CW i MI

Centre for Water in the Minerals Industry Sustainable Minerals Institute University of Queensland Brisbane 4072 Australia

Purpose: To facilitate the capture of benefits from opportunities associated with sustainable water management in the minerals industry

CWiMI

Scoping Study: Groundwater Impacts of Coal Seam Gas Development – Assessment and Monitoring Mr. Mal Helmuth Executive Director LNG Projects Infrastructure and Economic Development Group Department of Infrastructure and Planning

Document reference: P08-010-002.doc Project Team: Sue Vink, Nadja Kunz, Damian Barrett, Chris Moran Edited: 22 December 2008

Table of Contents Executive Summary ................................................................................................................................ 5 1

Background ................................................................................................................................... 13

2

Scope and Objectives ................................................................................................................... 13

3

Approach ....................................................................................................................................... 14

4

5

3.1

Assumptions and limitations.................................................................................................. 14

4.1

Location and characteristics of coal seams in Queensland .................................................. 14

4.3

Coal seam gas formation, retention and extraction .............................................................. 18

Coal seam gas in Queensland ...................................................................................................... 14

4.2

Coal Seam Gas Production in Queensland .......................................................................... 16

4.4

Gas and water production from CSG in Queensland ........................................................... 18

5.1

Model Structure and modelling methodology........................................................................ 21

5.3

Assumptions and uncertainties ............................................................................................. 29

6.1

Determining risks to groundwater aquifers resulting from CSG extraction ........................... 35

6.3

Hydrological Risk Formulation for Bowen and Surat Basins ................................................ 39

6.5

Summary of Aquifer Risks ..................................................................................................... 45

CSG Industry Development Scenarios and Implications for Water Production ............................ 20

5.2 6

7

Results and Discussion ......................................................................................................... 24

Assessment of Aquifer Risks ........................................................................................................ 32

6.2

Hydrological risks .................................................................................................................. 36

6.4

Risk assessment assumptions and data gaps ...................................................................... 42

8

Monitoring to manage risks ........................................................................................................... 45 Conclusions ................................................................................................................................... 48

References ............................................................................................................................................ 50 Appendix A – Areal Extents of Aquifers Relative to the Surat and Bowen Basin coal measures ........ 54 Appendix B – Aquifer Risks and Factors Contributing to Aquifer Risks. .............................................. 57 Appendix C– Using isotopic techniques to trace flows between coal seams and aquifers. ................. 61 Appendix D. Monitoring impacts on groundwater dependent ecosystems ........................................... 62

CWiMI

2

Figures Figure 1. Location of shallow coal bearing regions throughout Queensland (A) and major CSG projects (B) .............................................................................................................................................. 5 Figure 2. Gas and water production reported in 2007 from CSG tenements in the Bowen and Surat basins.. .................................................................................................................................................... 6 Figure 3. Coal seam gas production scenarios for period 2008 – 2020 (A) and corresponding estimates of CSG water produced (B) based on possible production figures of 10, 28 and 40.8 Mt per annum provided by the Queensland Department of Infrastructure and Planning ................................... 6 Figure 4. Spatial distribution of water production in 2020 for three possible CSG development scenarios of 10Mtpa (A), 28Mtpa (B) and 40 Mtpa (C).. ......................................................................... 8 Figure 6. Relative risk assessment for a) aquifers overlying coal measures and b) aquifers underlying coal measures. ...................................................................................................................................... 11 Figure 7. Distribution of shallow coal bearing areas in the depositional basins in Queensland. ......... 15 Figure 8. a) Distribution of coal seam gas tenements and wells in the Bowen and Surat Basins. The names of the major producing fields are noted. b) The 2P reserves estimated for currently active tenements are shown by the shaded colour scale. Exploration and currently non-producing tenements are shown unshaded, existing gas pipelines and towns. c) Tenements in relation to coal measures from which gas is being extracted. ........................................................................................................ 17 Figure 9. Typical gas and water production profile for a CSG well ...................................................... 19 Figure 10. Gas and water production reported in 2007 from CSG tenements in the Bowen and Surat basins. Aggregate production values are reported every 6 months by companies for each tenement. .............................................................................................................................................................. 20 Figure 11. Schematic of the coal seam gas and water production model for the Bowen and Surat basins in central Queensland. ............................................................................................................... 23 Figure 12. The ‘cost function’ per tenement used to determine the order in which tenements come online to meet gas production scenarios.. ............................................................................................ 23 Figure 13. Coal seam gas production scenarios for period 2008 – 2020 based on possible production figures of 10, 28 and 40.8 Mt per annum provided by the Queensland Department of Infrastructure and Planning. ........................................................................................................................................ 24 Figure 14. Trajectories of model output for coal seam gas water production (ML/year).. .................... 25 Figure 15. Spatial estimates of water production rates (ML/year) in 2020 for tenements in the Bowen and Surat basins of central Queensland.. ............................................................................................. 28 Figure 16. Effect of uncertainty in 2P gas reserves on number of active tenements required to meet gas production targets of 10, 28, and 40.8 Mtpa.. ................................................................................ 31 Figure 17. Effect of uncertainty in 2P gas reserve on spatial distribution of water production by tenement (ML/yr) for gas production targets of (a) 10, (b) 28, and (c) 40.8 Mtpa.. .............................. 32 Figure 18. Location of GAB management areas showing recharge springs in relation to CSG tenements.............................................................................................................................................. 33 Figure 19. Typical stratigraphic sequences observed in the areas of CSG production. ....................... 34

CWiMI

3

Figure 20. Summary of the hypothesised risks to aquifers in the Bowen and Surat Basins ............... 37 Figure 21. Schematic of hypothesised risks to aquifers from dewatering during CSG extraction.. ..... 38 Figure 22. Hypothesised risk to other overlying aquifers ..................................................................... 39 Figure 23. Relative risk to aquifers a) overlying coal seams and b) underlying coal seam in the Bowen and Surat Basins. .................................................................................................................................. 41 Figure A 24 – Areal extent of Hutton (A) and Springbok (B) aquifers relative to the shallow coal extent in the Surat and Bowen Basins. ............................................................................................................ 55 Figure A 25 – Areal extent of Condamine Alluvium (C) and Precipice Aquifer (D) relative to the shallow coal extent in the Surat and Bowen Basins ............................................................................. 56 Figure B 26. Calculated aquifer storage for aquifers overlying coal seams (A) and Hutton aquifer that underlies Walloon coal measures (B). .................................................................................................. 58 Figure B 27. Calculated coal storage for aquifers that overlie and underlie coal seams. .................... 59 Figure B 28. Distance between coal and aquifer for aquifers overlying coal seams (A) and Hutton aquifer that underlies Walloon coal measures (B). ............................................................................... 60

Tables Table 1. Summary of assumptions and uncertainties used in spatial analysis of gas and water production................................................................................................................................................ 7 Table 2. Assumptions made and associated uncertainties in the aquifer risk assessment. ................ 10 Table 3. Characteristics of the coal measures of Bowen and Surat Basins (Draper and Boreham 2006) ..................................................................................................................................................... 16 Table 4. Gas and water production and 2P (‘proved’ and ‘probable’) reserves per unit area for four coal measures from the Bowen and Surat Basins.. .............................................................................. 20 Table 5. Water production rates in 2020 for three gas production target scenarios (10, 28, 40.8 Mtpa).. .............................................................................................................................................................. 26 Table 6. Summary of assumptions and uncertainties used in spatial analysis of gas and water production.............................................................................................................................................. 29 Table 7. Average aquifer properties, coal measure thickness and distance between aquifers and coal measures............................................................................................................................................... 42 Table 8. Summary of assumptions and uncertainties used in risk aquifer assessment. ...................... 43

CWiMI

4

Executive Summary This scoping paper was commissioned by the Department of Infrastructure and Planning (DIP) to develop a better understanding of the potential risks posed to regional and local aquifer systems by the development of a coal seam gas-based Liquefied Natural Gas (LNG) industry in Queensland. The objectives of the study were to: Provide background information on potential groundwater impacts resulting from the expansion of the coal seam gas (CSG) industry; Provide a broad estimate of water production (and uncertainties) resulting from expansion of the CSG industry; and Propose an approach for effective monitoring of groundwater impacts due to CSG production. The areas considered in this paper were the Surat and Bowen Basins. Assessments of both estimated water production and aquifer risks were made using data provided by Queensland Department of Mines and Energy (DME) and Natural Resources and Water (NRW). Scenarios of CSG industry expansion were provided by Department of Infrastructure and Planning (DIP). CSG tenements have been granted by the Queensland Government throughout the Bowen and Surat Basins. CSG is being produced from four coal measures (Figure 1A) which have distinct depositional and tectonic histories and therefore different properties including gas and water content, permeability and porosity. The relationship between gas and water production is variable both within and between measures (Figure 2). Current gas production is concentrated in six major developments (Figure 1B) and so any impacts on groundwater resources are likely to be greatest in these areas.

Figure 1. Location of shallow coal bearing regions throughout Queensland (A) and major CSG projects (B)

CWiMI

5

Figure 2. Gas and water production reported in 2007 from CSG tenements in the Bowen and Surat basins. Aggregate production values are reported every 6 months by companies for each tenement.

Water Production Given the variability in gas and water relationships across the Basins (Figure 2) and other uncertainties (discussed below), accurately predicting the quantities of water likely to be produced during industry expansion is challenging. Therefore, broad estimates of water production under three possible CSG development scenarios (as provided by the DIP) were made by developing a simple ‘conceptual’ model. This generalised water and gas accounting model determined water yield from CSG production by tracking the addition of new wells on current production tenements and the ‘activation’ of new tenements brought online to meet gas production targets over a 20 year time horizon. The three scenarios were based on the stated development goals of proponent companies that will result in up to six LNG production plants over the next 20 years generating 550, 1885 and 2262 PJ equivalents of LNG in 2025 (Figure 3A). Resulting estimates of water co-produced with CSG are presented in Figure 3B.

Figure 3. Coal seam gas production scenarios for period 2008 – 2020 (A) and corresponding estimates of CSG water produced (B) based on possible production figures of 10, 28 and 40.8 Mt per annum provided by the Queensland Department of Infrastructure and Planning

CWiMI

6

The model makes simplifying assumptions (Table 1) about gas and water production per well, where new wells are added, the domestic and international demand for gas production and what determines the addition of wells onto tenements that are currently ‘inactive’ (i.e. tenements that have been explored but are currently not used to meet domestic gas production targets). This model can be used to explore scenarios for assessing possible water production rates throughout the Bowen and Surat Basins but does not constitute an accurate prediction of how the exploitation of gas reserves, the concurrent production of coal seam water and regional effects on surface aquifers will evolve through time. This approach was deemed to be the best compromise between achieving the outputs sought by the client and the limited availability of data with which to constrain the model. Table 1. Summary of assumptions and uncertainties used in spatial analysis of gas and water production. Assumptions

Associated Uncertainty/Implications

Well production rates for each measure were set at an average rate based on available well data. The model did not account for variation in gas and water production over time.

