Characterisation of induced fracture networks within an enhanced geothermal system using anisotropic electromagnetic modelling

Journal of Volcanology and Geothermal Research 288 (2014) 1–7

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research journal homepage: www.elsevier.com/locate/jvolgeores

Characterisation of induced fracture networks within an enhanced geothermal system using anisotropic electromagnetic modelling Jake MacFarlane a,⁎, Stephan Thiel b,a, Josef Pek c, Jared Peacock d,a, Graham Heinson a a

University of Adelaide, School of Earth & Environmental Sciences, Adelaide, South Australia, Australia Geological Survey of South Australia, GPO Box 320, Adelaide, South Australia 5001, Australia Institute of Geophysics AS CR, Bocni II/1401, Prague 4 Czech Republic d U.S. Geological Survey, 345 Middlefield Road, Menlo Park, CA, USA b c

a r t i c l e

i n f o

Article history: Received 5 March 2014 Accepted 5 October 2014 Available online 12 October 2014 Keywords: Geothermal Magnetotellurics Fluid injection Fractures Forward modelling Electrical anisotropy

a b s t r a c t As opinions regarding the future of energy production shift towards renewable sources, enhanced geothermal systems (EGS) are becoming an attractive prospect. The characterisation of fracture permeability at depth is central to the success of EGS. Recent magnetotelluric (MT) studies of the Paralana geothermal system (PGS), an EGS in South Australia, have measured changes in MT responses which were attributed to fracture networks generated during fluid injection experiments. However, extracting permeabilities from these measurements remains problematic as conventional isotropic MT modelling is unable to accommodate for the complexities present within an EGS. To circumvent this problem, we introduce an electrical anisotropy representation to allow better characterisation of volumes at depth. Forward modelling shows that MT measurements are sensitive to subtle variations in anisotropy. Subsequent two-dimensional anisotropic forward modelling shows that electrical anisotropy is able to reproduce the directional response associated with fractures generated by fluid injection experiments at the PGS. As such, we conclude that MT monitoring combined with anisotropic modelling is a promising alternative to the micro-seismic method when characterising fluid reservoirs within geothermal and coal seam gas reservoirs. © 2014 Published by Elsevier B.V.

Introduction With global energy production shifting away from fossil fuels, the development of economically viable renewable energy sources is gaining significant interest. One method which is showing promising results is geothermal, more specifically enhanced geothermal systems (EGS), with the Habanero EGS in Australia generating net power output without assistance (Geodynamics Ltd, 2013) and power production from EGS estimated to provide 100,000 MW of energy to the United States by 2050 (Tester et al., 2006). In a geothermal system, the naturally occurring temperature gradient within the Earth is utilised as a means of generating electricity through the heating of injected fluids. To generate energy efficiently, permeable pathways within hot lithologies must first be established through hydraulic fracturing. As such, the monitoring of injected fluids during fracturing and the characterisation of the resulting permeable zones are both crucial to the feasibility of an EGS. Currently, micro-seismic tomography is the primary geophysical technique utilised when characterising fractures generated during hydraulic fracturing (House, 1987; Albaric et al., 2013). From this method we are able to estimate spatial distribution of fractures. However, as this method is not directly sensitive to the fluids present within those ⁎ Corresponding author. Tel.: +61 8 8313 5493. E-mail address: [email protected] (J. MacFarlane).

http://dx.doi.org/10.1016/j.jvolgeores.2014.10.002 0377-0273/© 2014 Published by Elsevier B.V.

