2.6.2.8.3 Qualitative uncertainty analysis


The major assumptions and model choices underpinning groundwater modelling in the Maranoa-Balonne-Condamine subregion are listed in Table 13. The goal of the table is to provide a non-technical audience with a systematic overview of the model assumptions, their justification and effect on predictions, as judged by the modelling team. This table is aimed to assist in an open and transparent review of the modelling. Each assumption is scored on four attributes using three levels: high, medium and low. Beneath the table, each of the assumptions are discussed in detail, including the rationale for the scoring.

The data column is the degree to which the question ‘if more or different data were available, would this assumption/choice still have been made?’ would be answered positively. A ‘low’ score means that the assumption is not influenced by data availability while a ‘high’ score would indicate that this choice would be revisited if more data were available. Closely related is the resources attribute. This column captures the extent to which resources available for the modelling, such as computing resources, personnel and time, influenced this assumption or model choice. Again, a ‘low’ score indicates the same assumption would have been made with unlimited resources, while a ‘high’ value indicates the assumption is driven by resource constraints. The third attribute deals with the technical and computational issues. ‘High’ is assigned to assumptions and model choices that are predominantly driven by computational or technical limitations of the model code. These include issues related to spatial and temporal resolution of the models. The final, and most important column, is the effect of the assumption or model choice on the predictions. This is a qualitative assessment by the modelling team of the extent to which a model choice will affect model predictions, with ‘low’ indicating a minimal effect and ‘high’ a large effect. The precautionary principle is applied for assumptions with a large potential impact on model predictions; that is, the hydrological change is overestimated, rather than underestimated.

Beneath the table, each of the assumptions is discussed in detail, including the rationale for the scoring. The goal of the table is to provide a non-technical audience with a systematic overview of the model assumptions, their justification and effect on predictions, as judged by the modelling team.

Table 13 Qualitative uncertainty analysis for the Maranoa-Balonne-Condamine subregion

In this table, each assumption is scored on four attributes using three levels: high, medium and low. For example, the data column is the degree to which the question ‘if more or different data were available, would this assumption/choice still have been made?’ would be answered positively. In other words, a ‘low’ score means that the assumption is not influenced by data availability while a ‘high’ score would indicate that this choice would be revisited if more data were available.


Assumption / Model choice

Data

Resources

Technical

Effect on predictions

Model horizontal and vertical discretisation

medium

high

high

medium

Model code and solver

low

low

high

low

Model boundary conditions

medium

low

low

low

Surface water – groundwater interactions

high

high

medium

low

Assume constant rate non-P&G extractions

high

low

low

low

Mine pit dewatering represented by MODFLOW drain cells

medium

high

medium

high

CSG activities simulated by MODFLOW evapotranspiration cells

low

medium

high

medium

Spatial interpolation of hydraulic conductivity, recharge and storage values

medium

low

low

low

Quantitative uncertainty analysis using 200 calibration-constrained parameter sets

medium

high

low

medium

CSG = coal seam gas, P&G = petroleum & gas

Model horizontal and vertical discretisation

The OGIA model is conceptualised with 19 layers consisting of 1.5 km x 1.5 km grid cells to represent all major aquifers and aquitards as described in detail in Section 2.6.2.3.3 . The vertical and horizontal discretisation is partly driven by technical limitations as a higher resolution model would have more grid cells, which would make model runtimes and storage requirements exceedingly difficult.

Horizontal discretisation affects predictions, particularly where large hydraulic gradients are to be expected. Model predictions are most likely to be affected by horizontal discretisation in the immediate vicinity of coal mines where hydraulic gradients are steepest. Local-scale modelling is the only way to resolve this issue. Relatively simple representations of the Main Range Volcanics and Condamine Alluvium have been adopted in the model with each unit represented by a single layer for each unit. The integrated approach that OGIA used to model groundwater levels in the Condamine Alluvium using the more detailed Condamine Model is an example of using a more detailed local-scale model to make predictions at an appropriate scale.

Vertical discretisation is driven by the hydrostratigraphic interpretation of the geological basin at a regional scale and available modelling resources. While more detailed interpretations may be possible at a local scale, this stratigraphic interpretation is justified at the regional scale and sufficient data are available to create the top and bottom surfaces for each hydrostratigraphic unit. The Walloon Coal Measures are represented in the OGIA model using three model layers: an upper and lower layer representing generally low permeability mudstone and a composite middle representing all productive coal seams and inter-bedded low permeability sediments. While an increased vertical resolution allows for a more accurate representation of hydraulic properties, especially storage values, there are insufficient data to justify parameterising and constraining a higher vertical resolution at the regional scale. This will affect model predictions, but is compensated for by the calibration-constrained hydraulic conductivity parameters (Kh and Kv) that vary over several orders of magnitude in each layer of the OGIA model.