This assumption is only reasonable if the well logs contain time dependent variation in well production, and if there is limited spatial variation in well production across a coal measure.

Gas and water production does not decline to zero due to depletion of all water reserves within a tenement.

This simplifying assumption is unlikely to be accurate; e.g. modelling predictions by Harbison et al. (2008) 2 show that 72 wells pumping within a 225 km area of the Surat basin would extract all water in the vicinity of those wells within 11 years.

New tenements were opened to gas production when either: - Average well density in a tenement exceeded 1km .

These assumptions do not consider true economic factors nor reflect individual company strategies towards CSG development. For example, the model does not consider rapid expansion of wells on tenements owned by any single company.

New tenements were added according to a cost function that considered the average distance per tenement from existing and planned LNG pipeline infrastructure and existing towns

As above, this assumption neglects the economic drivers of gas exploration and company strategies that would drive the actual spatial pattern of CSG development.

New wells were added to a tenement in order to meet the production target for each year

This assumption was made such that the model met projected CSG estimates; it is yet to be determined if recoverable reserves can support this level of development.

2P Reserves extrapolated from estimates for existing active tenements in 2007 based on coal measure and tenement area.

Highly uncertain estimates of gas reserve but used only because no other means of calculating reserve was available. May impact on water yield estimates by model if current reserves are grossly underestimated.

- Gas reserves in a tenement drop below 50%; or 2

The modelled spatial distribution of water production in 2020 is shown in Figure 4 for the three gas production scenarios. The effect of uncertainty in gas production data on water yield was also estimated (see dotted lines in Figure 3). Within the model, a ‘cost function’ was used to predict the spatial development of CSG by assuming that new tenements were exploited preferentially based on proximity to existing infrastructure. The model predicts that the increase in gas production, the opening of new tenements and concomitant increase in water yield occurs preferentially throughout

CWiMI

7

the Walloon and Bandanna coal measures towards the south and southeast of the Bowen-Surat Basins; this is largely due to the relatively dense network of LNG pipelines already existing in this area. The relatively high water production rates of wells in the Walloon and Bandanna coal measures resulted in larger water yields per unit gas production than would occur if these wells were preferentially sunk in the Moranbah or Baralaba measures. Exploitation of reserves in the northern part of the Basin would be promoted in the model if penetration of the LNG pipeline network into the Bowen tenements was more extensive. Because new wells and tenements were preferentially added to measures in the Surat Basin, the variation in water production to meet gas production targets was a linear function of well number. As a result, the proportion of wells located in the Bowen relative to Surat Basin was forecast to be less in 2020 than what it is today.

Figure 4. Spatial distribution of water production in 2020 for three possible CSG development scenarios of 10Mtpa (A), 28Mtpa (B) and 40 Mtpa (C). Red arrows depict tenements brought on-line to meet the gas production target; dark blue tenements produce the most water.

Aquifer Risks Much of the area currently under development for CSG (notably in the Surat and central Bowen Basins) is coincident with aquifers of the Great Artesian Basin. In addition, in the Surat Basin the Walloon Coal Measures are a GAB aquifer and also underlie the Condamine alluvium aquifer which is the highest allocated groundwater source in the State. While some evidence, based on water quality variations, exists to suggest that some coal seams are currently or have recently been recharged from overlying strata, at present no studies (at either a local or regional scale) have been undertaken to directly assess hydraulic connectivity between aquifers that overlie or underlie coal seams in the Surat and Bowen Basins.

CWiMI

8

The potential impacts of CSG extraction on surrounding aquifers are presented in this report as hypothesised risks. The key hypothesised risk is that connectivity between the aquifers and coal seam may be altered due to a change in hydraulic conditions that currently maintain equilibrium flow conditions within and between aquifers (termed hydrological risk). Three aquifer/coal measure relational cases were identified for the risk assessment. These were: 1) An aquifer overlying a coal seam; 2) an aquifer underlying a coal seam; and 3) two or more aquifers interacting. The magnitude of the hydrological risk to an aquifer at any location will depend primarily on the magnitude of the change in head, due to dewatering of the coal seam, and capillarity change due to both water and gas removal from the coal pores. The risk, however, will be moderated by the properties of the aquifers, the interbedding strata and coal seams, the geological setting of the formations and the vertical distance between the aquifer and the coal seam. The properties used in the formulation of relative risk in this study were: 1. The proximity of the coal seam and aquifers (i.e. vertical distance, d, between coal and aquifers, where increasing distance or thickness of intervening layers will decrease the risk of impact on a particular aquifer) 2. The amount of water available (aquifer storage = porosity ( ) x thickness (T)) 3. The speed with which water can move (i.e. permeability (K)) The output of the risk analysis was a set of maps that spatially displayed point estimation of relative risk to underlying and overlying aquifers that may result from CSG extraction. From the bore and well drilling data acquired from DME and NRW, there was limited spatial coverage of data for risk parameters and also very little basic data available on either coal aquifers or groundwater aquifers upon which to develop meaningful contoured risk surfaces. In particular, there was limited data available to assess the northern Bowen Basin, as this area is outside the GAB and groundwater supply is not considered a major water source for the region. The assumptions that were made in formulating the relative risk assessment are presented in Table 2 and arise primarily from insufficient data. These knowledge gaps need to be addressed both through acquisition of better quality (either greater spatial coverage or detailed stratigraphy) data and on-going monitoring in future assessments.

CWiMI

9

Table 2. Assumptions made and associated uncertainties in the aquifer risk assessment. Assumptions Average properties of permeability, connectivity and porosity apply across each coal measure. Permeability is also assumed to be isotropic.

Associated Uncertainty/Implications Uncertainty over the movement of water between interburden and coal seams. The likelihood of different lateral and vertical flows, their spatial arrangement and change of hydrological conditions needs to be understood and quantified. The implications range from faster water movement than the risk assessment implies to potentially no water movement.

Average porosity and permeability values were used for each coal measure and aquifer.

Calculated risks do not demonstrate variation in risk within a coal measure due to differences in porosity and permeability.

Porosity and permeability for the Moranbah coal measures were assumed to be the same as for the overlying Rangal coal measures.

No other data was available.

Porosity and permeability was assumed for the Condamine Alluvium and Tertiary Basalt based on book values for similar strata (Fetter, 2001).

No other data was available.

The entire coal measure will be dewatered during CSG extraction.

This assumption was necessary because stratigraphic data for the cores only reported depths to the top and bottom of each coal measure, not individual seams. The change in hydraulic pressure (head) caused by dewatering the coal seam needs to be quantified. There is also a need to determine the lateral extent

of coal seam dewatering. Porosity and permeability of the interburden and coal seams is similar.

The association of aquifers and coal facies within measures is largely unknown. More detailed stratigraphy is required to accurately determine the distance and type of strata that lie between aquifers and dewatered coal seams. Depending on the properties of such strata, the time scales required for movement of water between aquifers and coal seams may be very slow and risks could be significantly less than what is hypothesised in this report.

The confined aquifers are full of water.

This assumption is reasonably robust for most areas but would clearly not be the case in recharge areas.

The thickness of the strata represents the thickness of the aquifer. Fracturing and folding was not considered.

There was insufficient data to estimate the impact of fracturing and folding. Further, the impacts of fracturing depend on localized conditions; it may in fact decrease connectivity through realignment of strata and re-mineralisation within fractures.

The normalised hydrological risk to each aquifer is presented in Figure 5. It must be emphasised that the risk ratings are a relative measure and should thus be used only as a guide for further investigation, monitoring and risk assessment.

CWiMI

10

Figure 5. Relative risk assessment for a) aquifers overlying coal measures and b) aquifers underlying coal measures.

It can be seen from these figures that the highest relative risks are generally to the Hutton, Springbok and Condamine Alluvium aquifers from dewatering of the Walloon Coal Measures in the Surat Basin. This is largely due to the relatively close vertical distance between these aquifers and coal measures. The risk to the Condamine aquifer is also relatively higher in areas where the Springbok Sandstone is thin or absent. Overall there is a higher relative risk to aquifers in the Surat Basin compared to other aquifer /coal associations. With the exception of the area around Moranbah, most of the Bowen Basin could not be assessed due to lack of stratigraphic and aquifer information. Monitoring Risks To manage these potential risks and sustainably manage the potential water resource derived from CSG production additional work is required in three areas. These are: 1. Fill key knowledge gaps; 2. Monitoring; and 3. Cumulative impacts assessment. It is highly likely that the CSG companies hold some of the information that can be used to fill some of the key knowledge gaps and uncertainties in the analyses presented in this paper. In particular water production during exploration and detailed stratigraphy and porosity/permeability and fracturing are

CWiMI

11

both critical elements required to enable gas production for the companies. This information would be useful in providing a better assessment of potential risks to aquifers in areas close to the gas fields. Engagement with CSG companies should be considered as a crucial early step to enable further analysis and design the monitoring program. Monitoring should be undertaken in a phased approach using an adaptive management framework. In this way new findings from a number of different lines of evidence can be incorporated into resource planning decisions as the information becomes available. Any monitoring program designed to evaluate impacts on aquifers from CSG development should include conventional well/bore monitoring and hydrological modelling combined with new approaches and technologies including geochemical and isotopic tracing and dating techniques and remote sensing. An assessment of the positive and negative cumulative impacts of CSG development in the Bowen and Surat Basins should also be considered as imperative to guiding policy decisions aimed at successfully and sustainably managing CSG extraction and long term land uses.

CWiMI

12

1 Background This scoping paper was commissioned by the Department of Infrastructure and Planning (DIP) to develop a better understanding of the risks posed to regional and local aquifer systems by the development of a coal seam gas-based Liquefied Natural Gas (LNG) industry in Queensland. The coal seam gas (CSG) industry is developing rapidly in Queensland, with known reserves of CSG currently exceeding domestic demand for the resource. Producers are seeking new markets for CSG by establishing an LNG industry. LNG attracts a higher price for gas on the global market than on Queensland’s domestic market. Increasing global demand for LNG has stimulated CSG exploration and more CSG reserves are viewed as commercially recoverable due to higher world prices for LNG. Cumulative CSG production from the Surat and Bowen basins from 2000 (when commercial production began) to June 2007 was about 646 petajoules (PJ) (11.7 Mt). Annual production for 2007-08 was 150PJ (2.7Mt). It is estimated that CSG requirements for six announced LNG plants under a scenario of fully expanded production over a 20 year period could reach approximately 46,500 PJ (around 42Mtpa). While it is yet to be determined whether recoverable reserves of CSG can support this level of development, it is clear there is pressure for rapid, ongoing development of CSG. Significant quantities of water are produced during CSG production as a result of dewatering coal seams to release the gas. In 2007, 12.5 gigalitres (GL) of water were produced in Queensland as a result of CSG production. Production of CSG from the Surat basin for domestic production will produce an estimated annual average of 25GL of CSG water for the next 25 years. A CSG-based LNG industry of between 10 Mtpa and 40 Mtpa (equivalent to between 550 PJ and 2,220 PJ) could result in between 55 GL and 540 GL of additional CSG water each year. Over a 20 year period, a LNG industry of between 1,100 PJ and 44,300 PJ could produce between 110 GL and 11,200 GL of CSG water. The potential scale of development of CSG and the associated dewatering of the coal seam aquifers has raised concerns about impacts on surrounding aquifers. Currently, there is limited understanding of the connectivity between coal seam aquifers and regional and local aquifer systems.