fractures, it is difficult to gain any information regarding the permeability or fluid motion occurring at depth. To rectify this issue, studies have begun to use electromagnetic (EM) methods which are directly sensitive to subsurface conductivity contrasts (Zlotnicki et al., 2003; Aizawa et al., 2005; Yasukawa et al., 2005; Peacock et al., 2012, 2013). One such method showing potential is the passive EM technique, known as MT with multiple authors using this method to locate and characterise potential geothermal targets (Heise et al., 2008; Newman et al., 2008; Arango et al., 2009; Spichak and Manzella, 2009; Geiermann and Schill, 2010). Furthermore, studies have also begun to identify the monitoring capabilities of MT Aizawa et al., 2011 with Peacock et al. (2012, 2013) measuring subtle conductivity contrasts within an EGS which were attributable to hydraulic fracturing. However, the spatial resolution of MT measurements decreases with depth. Therefore, EGS targets which are typically small and located at 3–5 km depth are unresolvable using conventional isotropic modelling. One recent example of this is a study by Peacock et al. (2012, 2013) of the Paralana geothermal system (PGS), an EGS at Paralana, South Australia. During July, 2011, Petratherm Ltd. and joint venture partners Beach Energy and TruEnergy injected 3.1 million litres of fluid over 4 days into a metasedimentary package at 3680 m depth. The PGS is situated in a dilational zone along a splay off the eastward thrusting Paralana fault system (Paul et al., 1999; McLaren et al., 2002; Brugger et al., 2005) bounding the eastern margin of the Mt. Painter Domain

2

J. MacFarlane et al. / Journal of Volcanology and Geothermal Research 288 (2014) 1–7

Fig. 1. Topographic location map of the Paralana geothermal system in South Australia with MT stations (from Peacock et al., 2012) displayed as black triangles and a star marking the Mount Painter Domain. The top right inset displays a map of Australia specifying the location of the Paralana geothermal system using a star. The bottom right inset displays a section of the station array with increased magnification. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(Brugger et al., 2005) in the Northern Flinders Ranges, South Australia (Fig. 1). The Mt. Painter Domain consists of granites, gneisses and meta sediments dated at approximately 1600 Ma to 1580 Ma in age (Kromkhun, 2010). These units were overlain by sediments (Paul et al., 1999; McLaren et al., 2002; Brugger et al., 2005; Wülser, 2009) with a maximum age of 800 Ma (Wülser, 2009) and a lower age limit constrained to the initiation of the Delamerian Orogen (Wülser, 2009), which occurred between 514 Ma and 492 Ma (Foden et al., 2006). Further granitic intrusions and tectonothermal events have been recorded throughout the history of the Mt. Painter Domain (Wülser, 2009) with the British Empire Granite intruding at approximately 460 to 440 Ma (Elburg et al., 2003; McLaren et al., 2006; Wülser, 2009). To understand the pre- and post-fluid injection MT data collected at the PGS, micro-seismic (Albaric et al., 2013) and MT (Peacock et al., 2012) responses were collected before and after hydraulic fracture stimulation. From these responses a series of micro-fractures preferentially aligned towards the north-east were interpreted to have opened in response to the injected fluids from phase tensor misfit ellipses calculated from MT data measured before and after fluid injection. However, subsequent isotropic forward MT modelling did not adequately reproduce the measured responses (Peacock et al., 2013). As such, this study utilises results presented by Wannamaker (2005) to hypothesise that the preferentially orientated fracture network generated within the PGS produces an anisotropic response which can be characterised by a 2-dimensional anisotropic forward model. A series of initial forward models are first generated to understand how each anisotropic modelling parameter influences the synthetic MT response. Preliminary results from these models along with results presented by Albaric et al. (2013) and Peacock et al. (2012, 2013) are then used to constrain an anisotropic forward model which adequately approximates MT data obtained by Peacock et al. (2012, 2013). From these results, we show that anisotropic MT modelling is able to overcome the limitations inherent of isotropic modelling and characterise the permeable pathways generated during hydraulic fracture stimulation within an EGS.