Model code and solver

MODFLOW is one of the industry standard codes for solving the finite difference groundwater flow equations and has a large number of different modules available for simulating different groundwater flow processes. But MODFLOW and other similar groundwater flow models are only able to simulate water movement and hence are unable to simultaneously simulate dual-phase flow. In the short-term and at the local scale, interaction between gas and water phases is an important factor governing water flow from CSG wells. Hence, the model has limitations in accurately estimating the short-term well yields and pressure recovery.

Model boundary conditions

The no flow boundary conditions to the north-east and north-west of the model domain coincide with the Surat and Bowen geological basin boundaries. There is no evidence of groundwater flow across these boundaries, meaning that the no flow boundaries are appropriate in this part of the model. MODFLOW general head boundary cells are used to simulate groundwater flow across the Surat and Clarence-Moreton basin boundaries to match hydraulic gradients derived from observed groundwater levels. Parameterisation of this boundary condition is limited by analysis of available groundwater level data. However, the OGIA model boundaries are sufficiently distant from the modelled CSG and coal mine areas, which means that these boundary conditions are unlikely to affect model predictions.

Surface water – groundwater interactions

Surface water – groundwater interactions in the OGIA model are simulated using MODFLOW drain and river packages. It is assumed that all surface watercourses act as groundwater discharge boundaries, meaning that groundwater only flows from the aquifer into the watercourse and that groundwater does not recharge the aquifer from the watercourse. This conceptualisation is consistent with previous studies of groundwater fluxes in the alluvial systems overlying the Great Artesian Basin (GAB) (Hillier, 2010). Groundwater is recharged in the GAB aquifer outcrop areas, with some groundwater discharging to watercourses via the alluvium, particularly in wetter years to sustain baseflows (QWC, 2012). This is a conservative approach to predicting groundwater drawdown, as it means recharge from surface watercourses cannot affect predicted groundwater level drawdown due to licensed extractions or for petroleum and gas (P&G) extractions.

This approach is not appropriate for the Condamine Alluvium, which is an important water source for irrigation, stock and domestic and town water supplies. Instead, OGIA used an integrated modelling approach for the Condamine Alluvium. Groundwater fluxes between the Walloon Coal Measures and the Condamine Alluvium model layers predicted by the regional model are set as boundary conditions for the more detailed Condamine Model to predict alluvial groundwater levels. Minimal additional groundwater drawdown less than 1.0 m (p=0.05) is predicted in Model layer 10 – Walloon Coal Measures under the eastern edge of the Condamine Alluvium in the vicinity of the New Acland Coal Mine. This is not considered likely to cause additional groundwater drawdown in the Condamine Alluvium. For this reason, the more detailed Condamine Model is not used to revise surface water – groundwater interactions in the Condamine Alluvium for BA. Available data, resources and technical issues limit representation of surface water – groundwater interactions in the regional model. However, this is likely to have a minimal effect on the regional-scale model predictions, which are confined to the deeper model layers for which the OGIA model was developed.

The focus on the deep regional aquifers targeted by CSG development means that the OGIA model may not on its own be suitable for assessing hydrological changes in surficial aquifers that are important in representing impacts to surface water – groundwater interactions and groundwater-dependent ecosystems. Improved model predictions for risk and impact analysis of surface water – groundwater interactions and groundwater-dependent ecosystems in the surficial aquifers would require significant additional investment in modelling resources, but is likely to be limited by water quality and quantity data availability.

Assume constant rate non-P&G (petroleum and gas) extractions

All groundwater extractions except CSG depressurisation and mine dewatering are considered to be occurring at a constant rate for individual bores. This assumption was mainly driven by the limited availability of time series data of groundwater extractions. Information on the aquifer from which water is taken and the volumes of extraction were largely absent. Hence, the missing information is estimated from geological mapping and other information. Extraction volumes are estimated using the best available information and follow the methodology used in the preparation of the Great Artesian Basin Water Resources Plan and Murray Darling Basin Plan. In accordance with previous water resources studies, groundwater extraction is assumed to be at the full entitlement level. The trajectory modelling approach used by OGIA and BA, where reported drawdown is the difference between the baseline and the development case, means that non-P&G extraction volumes have a minimal impact on predicted cumulative impacts of coal resource development.

