (Pushpalatha et al., 2012). Pushpalatha et al. (2012) used a number of flow transformation criteria and Nash–Sutcliffe efficiency to evaluate the efficiency of hydrological models and found that for most of the criteria high flows still make a significant contribution to the objective function value (see also Zhang et al. (2014)).

Secondly, the quality of the data gets degraded at the lower end of streamflow measurement due to uncertainty related to the rating curve as a result of frequent changes of channel bed geometry caused by floods and large flows. For example, Tomkins (2014) investigated the rating curve uncertainties from the Namoi river basin, and found that most gauges showed significant deviations at low stages, affecting the determination of low streamflow.

In the bioregional assessment (BA), the effects of over estimation of low streamflow are lessened by taking outputs from 3000 simulations using unique parameter sets applied to both AWRA-L and AWRA-R. As discussed in Section, the selected 300 best parameters are used for the prediction providing suitable uncertainty bounds for the changes in hydrological response variables caused by additional coal resource development.

Furthermore, the BA determines the effects of additional coal resource development on the water resources given by the difference between baseline and CRDP futures. As the errors introduced due to the uncertainty of model parameters are common to both baseline and CRDP, they cancel out when the net effect is calculated by taking the difference between the two pathways.

Last updated:
6 December 2018
Thumbnail of the Namoi subregion

Product Finalisation date