Hypothesis testing with forward models To understand the fracture geometries generated at the PGS, a series of forward models are generated. Subtle variations in the modelled structures allowed us to test whether the responses associated with various anisotropic structures are capable of approaching the measured response. Each model calculated using the 2-dimensional MT direct code for conductors with arbitrary anisotropy of Pek and Verner (1997) has a common 1-dimensional, layered background defined by geological information from previous studies (Paul et al., 1999; McLaren et al., 2002; Brugger et al., 2005; Wülser, 2009; Kromkhun, 2010). This layered background consists of two sedimentary layers, the first extending to a depth of 1.06 km with a resistivity of 5 Ω m and the second from 1.06 km to 2.06 km with a resistivity of 10 Ω m following previous forward modelling by Peacock et al. (2013). These layers overlie a highly-compacted sedimentary layer from 2.06 km to 7.1 km with a resistivity of 200 Ω m, defined by 2-dimensional MT inversions of the PGS presented by Peacock et al. (2012). The remainder of the model, from 7.1 km to 417 km, is defined by a homogeneous resistive half space with a resistivity of 10,000 Ω m. In order to represent the fluid injection into a fractured rock measured by Albaric et al. (2013) and Peacock et al. (2012), a 1 km wide block is introduced from 3.66 km to 4.46 km depth (Fig. 2). As the conductivity defining this volume is dependent on numerous variables, − upper (σ+ HS) and lower (σHS) Hashin–Shtrikman bounds for the electrical conductivity are calculated by (Hashin and Shtrikman, 1962)

þ

σ HS ¼ σ w þ

1−ϕ þ 3σϕ

1 σ m −σ w

ð1Þ

w

and −

σ HS ¼ σ m þ

ϕ þ 1−ϕ 3σ

1 σ w −σ m

m

ð2Þ

J. MacFarlane et al. / Journal of Volcanology and Geothermal Research 288 (2014) 1–7

Horizontal distance (km) 258.2 263.0 265.0 267.0 269.0 271.0 273.0

−αz

2

3.66km 4.46km

1

5.3 7.10km

7.4

1km

0

Log Resistivity

1.06km 2.06km

16.4 -1

71.9 415.9 11

21

31

41

51

61

71

-2

81

Station number Fig. 2. Model of the resistivity structure from the Paralana geothermal system when a resistivity of 200 Ω m is assumed for the layer between 2.06 km and 7.10 km depths, following previous forward modelling by Peacock et al. (2013). This model was input into the 2-dimensional MT direct code for conductors with arbitrary anisotropy (Pek and Verner, 1997) to model the MT response associated with the post-fluid injection MT data measured by Peacock et al. (2012). The values displayed define the depth and width in kilometres assigned to each resistivity domain.

to define how the bulk conductivity of the modelled fracture network changes with depth for interconnected and disconnected fractures respectively. Here σm, σw and ϕ correspond to the host rock conductivity, fluid conductivity and porosity respectively. For these calculations, we define the conductivity of the fluid σw using a linear relationship interpreted from Fig. 1 of Nesbitt (1993). This linear relationship explains how the conductivity of a fluid σw varies with depth z for a surface fluid conductivity σ0, surface temperature T0 and geothermal gradient ΔT.

σw ¼ σ0 þ

T 0 þ zΔT : 10

ð4Þ

ϕ ¼ ϕ0 e

3

1.3 3.3

Furthermore, the porosity ϕ of the host rock is defined by Athy (1930)

4

0km

0

Depth (km)