Mine pit dewatering represented by MODFLOW drain cells

Simulation of dewatering of the open-cut coal mines using the MODFLOW drain package is affected by drain cell elevations, which are derived from the regional-scale hydrostratigraphic interpretation of the geological basin. More accurate drain cell elevations and pit geometry are available in local-scale models, such as those used during the environmental impact assessments. However, the regional-scale hydrostratigraphic interpretation and pit geometry is justified for the 1.5 km × 1.5 km grid cells used in the regional model. While data are available to improve model predictions of mine pit dewatering at a local scale, the conservative approach used for the BA modelling is driven by limited modelling resources and technical and computation requirements that are inherent in regional-scale modelling. The predictions of cumulative drawdown impacts are considered appropriate at a regional scale. However, predicting absolute drawdown in the vicinity of the mines is beyond the capability of this representation.

Coal seam gas activities simulated by MODFLOW evapotranspiration cells

The MODFLOW EVT package used to simulate groundwater extraction associated with petroleum and gas activities, including CSG, includes historical and planned extraction rates. This dataset is revised annually using the development scenarios provided by the tenure holders to OGIA each year (OGIA, 2014). The MODFLOW EVT package sets user defined volumes of water extraction that are subject to an additional control on resulting head. This is a superior approach for representing CSG depressurisation in MODFLOW models compared to using the well or drain packages.

While the representation of CSG depressurisation as single-phase flow in MODFLOW in the OGIA model is not limited by data availability, resources or technical issues, MODFLOW is not able to simulate dual-phase flow. CSG depressurisation using MODFLOW model code over-estimates produced water volumes and hence groundwater drawdown during the initial production periods. A better method for representing CSG depressurisation using the MODFLOW code is currently unavailable. Alternative modelling approaches using dual-phase model code have greater data, resources and technical or computational requirements and would improve the accuracy of predicted drawdowns.

Spatial interpolation of hydraulic conductivity, recharge and storage values

Hydraulic conductivity, recharge and storage values are known to be heterogeneous and spatially variable. The regional model is calibrated using water levels or water pressure measurements for 1541 bores. The parameter sets – horizontal (Kh) and vertical (Kv) hydraulic conductivity, vertical anisotropy, recharge rates and specific storage (Ss) / specific yield (Sy) – are included in the model using pilot points (GHD, 2012). Pilot points are located 45 km apart, reducing to 15 km or less in areas and layers of particular interest. This approach resulted in 200 sets of horizontal and vertical parameter values that varied spatially, where the overall difference between observed and predicted water levels is within the acceptable calibration limits.

Spatial parameterisation using the pilot point approach to define calibration-constrained parameter sets for the uncertainty analysis reduces the reliance of model predictions on heterogeneous parameters that are inherently data limited. This uncertainty analysis approach means that additional estimates of parameter values do not limit model predictions. However, additional time series of groundwater levels would potentially reduce predictive uncertainty.

Quantitative uncertainty analysis using 200 calibration-constrained parameter sets

The quantitative uncertainty analysis numerically evaluates the uncertainty in the hydraulic conductivity fields. However, this does not account for uncertainty in other hydraulic properties and boundary conditions. While it is technically possible to include these parameters, the number of model runs and data required are prohibitive in the context of a regional-scale model. The effect of the individual components on model predictions is discussed in 2.6.2.8.1 . The existing uncertainty analysis accounts for the potential effects of hydraulic conductivity, recharge and storage values on model predictions, but not model conceptualisation or the parameters used to specify drain and river boundary conditions.

The 200 calibration-constrained parameter sets obtained from OGIA are used for the quantitative uncertainty analysis. These parameter sets are calibrated to pre 1995 groundwater levels that represent a period before significant coal seam gas extraction (CSG) and open-cut coal mine development affected the regional groundwater levels. For this reason, the OGIA model was not recalibrated when the open-cut coal mine boundary conditions were added for the BA groundwater modelling in the Maranoa-Balonne-Condamine subregion. Model calibration with additional groundwater level or flux observations and a finer resolution model grid in the vicinity of the coal mines would improve model predictions, but is not feasible in the existing regional-scale groundwater model.

Last updated:
17 October 2018
Thumbnail of the Maranoa-Baloone-Condamine subregion

Product Finalisation date

2016
PRODUCT CONTENTS