2 Scope and Objectives This project was a scoping study with the purpose of collating and documenting existing information available through State Government sources and proposing a basis for future, more detailed assessment of potential impacts of the CSG industry growth on groundwater systems.

CWiMI

13

The objectives of the study were to: 1. Provide background information on potential groundwater impacts resulting from the expansion of the CSG industry; 2. Provide a broad assessment of the water supply options resulting from the expansion of the CSG industry; and 3. Propose an approach for on-going monitoring of groundwater impacts during development of the CSG industry.

3 Approach The project scope was geographically restricted to the Surat and Bowen Basins. The work has been conducted using open source and confidential data provided by Department of Natural Resources and Water (NRW) and Department of Mines and Energy (DME). While it is recognised that CSG companies are likely to hold more detailed and comprehensive data that would allow more precise and detailed assessment of impacts particularly at local scales, the timeframe set for this work precluded acquisition of company data. Economic forecasts of the likely expansion of the industry were provided by the Department of Infrastructure and Planning (DIP). The project was guided by a steering committee consisting of staff from NRW, DME and DIP. A review of issues surrounding coal seam gas industry was conducted using journal literature and government and industry websites. The project also benefitted from discussions with key individuals from University of Queensland Department of Earth Sciences and Chemical Engineering, DME, Geological Survey of Queensland (GSQ), and NRW.

3.1 Assumptions and limitations No hydrodynamic modelling was conducted in the project due to the lack of individual well data and detailed geological information and physical properties, particularly related to the permeability and porosity of overlying or underlying strata (particularly the confining layers) in the vicinity of the CSG wells. In addition there was no information available regarding the hydraulic properties of the interburden of the coal measures or details of the spatial relationship of coal facies within the measures.

4 Coal seam gas in Queensland 4.1 Location and characteristics of coal seams in Queensland The distribution of shallow coal seams (0 - ~600m depth) in Queensland is shown in Figure 6. The major economically important coal deposits occur in the Bowen and Surat/Clarence–Morton Basins. In Queensland, the principle coal formation periods occurred during the Permian, Triassic and

CWiMI

14

Jurassic periods (Queensland Department of Natural Resources and Mines, 2003). The older Permian and Triassic age coals formed in the Bowen and Galilee Basins. A second phase of coal formation occurred in the Surat and Clarence-Morton Basins during the Jurassic period. As the Surat Basin formed at a later geological stage than the Bowen Basin, the coals seams of the southern Bowen Basin extend at depth underneath the Surat Basin (Draper and Boreham, 2006). A description of the sequence of tectonic events that occurred during coal formation in these basins can be found in Esterle and Sliwa (2002) and Draper and Boreham (2006).

Figure 6. Distribution of shallow coal bearing areas in the depositional basins in Queensland.

General features of the areas currently under CSG development in the Bowen and Surat basins are summarised below. Throughout this report reference will be made to two general geographic areas in the Bowen Basin (Figure 7A). Firstly, the Northern Bowen Basin which extends from Collinsville through Moranbah to Emerald; and secondly, the central Bowen Basin which is the area south of

CWiMI

15

Emerald to a line of latitude located approximately at Taroom. This area includes the area where the Bowen Basin underlies the northern extent of the Surat Basin.

4.2 Coal Seam Gas Production in Queensland Coal Seam Gas (CSG) tenements have been granted throughout the Bowen and Surat Basins (Figure 7). Current production is concentrated in six areas, predominantly located in the central Bowen Basin and the Surat Basin (Figure 7a, b). CSG is being produced from four different coal measures (Figure 7c) which have distinct properties due to the depositional and structural setting of the measures (Table 3). Consequently, as shown in Section 4.4, the relationship between gas and water production is extremely variable both within and between measures. The concentration of current CSG production in particular areas (Figure 7a, b) means that there is a greater amount of better quality data in these areas and also that any impacts to overlying or underlying aquifers are likely to be greatest in these areas. Table 3. Characteristics of the coal measures of Bowen and Surat Basins (Draper and Boreham 2006)

Moranbah Coal Measures Baralaba Coal Measures Bandanna Formation Walloon Coal Measures

CWiMI

Horizontal permeability (mD)

Coal Porosity (vol %)

CSG fields

Tectonic setting

Structural setting

Gas content

Moranbah Project

Foreland

Broad syncline, stress shadow

High

Moura, Peat, Scotia

Foreland

Large anticline, monocline and fracturing

High (free gas)

0.29

14

Fairview, Spring Gully

Foreland

Large anticline (Comet Ridge)

High

2.61

10

Jurassic Projects

Intracratonic

Noses and anticlines

Low

80.39

16

0.24

17

16

CWiMI

Figure 7. a) Distribution of coal seam gas tenements and wells in the Bowen and Surat Basins. The names of the major producing fields are noted. b) The 2P reserves estimated for currently active tenements are shown by the shaded colour scale. Exploration and currently non-producing tenements are shown unshaded, existing gas pipelines and towns. c) Tenements in relation to coal measures from which gas is being extracted.

17

4.3 Coal seam gas formation, retention and extraction The generation and retention of gas in a coal seam is the result of a complex interplay of processes including gas generation rate, the gas holding capacity of the coal and its permeability, the sealing nature of the overlying rocks, and the pressure of the rocks and fluids in the overburden (Draper and Boreham, 2006). Coal seam gas is a mixture of methane (CH4) and carbon dioxide (CO2) that is produced during coal maturation through either thermogenic (burial heat and pressure) or biogenic (bacterial activity) processes (Green and Randall, 2008). The majority of coal seam gas is retained in the coal as an adsorbed gas in the coal matrix, i.e. the internal surface of the coal (Thomas, 2002). This internal matrix is comprised of micropores that range in size from < 2 - 20 angstroms. The gas is held in place by the hydrostatic pressure of water that fills the larger (>500 angstrom) pore spaces called cleats (Rightmire, 1984). Gas content in coal generally increases with depth of cover and/or increasing coal rank, due both to longer times for gas generation and retention by hydrostatic pressure (Rightmire, 1984). These factors can change throughout geological time, particularly if the strata are uplifted, folded and/or faulted, relieving pressure and allowing the gas to escape. Figure 7c shows the distribution of the coal measures from which gas is currently being extracted in

Queensland. The characteristics of the coal measures vary spatially and are reflected in gas and water production. Consequently, coal seams with seemingly low gas contents, as in the Surat Basin, are proving to be economical because permeability is higher than in the Bowen Basin. Additionally, in many cases different operational techniques can be used to overcome many of the constraints. For example in-seam drilling and fraccing (fracturing coal) can be used to increase permeability of the coal and therefore increase gas production.

4.4 Gas and water production from CSG in Queensland To extract coal seam gas, the hydrostatic pressure must be reduced so that gas desorbs from the coal matrix; this is achieved by pumping water out of the cleats. A typical production profile is presented in Figure 8. During the initial depressurising stage, the rate of water production is high; over time, gas production increases and water production decreases (Witherbee et al., 1992, Harrison et al., 2000).

CWiMI

18

Figure 8. Typical gas and water production profile for a CSG well (DME, 2008; original picture courtesy of CH4 Pty Ltd Arrow Energy Limited)

Although gas and water production data from individual wells was not available for this study, an indication of the variability of gas and water production observed in Queensland CSG developments can be seen in Figure 9. This data was taken from the Digital Exploration Report System (QDEX) held by the Department of Mines and Energy. Since 2005, companies have reported aggregate 6 monthly gas and water production for each producing tenement. Each data point in Figure 9 represents the gas and water produced per well in 2007 on each tenement producing gas. Similar variation was observed for each year that data has been reported. It is evident from Figure 9 that there is enormous variability in gas and water production both within and between measures. Some of this variability is due to inherent spatial variability in the factors controlling gas and water content of the seams. For example, the Baralaba Coal Measures are producing relatively high gas volumes but almost no water production. This is because production from these fields is from a gas cap, obviating the need for dewatering. Some of the variation may also reflect fields and wells in different stages of gas production (Figure 8). In addition, there are some inconsistencies in company reporting, where some companies report the total number of wells in a field for each period but note that some wells may not be producing during the period. In this work it has been assumed that all reported wells are producing during the period. Although variable, in general, the Walloon Coal Measures and Bandanna Coal Measures produce more water than the other measures. Average water and gas production for each measure are given in Table 4. The average ratio of gas:water production for each measure was used to model predicted water production for the industry expansion scenarios presented below (Section 5).

CWiMI

19

Figure 9. Gas and water production reported in 2007 from CSG tenements in the Bowen and Surat basins. Aggregate production values are reported every 6 months by companies for each tenement. Table 4. Gas and water production and 2P (‘proved’ and ‘probable’) reserves per unit area for four coal measures from the Bowen and Surat Basins. Gas and water production values were taken from 6 monthly well production data supplied by the Queensland Department of Mines and Energy (DME). Reserves per area were based on tenement areas and DME estimates of 2P reserves for active tenements in coal measures as of December 2007.

Gas production 3 (Mm /well/year)

Water production (ML/well/year)

Water: Gas production

Bandanna

20.0

64.6

3.2

40

Baralaba

4.3

0.5

0.1

14

Walloons

1.6

36.4

22.7

45

Moranbah

2.8

10.0

3.6

21

Measure

2P Reserve/area 3 2 (Mm /km )

5 CSG Industry Development Scenarios and Implications for Water Production The second objective of this work was to provide a broad assessment of the water production and spatial distribution resulting from the expansion of the coal seam gas industry. Using three potential 20 year LNG industry development scenarios of 10 Mtpa, 28 Mtpa and 40 Mtpa production targets with concomitant growth in the existing domestic market, the following parameters were estimated: 1. The spatial extent and pattern of CSG development in the Surat and Bowen Basins including a broad description of the possible location and characteristics of potential nodes; and 2. The likely quantity and distribution of CSG water, including if possible, an estimate of the reliability of and variability in quantity and quality, which may result from CSG development in the Surat and Bowen basins.