283.3

3

ð3Þ

where ϕ0 is the surface porosity, α is the compaction coefficient and z is the depth (Fig. 3a). To reduce ambiguity, we define the surface fluid conductivity (σw = 3 S/m) using conductivity measurements of the injected brine (Peacock et al., 2012) while the geothermal gradient (ΔT = 52 °C per km) is constrained by previous work within the region (Neumann et al., 2000). In addition, the host rock conductivity is defined by the compacted sediments which host the 1 where modelled fracture network σ m = 0.005 S/m (σm = ρ− m ρm = 200 Ω m). The surface porosity (ϕ0 = 50%), the compaction coefficient (α = 1500 m− 1) and surface temperature (T0 = 20 °C) were defined arbitrarily. Subsequently, the reciprocal of the conductivities calculated by Eqs. (1) and (2) generates the curves presented in Fig. 3(b) and (c) which represent the range of resistivities available to define the modelled fracture network (ρHS + =σ−1 HS + and ρHS − = σ−1 HS −). To calculate the synthetic MT responses, we use the 2-dimensional MT direct code for conductors with arbitrary anisotropy of Pek and Verner (1997) over a grid of 89 × 89 cells (horizontal × vertical). These include 15 air layers and 7 surface padding layers where these cells extend laterally across the entire model and define the first 1 km of the model. Underlying this is a block within the centre of the model, defined by 52 × 33 cells, with a constant mesh spacing of 200 m. The remaining cells define a further padding zone where multiplication factors of 1.5 and 1.2 are applied to the horizontal and vertical mesh spacings, respectively. Isotropic modelling As an initial hypothesis test, we attempt to replicate post-fluid injection MT data using four isotropic models with resistivities of 1 Ω m (ρHS +), 10 Ω m, 100 Ω m and 180 Ω m (ρHS − for σm = 0.005 S/m) defining the modelled fracture network within each model. A fifth isotropic model is also produced with the fracture network removed to

Porosity (%) 0

Depth (m)

1000

0

10

20

30

40

50

60

70

80

90

100

a)

2000 3000 4000 5000 6000 10-1 0

Depth (m)

1000

100

101

102

103

104

b)

2000 3000 4000 5000 6000

Fig. 3. (a) displays how porosity decreases with depth when a surface porosity of 50% is assumed, as per Eq. (4). Porosity information from (a) served as input to the calculations required to generate the curve presented in (b). (b) displays resistivity values calculated by Hashin–Shtrikman bounds for a host rock resistivity (ρm) of 200 Ω m. Note, ρHS + (solid line) corresponds to the Hashin–Shtrikman upper bound whereas ρHS − (broken line) corresponds to the Hashin–Shtrikman lower bound.

4

J. MacFarlane et al. / Journal of Volcanology and Geothermal Research 288 (2014) 1–7

approximate the pre-fluid injection MT data. In order to best compare the data generated by the synthetic models to the data collected by Peacock et al. (2012), we introduce the phase tensor (PT) defined by Caldwell et al. (2004). This property of the MT impedance tensor is resistant to near-surface distortions caused by unresolved structures and will minimise the influence that distortions will have on our final results. To accommodate for the 4D monitoring we are attempting to reproduce, we define the residual phase tensor (Heise et al., 2007). Residual PT ellipses Φ allow us to graphically represent how the phase changes between two MT data sets and are calculated for each model using the equation (Heise et al., 2007)
 −1 Φ ¼ I− Φ1 Φ2

ð5Þ

where I is the identity matrix, Φ−1 is the inverse of the pre-injection PT 1 and Φ2 is the post-injection PT. We then compare the PT misfit ellipse for each isotropic model (Fig. 4) and compare them to the MT responses measured by Peacock et al. (2012) (Fig. 5) to analyse whether an isotropic body is capable of reproducing the measured responses. From Eq. (5), we are able to derive the rotationally invariant parameters △ Φmax and △ Φmin which are the maximum misfit and minimum misfit respectively. These then relate to the size, representing the magnitude of the misfit, and face colour, representing the geometric mean of pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the misfit ( △Φ max △Φ min ), of each misfit ellipse for a given period (Heise et al., 2008). The major axis of the ellipse indicates the direction which experienced greatest change (Heise et al., 2007). Although the isotropic responses approximate the long axis magnitude when large changes in resistivity occur, the inability to reproduce both the ellipse orientation and minor axis magnitude leads us to believe that the measured response is not isotropic. This result coincides with 3-dimensional isotropic forward modelling presented by Peacock et al. (2013) which required the isotropic block to be larger than the block inferred from the micro-seismic response presented by Albaric et al. (2013) to reproduce the measured response.