CWiMI

20

Given the spatial and temporal variability in gas and water relationships across the Basins (Figure 9), predicting the quantities of water likely to be produced during industry expansion is difficult. To fully address this objective would require the development of a comprehensive spatial model of coal seam gas production, the concomitant water yields, aquifer hydraulics and the interactions among aquifers and coal seam dewatering, the economic drivers of gas exploration and company exploration and development strategies. For example, Harbison et al. (2008) demonstrated the utility of a coal seam gas groundwater model for a hypothetical well field in the Surat basin. However, the application of regional numerical models awaits availability of comprehensive field data. A simple ‘conceptual’ model of coal seam gas and water production was developed to estimate the possible magnitude and spatial extent of water production as coal seam gas production is increased to meet international and domestic market demands over the next decade. This generalised water and gas accounting model determined water yield from coal seam gas production by tracking the addition of new wells on current production tenements and the ‘activation’ of new tenements brought online to meet gas production targets over a 20 year time horizon. The model makes simple assumptions (refer to Table 6) about gas and water production per well, where new wells are added, the domestic and international demand for gas production and what determines the addition of wells onto tenements that are currently ‘inactive’ (i.e. tenements that have been explored but are currently not used to meet domestic gas production targets). As such this model can be used to explore scenarios for assessing possible water production rates throughout the Bowen and Surat basins but does not constitute a prescription of how the exploitation of gas reserves, the concurrent production of coal seam water and regional effects on surface aquifers will evolve through time. This approach was deemed to be the best compromise between achieving the outputs sought by the client and the availability of data with which to constrain the model.

5.1 Model Structure and modelling methodology The basic structure of the conceptual model is depicted in Figure 10. Starting with the current configuration of active wells on tenements throughout the Bowen and Surat Basins in 2007, the total annual gas production, F, was calculated as Fg(t) = where w

g

i,j

i j

wgi,j

(1) th

3

th

is the annual average gas production of the i well (Mm /well/year) on the j tenement.

Average gas and water production per well for each coal seam gas reservoir is shown in Table 4. The primary unit of calculation in the model is the tenure spatial element, or ‘tenement’, which is identified by a lease number. Hierarchically, wells are located within tenements which are located within reservoirs. Reservoirs are identified by the primary coal measures from which coal seam gas is extracted. Thus, each tenement takes on the average gas and water production characteristics of its

CWiMI

21

‘parent’ reservoir. Total water and gas production is a function of reservoir characteristics and number of wells. If total gas production in any year was less than the production target, a single new well was added to each ‘active’ tenement and the summation repeated until the production target was achieved. Where any single tenement in any year had lost more than 50% of its estimated 2007 gas reserve or had reached an average well density of more than one well per square kilometre, an ‘inactive’ tenement was ‘opened’ (brought online) and wells added to this new tenement on each iteration. No further 2

wells were added to existing tenements that had reached their 50% reserve or more than 1 well/km . Gas production proceeded only from existing wells at their average rate for the remainder of the model run on these tenements. No tenement exceeded its 2P reserves in the time horizon of the model. When the model was confined to only existing tenements it exhausted these 2P reserves within 4 – 7 years over the three LNG development scenarios. 2

The choice of which tenement became active following the 50% reserve or 1 well/km ‘trigger’ was based on the distance of that tenement to existing infrastructure. The type of infrastructure considered was current and planned gas pipelines, existing wells (active or inactive), roads and towns. Ultimately, only current and planned pipeline and existing towns were used to develop the distance function because other infrastructure was aliased with these. Figure 11 shows the average distance per tenement calculated using a Geographic Information System (GIS) from vector coverages of pipelines and township locations supplied by the Queensland Department of Mines and Energy and the Australian Survey and Land Information Group (1:2.5 million digital topographic dataset). Euclidean distances to infrastructure were calculated within the GIS for a 1 km grid laid over central Queensland and average distance per tenement was then determined from all grid cells within each tenement boundary. Thus, smaller tenements closer to the existing LNG pipeline and townships were preferentially selected. The assumption underlying the cost function is that the next tenement is selected on the basis of its proximity to existing infrastructure; specifically existing and proposed pipelines and townships. Once the target gas production in any year was achieved, annual water production was calculated as: Fw(t) =

i j

wwi,j

(2)

w

where w is the annual average water production per well. Annual gas production per tenement, Fg(t), was deducted from the tenement’s reserves. Tenement reserves for the start year 2007 were estimated as the ‘proved’ and ‘probable’ (2P) reserves for that year as reported for ‘active’ tenements in data supplied by the Queensland Department of Mines and Energy and Department of Infrastructure and Planning. A ‘reserve capacity’ for each measure was calculated based on the 2P reserves per unit area for all active tenements averaged across each measure (Table 4). On inactive tenements, where no 2P estimate was available, the total reserve was calculated as the reserve

CWiMI

22

capacity multiplied by tenement area. An attempt to relate 2P 2007 reserves to tenement area and coal measure thickness by regressions methods showed no relationship and so wasn’t pursued further.

Figure 10. Schematic of the coal seam gas and water production model for the Bowen and Surat basins in central Queensland.

Figure 11. The ‘cost function’ per tenement used to determine the order in which tenements come online to meet gas production scenarios. The cost function is calculated as the average distance per tenement from existing and planned LNG pipeline infrastructure and existing townships. Order of tenement selection proceeded form light green towards dark green shades.

CWiMI

23

The three target gas production scenarios, provided by the Queensland Department of Infrastructure and Planning, were based on the stated development goals of proponent companies that amount to six LNG production plants over the next 20 years generating 550, 1885 and 2262 PJ equivalents of LNG in 2019. The production scenarios in energy units were converted by the Department to annual targets by volume (Mm3/year) and these values were supplied in a spreadsheet to the project. The gas production targets are shown in Figure 12.

Figure 12. Coal seam gas production scenarios for period 2008 – 2020 based on possible production figures of 10, 28 and 40.8 Mt per annum provided by the Queensland Department of Infrastructure and Planning.

For each production scenario, three model runs were generated to account for uncertainty in model parameters. Values of well gas and water production from Table 4 were adjusted by +/- 25% to examine sensitivity of total annual water production and its spatial distribution to uncertainty in gas production rates per well.

5.2 Results and Discussion Results from the modelling are shown as water production trajectories in Figure 13 and as output statistics in Table 5. The ‘baseline’ model run generated water output from tenements considering only the domestic growth in gas production of approximately 0.51 Mtpa (yellow curve in Figure 13). The model predicted for this ‘baseline’ target in 2020, a water production rate of 63GL/year, requiring 2 new tenements and a total of 1725 wells distributed 63% in the Bowen and 37% in the Surat Basins. Note that the initial water production from the model in Figure 12 is constant because the baseline number of wells was deemed sufficient by the model to meet demand for the first few years of demand. Table 5 and Figure 13 summarise model output for all model runs that were done to meet the higher gas production targets of 10, 28 and 40.8 Mtpa allowing for comparison with baseline model output.

CWiMI

24

Figure 13. Trajectories of model output for coal seam gas water production (ML/year). The plot depicts three gas production target scenarios and the domestic production target for the Bowen and Surat basins (solid lines). Dotted black, red and blue lines show sensitivity of water production based on +/-25% uncertainty in average gas production rates per well. Vertical bars indicate the range in sensitivity of water production estimates for 2020.

CWiMI

25

CWiMI

Well distribution by basin

Distribution of water production (%)

34.9 62.6 2.3 0.2 60 40

Walloons

Bandanna

Moranbah

Baralaba

Bowen (%)

Surat (%)

4069

10

New tenements (number)

Total wells (number)

166

Water production (GL/year)

g

-w

Uncertainty

Gas production (Mtpa) g

38

62

0.2

2.8

62.4

34.6

3143

4

126

w

10 g

37

63

0.2

3.4

61.2

35.3

2617

4

102

+w

g

42

58

0.1

1.8

61.5

36.5

6297

18

264

-w

g

40

60

0.2

2.1

62.5

35.2

4698

12

196

w

28 g

39

61

0.2

2.5

62.1

35.2

3843

10

156

+w

g

52

48

0.1

2.9

48.7

48.3

11100

38

419

-w

g

42

58

0.1

1.9

61.7

36.3

6526

21

281

w

40.8 g

42

58

0.1

2.0

61.4

36.4

5282

16

227

+w

Table 5. Water production rates in 2020 for three gas production target scenarios (10, 28, 40.8 Mtpa). Also shown is model output predicting the number of new tenements added to meet production targets, the predicted total number of wells on tenements, and the proportion of these wells distributed between the Bowen and Surat basins. g Uncertainty values are given as ‘+’ or ‘-‘ values indicating a 25% variation on the nominal average gas production rate per reservoir (w ) from Table 4.

26

Water production increased from 126 to 281 GL/yr in 2020 as total gas production increased from 10 to 40.8 Mtpa representing a 2.0 – 4.5 fold increase in water production over the baseline value. The target gas production rates determined the rate at which new wells were added to tenements and it was found that new tenements were opened primarily in response to well densities exceeding 1 2

well/km rather than depletion of gas reserves. Uncertainty in the gas production rate per well had a countervailing influence on estimated water production. Figure 13 shows that if true gas production per well is 25% lower than the nominal values, water production was increased by 32%, 35%, and 49% for the 10, 28, and 40.8 Mtpa targets. Conversely, if true gas production per well is 25% higher than the nominal values, water production was 19%, 20%, and 19% lower, respectively. Hence, true water production will be higher than estimated here only if new data shows that average well gas production rates per measure are actually lower than those assumed in Figure 13. The asymmetric distribution of errors (dotted lines) around the expected values (Figure 13) is due to the way tenements were added to meet gas production targets. With lower gas production rates per well, more wells are required to meet a target and so new tenements need to be opened to meet these targets. The model cost function forces these new tenements to be preferentially opened in the Walloon and Bandanna coal measures which have relatively high water yields per unit gas production than the other coal measures (Figure 13). The spatial distribution of water production in 2020 is shown in Figure 14 for the three gas production scenarios (rows down Figure). The effect of uncertainty in the gas production rate per well on water yield is also presented (columns across Figure). The increase in gas production, opening of new tenements and concomitant increase in water yield occurs preferentially throughout the Walloon and Bandanna coal measures towards the south and southeast of the Bowen-Surat basins. The relatively dense network of LNG pipelines in this area (Figure 2b) is largely responsible for the rapid exploitation of these tenements by the model to meet projected gas production targets via the cost function (Figure 11). The relatively high water production rates of wells in the Walloon and Bandanna coal measures resulted in higher water yields per unit gas production than would occur if these wells were preferentially sunk in the Moranbah or Baralaba measures. Exploitation of reserves in the northern part of the basin would be promoted in the model if penetration of the LNG pipeline network into the Bowen tenements was more extensive. Because new wells and tenements were preferentially added to measures in the Surat Basin, the variation in water production to meet gas production targets was a o

2

linear function of well number (Water production [GL/yr] = 0.0378 x N . wells – 15.354; r = 0.985). As a result, the proportion of wells located in the Bowen relative to Surat Basin was also decreased (Table 5).

CWiMI

27

Figure 14. Spatial estimates of water production rates (ML/year) in 2020 for tenements in the Bowen and Surat basins of central Queensland. Each row depicts modelled water production for the 10 (a, b, c), 28 (d, e, f), and 40.8 (g, h, i) Mtpa gas production targets. Columns shows the effect of uncertainty in gas production rates per g well (w ) for -25% error (a, d, g,), nominal values (b, e, h), and +25% (c, f, i) error in gas production rate.

CWiMI

28

5.3 Assumptions and uncertainties Table 6 summarises the assumptions and associated uncertainties made in the spatial analysis of gas and water production Table 6. Summary of assumptions and uncertainties used in spatial analysis of gas and water production. Assumptions

Associated Uncertainty/Implications

Well production rates for each measure were set at an average rate based on available well data. The model did not account for variation in gas and water production over time.

This assumption is only reasonable if the well logs contain time dependent variation in well production, and if there is limited spatial variation in well production across a coal measure.

Gas and water production does not decline to zero due to depletion of all water reserves within a tenement.

This simplifying assumption is unlikely to be accurate; e.g. modelling predictions by Harbison et al. (2008) 2 show that 72 wells pumping within a 225 km area of the Surat basin would extract all water in the vicinity of those wells within 11 years.

New tenements were opened to gas production when either: - Average well density in a tenement exceeded 1km .

These assumptions do not consider true economic factors nor reflect individual company strategies towards CSG development. For example, the model does not consider rapid expansion of wells on tenements owned by any single company.

New tenements were added according to a cost function that considered the average distance per tenement from existing and planned LNG pipeline infrastructure and existing towns

As above, this assumption neglects the economic drivers of gas exploration and company strategies that would drive the actual spatial pattern of CSG development.

New wells were added to a tenement in order to meet the production target for each year

This assumption was made such that the model met projected CSG estimates; it is yet to be determined if recoverable reserves can support this level of development.

2P Reserves extrapolated from estimates for existing active tenements in 2007 based on coal measure and tenement area.

Highly uncertain estimates of gas reserve but used only because no other means of calculating reserve was available. May impact on water yield estimates by model if current reserves are grossly underestimated.

- Gas reserves in a tenement drop below 50%; or 2

The limited data available on individual well production rates and their spatial variability is likely a major source of uncertainty in this analysis and so conclusions based on model output should be interpreted with this level of uncertainty in mind. There is also no data available to determine the amount of water produced during exploration. As shown in Figure 8 this may be a significant volume of water. Additionally, major uncertainties reside with the estimation of 2P reserves on tenements. Uncertainties in well production rates influence the rate at which the 2P reserves were depleted of gas in the model and this can potentially influence the timing of when new tenements were brought online particularly if reserves are relatively small (or overestimated) and depleted quickly. These uncertainties impact on the number of tenements exploited for new production and the spatial distribution of these active tenements at the end of a model run.

CWiMI

29

As noted above, well production rates were set at an average rate based on well data. Hence, these rates did not vary throughout model runs nor did they take into account variation in well gas and water production over time. This approach assumes that averages calculated from well logs implicitly contain time dependent variation in well production. While any single well may have production rates above or below the assigned value in Table 4 when averaged over all wells on a tenement at any time these production rates should converge to the average value if it is unbiased. If the values in Table 4 are actually representative of wells across a coal measure then the model will yield reasonable results. Clearly, there will be some spatial variation in well production and this will influence rates of water production; however, the level of this variability could not be ascertained with the data at hand. In particular, variation in aquifer potentiometric surface throughout measures is unknown and if high well densities per tenement impart significant effects on aquifer hydraulics then model results here will 2

depart from reality. Harbison et al (2008) note that 72 wells pumping within a 225 km area of the Surat basin would extract all water in the vicinity of those wells within 11 years. Given that the model runs here were 12 years duration, we made the simplifying assumption that well gas and water production did not decline to zero due to depletion of all water reserves in a tenement. We were able to deplete gas reserves only by forcing the model to meet gas production targets from existing active 3

tenements (amounting to a total 2P estimate in 2007 of 188,241 Mm ). Depletion occurred in 5, 6 and 7 years for 40.8, 28, and 10 Mtpa gas targets. Note, however, that these estimates do not take into account effects of extraction on flow and aquifer hydraulics. They simply amount to a mass balance calculation. Errors in well water production rates per unit time will impact directly on the estimates of total water production in this study because of the direct linear relationship between rate and total yield. There is also considerable uncertainty in the ‘trigger’ that opens new tenements to gas production based on the extent of exploitation of existing gas reserves. The current ‘trigger’ mechanism in the model attempts to capture some measure of both the extent to which a resource has been exhausted (through estimation of gas reserve and density of wells on a tenement) and the likelihood of tenement -2

activation based on proximity to infrastructure. A threshold well density of less than 1 km does not change substantially the results of model runs in terms of total water production. However, a lower well density will open more tenements (particularly small tenements in the vicinity of the LNG pipeline network) and these will have a more even distribution of wells per tenement.

Figure 15 shows the effect of uncertainty in 2P gas reserves on the number of active tenements needed to meet gas production targets of 40.8, 28 and 10 Mtpa. If existing tenements continued to have wells added until estimated 2P gas reserves are between 45% and 100% exhausted there is no change in the total number of tenements needed to meet gas production targets or their estimated water yields. This is because the activation of new tenements is dependent only on the number of wells per unit area. However, where the threshold for activation of new tenements falls below 40% of 2P reserve, the total number of active tenement increases linearly with the threshold value at which a new tenement is activated (Figure 14).

CWiMI

30

Figure 15. Effect of uncertainty in 2P gas reserves on number of active tenements required to meet gas production targets of 10, 28, and 40.8 Mtpa. X-axis shows the proportional depletion point of any tenement’s 2P reserves beyond which no new wells are added to that tenement and a new tenement is brought into production. Y-axis shows the number of active tenements at end of the model run.

Thus, if 2P gas reserves are currently greatly overestimated (and hence the trigger to open new tenements is considerably lower than the 50% assumption used in the standard model runs in this study) more tenements will need to become active to meet the specified gas production targets. The net result is to increase water production above the values estimated in the standard model runs (Figure 12; Figure 13; Table 5) because new tenements are preferentially opened in the higher water yielding Surat Basin (Figure 14). With an increase in gas production target (from 10 to 40.8 Mtpa) the model activates new tenements preferentially along existing infrastructure. Extensive activation of tenements in the Moranbah measures occurs only in the 40.8 Mtpa scenario (Figure 16c). Total water yields are not greatly increased over the current model predictions (and are within the uncertainty bands presented above; Table 3) because these new tenements have the same gas and water yields per well. Only the spatial extent of production tenements is more extensive in Figure 16 compared with Figure 14.

CWiMI

31

Figure 16. Effect of uncertainty in 2P gas reserve on spatial distribution of water production by tenement (ML/yr) for gas production targets of (a) 10, (b) 28, and (c) 40.8 Mtpa. In these model runs, the trigger for activation of a new tenement was the depletion of existing 2P gas reserves on any tenement by 25% (rather than 50% used in the standard model run; Figure 14).

In reality, economic factors coupled with different company’s risk management strategies, stratigraphic characteristics of the coal measures, their gas production rates, and the location of existing infrastructure all bear on the probability that a tenement will be brought into production at any point in time. Currently, no differentiation among the six lease-holding gas companies is included in the model. For example, the model does not include rapid expansion of wells on tenements owned by any single company should they decide to do this. The actual response by companies to economic, geologic and hydrologic risk factors when establishing wells on tenements will alter the pattern of exploitation of tenements seen here. Such a different spatial pattern of exploitation of reserves would yield a different pattern of water production if it led to substantially different coal measures being chosen over what the model has selected. In particular, where the Moranbah measure became exploited preferentially over the more southern coal measures, a lower water production rate might occur than is predicted by the model.

6 Assessment of Aquifer Risks Much of the area currently under development for CSG lies within the Great Artesian Basin (GAB) water resource management area (Figure 17). The areal extent of two of the GAB aquifers (Hutton and Precipice) is shown relative to the extent of the coal seams under development for CSG in Appendix A. The aquifers of the GAB are the pre-eminent groundwater supply area for Queensland supporting agricultural production and human consumption. The Walloon Coal Measures are also a GAB aquifer. In addition to the GAB aquifers, the Condamine alluvial aquifer system, also overlies the coal seams in the Surat Basin (Figure 18; Appendix A). This alluvial aquifer is the highest allocated groundwater supply in the state.

CWiMI

32

The analysis presented above shows that by 2020 between 63 GL/yr and 275 GL/yr may be dewatered from the coal seams of the basins. Current estimates of water use for stock and domestic from the GAB aquifers in the Surat Basin is ~74 GL/yr (Queensland Department of Natural Resources and Mines, 2005). If dewatering the coal seams does compromise connectivity with GAB aquifers, the loss of water from the aquifers to the coal seams could potentially be of the same order as current groundwater use. It is therefore important to establish a profile of the aquifers potentially at risk and the factors that may contribute to that risk. In this work analysis was restricted to the significant aquifers that were in closest proximity to the coal measures used for CSG extraction. The significance of the aquifers was taken as the ranking of the GAB aquifers provided by NRW or aquifers where there is public concern over CSG activities (e.g. Condamine alluvial system).

Figure 17. Location of GAB management areas showing recharge springs in relation to CSG tenements.

CWiMI

33

Figure 18. Typical stratigraphic sequences observed in the areas of CSG production.

Typical stratigraphic profiles showing the relationship between the coal measures and aquifers of closest vertical distance for the main areas of CSG production are shown in Figure 18. Throughout

CWiMI

34

the Surat basin the Walloon Coal Measures lie between the Hutton aquifer and the Springbok aquifer. In the eastern part of the Surat Basin, the Condamine alluvium aquifer, which is not part of the GAB but is the highest allocated groundwater source in the state, can either directly overlie the Walloon Coal Measures or the Springbok sandstone (refer to Figure 18). The Walloon Coal Measures are also used as a groundwater source in some areas, however, in general this occurs in areas where the coal measures are too shallow to provide a significant gas supply and hence these water users are not likely to be at risk (Parsons Brinckerhoff, 2004). Groundwater use is less developed in the northern Bowen Basin. In general the tertiary basalts may contain useful supplies in some areas. Around Moranbah the tertiary basalts may directly overlie the coal measures. The Isaac River alluvium which also overlies the coal measures, particularly in the vicinity of the Moranbah gas fields, is also used as a groundwater source and additionally can provide a conduit for loss of river flow if connectivity is established with the coal measures.

6.1 Determining risks to groundwater aquifers resulting from CSG extraction The connectivity of coal seam aquifers with overlying and underlying groundwater aquifers is not well known in Queensland. Indeed, even in the United States, where commercial CSG production has been occurring since the 1980s (Rightmire, 1984), there is a lack of scientific information regarding the impacts of CSG extraction on hydraulically connected aquifers (CBM/NGC Multi-Stakeholder Advisory Committee, 2006). Community concerns have been raised in the Surat basin where there is anecdotal evidence that withdrawal of water from CSG wells has impacted local groundwater bores. The water table in bores located at the margins of the groundwater aquifers in the area have certainly decreased in recent years (NRW water level data). However, given that the area has been in severe drought for the last 10 years it is difficult to directly attribute the water table drawdown to CSG development. At present no studies have been undertaken to directly assess hydraulic connectivity between overlying or underlying aquifers and the coal seams at either a local or regional scale. However, in a few cases evidence of connectivity has been suggested from water quality analysis. For example the wide range of total dissolved solids and ionic composition of waters from the Fairview and Spring Gully CSG fields in the Bowen Basin has been interpreted to reflect the influx of water to the coal seam from shallower depths (Draper and Boreham, 2006). It is not clear when this influx occurred, but the data suggests that there may currently be, and likely has been, incomplete aquifer isolation in this area. In contrast, variation in the quantity and quality of produced water from the Moranbah field in the northern Bowen Basin is interpreted to reflect a lack of connectivity between coal seams at the local scale due to splitting and loss of permeability from increased mineral matter deposited in cleats that is compounded by faulting (Kinnon et al., 2008). However, there is evidence that suggests that the coal seam water may be recharged from surface or other groundwater aquifer water sources in some

CWiMI

35

areas of the field (Kinnon et al., 2008). Both these fields have been noted to be highly fractured which may lead to enhanced connectivity. With the exception of the Condamine and Isaac River alluvium and Tertiary Basalts the aquifers noted in Figure 18 are confined aquifers and are therefore bounded by relatively impermeable layers. In the central Bowen Basin, these confining layers are the Rewan formation and Moolayember formation. In the Surat Basin the Walloon Coal Measures form the confining layer between the Hutton and Springbok aquifers. Under normal circumstances the less permeable confining layer would impede water flow between different strata. Consequently, it is generally thought that aquifers are not in hydrological connection to the coal seam aquifers. For this reason it has previously been expected that there would only be a low impact of CSG development on surrounding aquifers in the Surat basin (Parsons Brinckerhoff, 2004). However, because dewatering of the coal seams might disrupt the hydraulic conditions that maintain the dynamic flow equilibrium in aquifers, connectivity may be established posing a risk to the aquifer water supply. Coal seam gas extraction may alter the connectivity between coal seams and aquifers through the following means: 1. Changes in hydraulic conditions that control water movement within and between aquifers = Hydrological risk ; and /or 2. Permanent physical changes to the strata containing the aquifers = Physical risk. The hydrological risks expected in the context of the Surat and Bowen Basin gas fields are discussed further below. A preliminary formulation for determining the relative importance of these risks and the contributing factors is presented. The causes of the physical risks include operational methods such as where coal is fractured deliberately to improve the extraction efficiency of gas. It is not known whether this fracturing may extend to overlying strata, however if this occurs fractures can provide a direct pathway for water to move from the aquifer to the coal seam. Similarly, release of hydrostatic and gas pressure from the coal may physically alter the coal pore structure through compaction of the seam. These changes may have implications for re-injection as a disposal method for extracted water as it is likely that the coal does not refill with water in the same way that it dewatered (hysteresis effects). There is currently not enough basic research conducted to determine if hysteresis is likely to occur.

6.2 Hydrological risks Aquifers exist in a state of dynamic equilibrium established over long time periods. Flow within an aquifer is governed by rate of recharge, discharge, the hydraulic gradient (or head) and physical properties such as permeability and porosity of the strata. As noted above, flow between aquifers is typically considered to not occur due to the presence of relatively impermeable (confining) layers between aquifers. However, vertical leakage from GAB aquifers through the confining beds could

CWiMI

36

occur where the confining beds are relatively thin, pressures are high, and/or in marginal areas of the Basins (Woods et al., 1990). A steady state groundwater flow model (Welsh, 2000) estimated the net leakage loss from GAB aquifers across the confining layers to be approximately 440 ML/d. In the context of the Bowen and Surat basins, three potential cases were identified in which changes to hydrological conditions in the coal seam may present a risk to aquifers. These cases and the hypothesised risks to the aquifers are summarised in Figure 19.

Figure 19. Summary of the hypothesised risks to aquifers in the Bowen and Surat Basins

Case A: One aquifer overlying the coal is exemplified by the Springbok aquifer overlying the Walloon Coal Measures in the Surat basin and the Clematis or Precipice aquifers overlying the Bandanna Coal Measures in the Bowen Basin (Figure 18). Case B: is exemplified by the Hutton aquifer underlying the Walloon Coal measures in the Surat Basin and an example of Case C is shown by the Condamine alluvium overlying the Springbok aquifer and Walloon Coal Measures (Figure 18). During CSG extraction, hydrological connectivity of the coal seam with either overlying or underlying aquifers may be impacted due to: 1. Depressurisation from initial dewatering causing a lowering of the head in the coal seam relative to the head in the overlying aquifer thereby creating a hydraulic gradient across the aquitard (termed Head Risk, Rh); and/or 2. Local changes in capillarity where the smaller empty pores of the coal will exert a capillary pull on water held in larger pores of both the aquitard and the sandstone (termed Capillarity Risk, Rc). The above risks may be compounded in areas where there is fracturing or faulting that can provide fast connective flow paths. Increased pressure regimes present in folded areas may also force water movement from aquifers into the coal seam.

CWiMI

37

Figure 20. Schematic of hypothesised risks to aquifers from dewatering during CSG extraction. The coal seam is shown as black, aquifer as grey and aquitard as orange. Other strata such as alluvial clay are displayed in shades of brown. Blue dots represent water contained within the coal seam (circles) and aquifer (squares); white dots denote dewatered sections of the strata. Water flow resulting from both decreased head and capillary pull from the empty coal pores is shown by the red arrows. The top panel (A, B and C) demonstrates how a change in hydraulic head within a coal seam may result in water movement from an overlying aquifer. Figure A depicts the landscape prior to CSG extraction, whereby the coal and aquifer share a similar hydraulic head and are in tension equilibrium. During dewatering, the head within the coal seam reduces (B) resulting in a head difference between the aquifer and coal. This could cause water movement from the aquifer, potentially through an aquitard, into the coal seam (C). The lower panel (D and E) illustrate that water from an overlying aquifer may also be pulled into a coal seam due to capillary forces.

The expected changes to water flow between the aquifers and the coal seam are schematically represented in Figure 20. Both the decrease in head and the increase in capillarity exerted by the small coal pores may cause water to be drawn through the aquitard from the overlying or underlying aquifer into the coal seam. With continuing gas and water extraction from a coal seam in an area, is it also possible that the effect of decreasing head and increasing capillary pull due to dewatering may reduce the head in the overlying aquifer sufficiently that this effect may cascade through to other aquifers (Figure 21). In this case it should also be recognised that water quality of the productive aquifer may be compromised if the secondary connected aquifers have lower water quality or if water quality changes as a result of interaction with intervening strata.

CWiMI

38

Figure 21. Hypothesised risk to other overlying aquifers

6.3 Hydrological Risk Formulation for Bowen and Surat Basins As outlined above the magnitude of the hydrological risk to an aquifer at any location will depend primarily on the magnitude of the change in head, due to dewatering of the coal seam, and capillarity change due to both water and gas removal from the coal pores. The risk, however, will be moderated by the properties of the aquifers, the interbedding strata and coal seams, the geological setting of the formations and the proximity of the aquifer to the coal seam. The properties used in the formulation of risk in this study were: 1. The proximity of the coal seam and aquifers (i.e. vertical distance, d, between coal and aquifers, where increasing distance or thickness of intervening layers will decrease the risk of impact on a particular aquifer) 2. The amount of water available (aquifer storage = porosity ( ) x thickness (T)) 3. The speed with which water can move (i.e. permeability (K)) 4. Amount of fracturing and folding in the region (F) The normalised hydrological risk to an aquifer, Ra, was computed using the following equation: 2

Ra = [(ε x T)c x Kc x C ] x [(ε x T)a x Ka x (1/d) ] x h x F

(3)

Where the subscript ‘c’ refers to the coal seam property and subscript ‘a’ refers to the aquifer property; C is a capillarity factor (assumed = 1 as no data is available to define this variable); h is the head of the aquifer. Properties of the interbedding strata were unavailable. The amount of fracturing or folding was not considered in this study due to lack of sufficient detailed information. It should also be noted that fracturing may decrease connectivity through realignment of strata and re-mineralisation within fractures. Thus simply estimating the number of fractures or degree of fracturing in an area cannot be reliably used to estimate the impact of this factor.

CWiMI

39

The aquifer risk was computed at each location that core stratigraphy was available from DME or NRW databases. For each core that contained the aquifer of interest and where the coal was within 200 – 600m of the surface, the thickness of the coal measure, thickness of the aquifer(s) of interest and distance between the coal measure and aquifer were calculated from stratigraphic data. Average porosity and permeability were assigned for individual strata. The aquifer head at each location was estimated as the difference between the depth to the top of the aquifer at the core location and the minimum depth in the recharge area. The average value of each of the parameters used to compute the risk is summarised in Table 7. Each of these parameters was then normalised over the entire data set. The normalised hydrological risk to each aquifer was computed using Equation 3. The normalised hydrological risk to each aquifer is presented in Figure 22. The risks are presented as: a.

Risks to overlying aquifers that includes the Springbok and Condamine overlying the Walloon Coal Measures and the Precipice and Clematis aquifers overlying the Bandanna Coal Measures; and

b. Risks to aquifers underlying coal measures (i.e. the Hutton aquifer underlying the Walloon Coal Measures). It can be seen for these figures that the highest risks are generally to the Hutton, Springbok and Condamine Alluvium aquifers from dewatering of the Walloon Coal Measures in the Surat Basin. This is predominantly due to the relatively close vertical distance between these aquifers and the coal measures compared to the other aquifer/ coal measure relationships (Table 7, Appendix B, Figure 3). The risk to the Condamine aquifer is also relatively higher in areas where the Springbok Sandstone is thin or absent. In general, risks to aquifers in the Surat Basin are higher compared to other aquifer /coal associations located outside of the Surat Basin. With the exception of the area around Moranbah, most of the Bowen Basin could not be assessed due to lack of stratigraphic and aquifer information. The risks to aquifers around Moranbah were assessed as medium for areas where tertiary basalt aquifers directly overlie the coal measures. Risks to aquifers around Moura were not assessed due to insufficient information for this area. Although the distance between the aquifer and the coal measure was a key factor in determining the relative risk, Comparison of the relative risks between the Moranbah coal measures and the tertiary basalts and the Springbok/ Walloon risks shows that distance is not the only factor influencing the risk assessment since both these aquifers directly overly the coal measures The amount of fracturing or folding was not considered in this study due to lack of sufficient detailed information. It should also be noted that fracturing may decrease connectivity through realignment of strata and re-mineralisation within fractures. Thus simply estimating the number of fractures or degree of fracturing in an area cannot be reliably used to estimate the impact of this factor.

CWiMI

40

CWiMI

Figure 22. Relative risk to aquifers a) overlying coal seams and b) underlying coal seam in the Bowen and Surat Basins.

41

Table 7. Average aquifer properties, coal measure thickness and distance between aquifers and coal measures.

Coal measure

Walloon

Moranbah Bandanna

Coal measure thickness in vicinity of aquifer (m)

Aquifer Thickness (m)

Springbok

318

Hutton

Head (m)

Distance between coal and aquifer (m)

Aquifer Porosity (vol%)

Aquifer Permeability (mD)

57

617

0

22

84

290

199

910

28

22

427

Condamine

273

34

33

78

30

700

Tertiary Basalt

60

40

590

0

12

100

Precipice

110

81

418

365

16

346

Clematis

89

115

340

179

19

446

Aquifer

6.4 Risk assessment assumptions and data gaps The specified output of the risk analysis was a set of maps that spatially displayed areas of relative high and low risk to underlying and overlying aquifers that could result from CSG extraction. In order to meaningfully display the results as maps, spatially explicit data is required for each parameter outlined in the risk formulation presented above. Data was acquired from DME and NRW from bore and well drilling results. After reviewing available data it was apparent that appropriate data was not available to produce maps of contoured risk surfaces. This was because either the spatial coverage of data for any given parameter was not available or for some strata or regions very little basic data was available for either coal aquifers or groundwater aquifers. This was particularly a problem for assessing the northern Bowen Basin, as this area is outside the GAB and groundwater supply is not considered a major water source for the region. Consequently, relatively little information has been gathered on the aquifers in the area. Based on data availability, it was decided that the risks should be calculated as “point” risks based on individual core data. The results of the risk assessment presented above should be viewed as a preliminary assessment of the potential risks. It should also be emphasised that the risk rating is a relative measure and should be used only to guide further investigation and monitoring. Table 8 summarises the assumptions and uncertainties in the aquifer risk assessments. Most of the assumptions and uncertainties were due to limitations in the data available for the assessments. These knowledge gaps should be addressed either through acquisition of better quality (either greater spatial coverage or detailed stratigraphy) data and on-going monitoring in future assessments.

CWiMI

42

Table 8. Summary of assumptions and uncertainties used in risk aquifer assessment. Assumptions Average properties of permeability, connectivity and porosity apply across each coal measure. Permeability is also assumed to be isotropic.

Associated Uncertainty/Implications Uncertainty over the movement of water between interburden and coal seams. The likelihood of different lateral and vertical flows, their spatial arrangement and change of hydrological conditions needs to be understood and quantified. The implications range from faster water movement than the risk assessment implies to potentially no water movement.

Average porosity and permeability values were used for each coal measure and aquifer.

Calculated risks do not demonstrate variation in risk within a coal measure due to differences in porosity and permeability.

Porosity and permeability for the Moranbah coal measures were assumed to be the same as for the overlying Rangal coal measures.

No other data was available.

Porosity and permeability was assumed for the Condamine Alluvium and Tertiary Basalt based on book values for similar strata (Fetter, 2001).

No other data was available.

The entire coal measure will be dewatered during CSG extraction.

This assumption was necessary because stratigraphic data for the cores only reported depths to the top and bottom of each coal measure, not individual seams. The change in hydraulic pressure (head) caused by dewatering the coal seam needs to be quantified. There is also a need to determine the lateral extent of coal seam dewatering.

Porosity and permeability of the interburden and coal seams is similar.

The association of aquifers and coal facies within measures is largely unknown. More detailed stratigraphy is required to accurately determine the distance and type of strata that lie between aquifers and dewatered coal seams. Depending on the properties of such strata, the time scales required for movement of water between aquifers and coal seams may be very slow and risks could be significantly less than what is hypothesised in this report.

The confined aquifers are full of water.

This assumption is reasonably robust for most areas but would clearly not be the case in recharge areas.

The thickness of the strata represents the thickness of the aquifer. Fracturing and folding was not considered.

There was insufficient data to estimate the impact of fracturing and folding. Further, the impacts of fracturing depend on localized conditions; it may in fact decrease connectivity through realignment of strata and re-mineralisation within fractures.

CWiMI

43

Key assumptions and data gaps in the analysis are: Average porosity and permeability values were used for each coal measure and aquifer. There were comparatively few permeability data available from which to compute the average permeability for either the coal measures or the aquifers. The porosity and permeability of the Moranbah Coal Measures is assumed to be the same as the overlying Rangal Coal Measures. Values for the Tertiary Basalt and Condamine Alluvium were assumed to be commonly used average values for these strata (Fetter, 2001). It was assumed that the entire coal measure was dewatered and that the porosity and permeability of the interburden and coal seams are not significantly different. This assumption was necessary because most of the stratigraphic data for the cores only reported depths to the top and bottom of each coal measure, not individual seams within the measure. While the distribution of coal seams and cumulative coal seam thickness has been well quantified for the Bowen Basin Moranbah/German Creek Measures (Esterle and Sliwa, 2002), only a few reported values exist for average seam thickness for the other coal measures. These items represent significant data gaps that should be constrained in any future analysis with more detailed data on the physical properties that affect water flow of interbedding strata at least for a few locations within each measure. Association of aquifers and coal facies within the measures is largely unknown. More detailed stratigraphy is required to accurately determine the distance and type of strata between dewatered coal seams and aquifers. Experimental work needs to be undertaken to determine the flow properties of dewatered and wet coal. The lateral extent of coal seam dewatering is not known. The change in hydraulic pressure (head) caused by dewatering the coal seam needs to be quantified. Capillarity effect was assumed to be 1 in this analysis and needs to be quantified with laboratory measurements. Some work is being undertaken at the University of Queensland in relation to this factor. It was assumed that the aquifers were full and that the thickness of the strata represented the thickness of the aquifer. This assumption is reasonably robust for most areas, but would clearly not be the case in recharge areas.

CWiMI

44

Fracturing and folding was not considered in the formulation as we did not have sufficient data to estimate the impact of fracturing and folding in this work. In addition, fracturing may decrease connectivity through realignment of strata and re-mineralisation within fractures. Thus simply estimating the number or degree of fracturing in an area cannot be reliably used to estimate the impact of this factor.

6.5 Summary of Aquifer Risks In summary, there are risks associated with aquifer and coal seam hydraulic connectivity when long term hydrological equilibrium situations are altered such as during dewatering of coal seams to extract coal seam gas. As shown in Figure 22, the risks are spatially heterogeneous. Broadly, aquifers that immediately overlie or underlie the Walloon Coal Measures in the Surat Basin are potentially more at risk from CSG development than other aquifers. More detailed investigation should be conducted to determine the actual level of impact. Given that this analysis assigns a relative risk it might turn out that the actual impact may be low for much of the area and hence while the risk rating may be high the impact may be low. It is highly likely that the companies currently hold more detailed hydrogeological information identified as key knowledge gaps in the analysis presented above. Inclusion of this data in the risk analysis would better constrain the risks to aquifer connectivity. A monitoring strategy, as outlined below, should be implemented to provide baseline information that can be used to help identify and monitor future impacts. Filling the key knowledge gaps and implementation of a monitoring program will ensure that robust policy is formulated so that industry, community and government are confident that risks are being managed.

7 Monitoring to manage risks As presented above dewatering of coal seams for CSG extraction could potentially present risks to overlying and underlying aquifers by changing the hydraulic conditions controlling connectivity. In order to manage these potential risks and sustainably manage the potential water resource derived from CSG production additional work is required in three areas. These are: 1. Fill key knowledge gaps; 2. Monitoring; and 3. Cumulative impacts assessment. It is highly likely that the CSG companies hold some of the information that can be used to fill some of the key knowledge gaps and uncertainties in the analyses presented in this paper. In particular water production during exploration and detailed stratigraphy and porosity/permeability and fracturing are both critical elements required to enable gas production for the companies. This information would be important in providing a better assessment of potential risks to aquifers in areas close to the gas

CWiMI

45

fields. Engagement with CSG companies should be considered as a crucial early step to enable further analysis and design the monitoring program. Monitoring should be undertaken in a phased approach using an adaptive management framework. In this way new findings from a number of different lines of evidence can be incorporated into resource planning decisions as the information becomes available. Under an adaptive management strategy decisions are made in a stepwise manner taking into account uncertainty when there is not comprehensive information available. Formal adaptive management uses rigorous methods to establish a hypothesis of how a system will behave under a given decision (or decisions). Once the decision(s) are taken the system is then monitored to observe whether the hypothesis is met or not. Over time, as more information becomes available, via monitoring, alternative hypotheses and decisions are made. The process is then repeated. Therefore, adaptive management is a process of guiding ongoing system evolution using step-by-step decision making as opposed to deterministic rigid planning. Any monitoring program designed to evaluate impacts on aquifers from CSG development should include conventional well/bore monitoring and hydrological modelling combined with new approaches and technologies including geochemical and isotopic tracing and dating techniques and remote sensing. The modelling and field monitoring should be undertaken at both regional scale and local (gas field) scales. A well structured observation and modelling program for the area under CSG development along with metered abstractions by the irrigated agriculture sector would be able to identify regional trends in groundwater resources through time. This work would require a coordinated program involving industry, government departments (NRW, DME and DIP) as well as researchers/consultants. Engagement with the CSG companies is critical to ascertain whether more detailed hydrological and geological information or models are currently available to reduce the uncertainty in the aquifer risk assessments at key sites and gain a better understanding of the dynamics of water production for the CSG fields. Such a comprehensive approach is the only means by which the significant uncertainties in current understanding of the groundwater system(s) can be reduced in order to assess whether water extraction accompanying coal seam gas production will have deleterious impacts. Key improvements in the prediction of coal seam gas water production rates and assessment of risks to groundwater resources could be achieved through adoption of the following initiatives: 1. Coal Seam Water Resources: Establishment of a program for baseline coal seam water resources to better determine the size of the water store in coal measures and hence the potential of these seams for water production in the future. Improved monitoring of water extraction from existing and future wells, including mandatory reporting, to better understand the distribution of well gas and water production per unit time and through this improve understanding of local aquifer hydraulics in the vicinity of well locations. This reporting should include water produced during the exploration phase.

CWiMI

46

The development of a central repository for data and analysis of well production rates and their distribution throughout the Bowen and Surat basins. These data are vital for iterative comparison, validation and improvement of water and gas production models from coal seams. 2. Impacts on Aquifers: Establishment of a baseline program monitoring coal seam water and other groundwater aquifers to establish the geochemical and isotopic signatures of these systems. An initial synoptic survey of coal seams and aquifers that includes dating techniques will provide a baseline for on-going monitoring and help constrain recharge rates and timing in the coal seams as well as connectivity with aquifers. On-going routine monitoring of these geochemical and isotopic properties could then be used to trace water exchange between coal seams and aquifers. The development of regional and local groundwater models that are conditioned using geochemical and isotopic information as well as remote sensing data. A summary of the likely useful isotopic tracers and dating techniques is detailed in Appendix 3. Determining the changes in potentiometric surface (head changes) in the gas fields will provide an estimate of the likely area of impact. The development of a comprehensive modelling scheme for improved forecasting of water production from the exploitation of coal seam gas reserves that includes forecasting of cumulative impacts (positive and negative) on groundwater resources and surface water/groundwater interactions. In addition there are a number of the key issues not considered in this work that should be addressed in future monitoring. These include: 1. Water Quality analysis. Draper and Boreham (2006) showed that the quality of water produced from CSG production varies widely both within fields and between fields. Water produced around Moranbah, for example, is generally saltier than water produced from either the central Bowen Basin or the Surat Basin. If connectivity is established between coal seams and multiple aquifers it is then possible that the water quality of productive aquifers may be compromised if connection is established with poorer quality aquifers. Additionally, better understanding of the spatial distribution of water quality will improve planning for beneficial water re-use and/or treatment options and may provide insights into aquifer connectivity. 2. Impact on aquifer recharge and groundwater dependent ecosystems. A remote sensing approach to monitor these potential impacts is outlined in Appendix 4. 3. Physical risks. There is little information available on the changes to physical properties of coal seams during CSG extraction. Some work is currently being undertaken at the University of Queensland in this regard.

CWiMI

47

Finally, an assessment of the positive and negative cumulative impacts of CSG development in the Bowen and Surat Basins should also be considered as imperative to guiding policy decisions aimed at successfully and sustainably managing CSG extraction and long term landuses. Balancing the costs of effective monitoring and solutions with adequate risk management is always challenging. A cumulative impact assessment would allow evaluation of the economic, social and environmental costs and benefits across the basins and interstate when considering downstream users of the GAB waters.

8 Conclusions It is clear from this scoping study that there are significant data limitations relating to coal seams and surrounding aquifers that must be dealt with to inform policy development with confidence. This report has presented a summary of these and other data gaps that were identified throughout the project and provided recommendations as to how this information could be obtained through an ongoing monitoring program involving collaboration with CSG companies. In the aquifer risk assessment it is hypothesised that there may be risks associated with aquifer connectivity when long term equilibrium conditions are altered due to gas extraction. A point-based risk assessment demonstrated that risks are spatially heterogeneous, with the Hutton and Springbok aquifers generally at higher risk relative to other aquifers due to the close vertical distance between the aquifers and coal seams. Further information on the properties of coal seams and surrounding aquifers and the connectivity between them is needed in order to more reliably predict the risks that may result due to dewatering of coal seams and subsequent gas extraction. It is also recommended that future risk assessments consider the impacts of well completion techniques used to enhance gas recovery. Given the spatial and temporal variability in gas and water relationships across the Surat and Bowen Basins, difficulties arose in predicting the quantities of water likely to be produced during industry expansion. A model was therefore created to provide broad estimates under the three possible CSG development scenarios provided by the DIP (10, 28 and 40Mtpa). At 2020, water co-produced with coal seam gas is estimated at 126, 196 and 281GL/year for the three respective CSG scenarios. Significant knowledge gaps also arose during this analysis including uncertainty around estimates of gas reserves, the relationship between water: gas production and the likely spatial pattern of tenure development. Further, there was uncertainty as to whether water extraction during exploration is included in company production reports provided to the DME. This emphasises a strong need for further information about the water extraction profile over the life cycle of individual wells. It is recommended that an on-going monitoring program should combine multiple approaches in order to obtain sufficient data to manage risks effectively. An adaptive management approach is recommended so that new findings from a number of different lines of evidence can be incorporated

CWiMI

48

into resource planning decisions as the information becomes available. The need for involvement from stakeholders (particularly the CSG industry) in formulating and implementing a monitoring program is essential. A major challenge in developing such a monitoring program will be around matching cost effectiveness with adequate risk management.

CWiMI

49

References BRUNNER, P., HENDRICKS FRANSSEN, H. J., KGOTLHANG, L., BAUERGOTTWEIN & KINZELBACH, W. (2007) How can remote sensing contribute to groundwater modelling? . Hydrologeology Journal, 15, 5-18. CBM/NGC MULTI-STAKEHOLDER ADVISORY COMMITTEE (2006) Coalbed Methane/ Natural Gas in Coal: Final Report. Government of Alberta. COLLON, P., KUTSCHERA, W., LOOSLI, H. H., LEHMANN, B. E., PURTSCHERT, R., LOVE, A., SAMPSON, L., ANTHONY, D., COLE, D., DAVIDS, B., MORRISSEY, D. J., SHERRILL, B. M., STEINER, M., PARDO, R. C. & PAUL, M. (2000) Kr-81 in the Great Artesian Basin, Australia: a new method for dating very old groundwater. Earth and Planetary Science Letters, 182, 103-113. DE CARITAT, P., KIRSTE, D., CARR, G. & MCCULLOCH, M. (2005) Groundwater in the Broken Hill region, Australia: recognising interaction with bedrock and mineralisation using S, Sr and Pb isotopes. Pergamon-Elsevier Science Ltd. DIXON, B. (2004) Prediction of ground water vulnerability using integrated GIS-based Neuro-Fuzzy techniques. Journal of Spatial Hydrology, 1-38. DIXON, W. & CHISWELL, B. (1994) Isotopic study of alluvial groundwaters, southwest lockyer Valley, Queensland, Australia. Hydrological Processes, 8, 359-367. DRAPER, J. J. & BOREHAM, C. J. (2006) Geological controls on exploitable coal seam gas distribution in Queensland. The APPEA Journal, 46, 343-366. ESTERLE, J. & SLIWA, R. (2002) Bowen Basin Supermodel 2000. ACARP Project C9021 Exploration and Mining Report 976C. Kenmore, Queensland, CSIRO Exploration and Mining. FETTER (2001) Applied Hydrogeology, Upper Saddle River, New Jersey, Prentice Hill. CWiMI

50

FRIEDL, M. A. (2002) Forward and inverse modeling of land surface energy balance using surface temperature measurements. Remote Sens. Environ., 79, 344-354. GREEN, P. & RANDALL, R. (2008) Queensland's coal seam gas industry continues to brighten. Queensland Government Mining Journal. HARBISON, J., O'NEAL, B. & HAZEL, C. (2008) Regulation of Groundwater Resources Management with Regard to Coal Seam Gas Development for Queensland. Asia Pacific Coalbed Methane Symposium. Brisbane, Australia, University of Queensland. HARRISON, S., MOLSON, J., ABERCROMBIE, H. & BARKER, J. (2000) Hydrogeology of a coal-seam gas exploration area, southeastern British Columbia, Canada: Part 2. Modeling potential hydrogeological impacts associated with depressurizing. Hydrogeology Journal, 8, 623-635. JACKSON, T. J. (2002) Remote sensing of soil moisture: Implications for groundwater recharge. Hydrogeology Journal, 10, 40-51. KINNON, E. C. P., GOLDING, S. D., BOREHAM, C. J., BAUBLYS, K. A. & ESTERLE, J. S. (2008) Stable isotope and water quality analysis of production waters and gases from coal bed methane production. Asia Pacific Coalbed Methane Symposium. Brisbane, Australia, The University of Queensland. LEBLANC, M., RAMILLIEN, G., TREGONING, P., TWEED, S. & FAKES, A. (2008) Drought detection in the Murray-Darling basin from space gravity and hydrologic observations. Western Pacific Geophysics Meeting. Cairns. NORMAN, J. M., ANDERSON, M. C., KUSTAS, W. P., FRENCH, A. N., MECIKALSKI, R., TORN, R., DIAK, G. R., SCHMUGGE, T. J. & TANNER, B. C. W. (2003) Remote sensing of surface energy fluxes at 101-m pixel resolution. Water Resour. Res., 39. PARSONS BRINCKERHOFF (2004) Coal Seam Gas Water Management Study NRO0011. Brisbane, Department of Natural Resources, Mines and Energy. CWiMI

51

QUEENSLAND DEPARTMENT OF NATURAL RESOURCES AND MINES (2003) Bowen Basin Geology: Geological Distribution of Coal in the Bowen Basin. IN MUTTON, A. J. (Ed.) Queensland Coals 14th Edition. QUEENSLAND DEPARTMENT OF NATURAL RESOURCES AND MINES (2005) Hydrogeological Framework Report for the Great Artesian Basin Water Resource Plan Area. RENZULLO, L. J., BARRETT, D. J., MARKS, A. S., HILL, M. J., GUERSCHMAN, J. P., MU, Q. Z. & RUNNING, S. W. (2008) Multi-sensor model-data fusion for estimation of hydrologic and energy flux parameters. Remote Sensing of Environment, 112, 1306-1319. RIGHTMIRE, C. T. (1984) Coalbed Methane Resource. IN RIGHTMIRE, C. T., EDDY, G. E. & KIRR, J. N. (Eds.) Coalbed Methane Resources of the United States. Oklahoma, The American Association of Petroleum Geologists (AAPG). RODELL, M., FAMIGLIETTI, J. S., CHEN, J., SENEVIRATNE, S. I., VITERBO, P., HOLL, S. & WILSON, C. R. (2004) Basin scale estimates of evapotranspiration using GRACE and other observations. Geophysical Research Letters, 31, 4. SANDER, P. (2007) Lineaments in groundwater exploration: a review of applications and limitations. Hydrogeology Journal, 15, 71-74. THOMAS, L. (2002) Coal Geology, West Sussex, John Wiley & Sons. TWEED, S. O., LEBLANC, M., WEBB, J. A. & LUBCZYNSKI, M. W. (2007) Remote sensing and GIS for mapping groundwater recharge and discharge areas in salinity prone catchments, southeastern Australia. Hydrogeology Journal, 15, 75-96. WELSH, W. D. (2000) A Steady State Groundwater Flow Model of the Great Artesian Basin. Canberra, Bureau of Rural Sciences. CWiMI

52

WITHERBEE, K. G., SALWEROWICZ, F. A. & HOFFMAN, K. L. (1992) Environmental and regulatory aspects of the management of coalbed methane development and production, northern San Juan Basin, Colorado. Environmental Issues and Waste Management in Energy and Minerals Production. Rotterdam, Balkema. WOODS, P. H., WALKER, G. R. & ALLISON, G. B. (1990) Estimating Groundwater Discharge at the Southern Margin of the Great Artesian Basin near Lake Eyre, South Australia. International Conference on Groundwater in Large Sedimentary Basins. Perth, Australian Water Resources Council.

CWiMI

53