Fig. 4. Pseudosection of PT misfit ellipses calculated from station 45 for four isotropic forward models with resistivities of 1 Ω m (ρHS −), 10 Ω m, 100 Ω m and 180 Ω m (ρHS + for σm = 0.005 S/m) defining the fracture network within each model. PT misfit ellipses are calculated with respect to a base model which is defined by an isotropic fracture network with a resistivity of 200 Ω m. The ellipse face colour indicates the geometric mean of the misfit while its ellipticity represents the orientation and magnitude of the maximum misfit and minimum misfit to three times the median of all stations' maximum principal direction, represented by the blue reference ellipse on the top left side of the plot. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5. Pseudosection of PT misfit ellipses calculated for pre- and post-fluid injection data from the east–west profile measured by Peacock et al. (2012). The bottom right inset displays a section of the station array at the Paralana geothermal system with MT stations plotted by triangles. Red triangles correspond to the MT stations plotted by the pseudosection. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Influence of anisotropic modelling parameters Following the inability of our isotropic models and those presented by Peacock et al. (2013) to reproduce the measured data, we then attempt to reproduce the measured data using a model where the fracture network is represented by anisotropic resistivity. Electrical anisotropy, defined as the directional dependence of electrical resistivity within a medium, is an important property to consider when studying the electrical properties of the Earth's interior (Rikitake, 1948; Martí, 2014). Central to electrical anisotropy is the diagonalisability of the 3 × 3 resistivity tensor. This allows anisotropic structures to be defined by three principal resistivities ρx, ρy and ρz and three rotation angles αS, αD, αSL (Fig. 6). According to their physical relation, these angles can be identified as geological strike, dip and slant while x, y and z are mutually orthogonal directions with positive z downward allowing us to define the resistivity parallel and perpendicular to the fault plane. However before attempting to reproduce the responses measured at the PGS using an anisotropic model, we investigate how synthetic responses change when the anisotropy and the geometry of the modelled fracture network are varied. The anisotropic model from which the following scenarios will vary from is defined by an anisotropic strike of 30° to approximate the major ellipse axis presented by Peacock et al. (2012) with principal resistivities assigned to the fault network are constrained by the Hashin– Shtrikman bounds displayed in Fig. 3. As ρy is modelled as being perpendicular to the fault plane, we define it using the reciprocal of the Hashin–Shtrikman lower bounds (ρy = ρHS −) of 180 Ω m such that it approximates the host layer with disconnected porosity. The remaining principal resistivities, ρx and ρz, are within modelled to be within the fault plane and therefore approximate the host layer with interconnected porosity. As such we define these resistivities using the reciprocal of the Hashin–Shtrikman upper bound (ρx = ρz = ρHS +) of 1 Ω m. A dip of 40° towards 300° is used to approximate the dip interpreted by Albaric et al. (2013). Finally the slant is set to 0° as no evidence supporting the presence of slant is present. Fig. 7(a) shows how the synthetic MT response behaves when varying the strike of the fracture network. For all modelled scenarios presented in Fig. 7(a), an observable relationship between the strike and the ellipse orientation exists which causes the long axis of the ellipse to orientate itself parallel to the anisotropy strike. In contrast to strike, Fig. 7(b) shows how the synthetic MT response behaves when varying the dip of the fracture network for a constant strike angle of 30°. Low dip angles continue to produce highly elliptical responses, similar to

J. MacFarlane et al. / Journal of Volcanology and Geothermal Research 288 (2014) 1–7

a)

b) rotation around z

c) rotation around x

5

d) rotation around z

Fig. 6. Successive rotations which define the resistivity tensor. The rotations are applied to the Cartesian coordinate system, {x,y,z}. The rotation angles αS, αD and αSL, which are the ‘strike’, ‘dip’ and ‘slant’ anisotropy angles, define a rotation around the z-, x′- and z″-axes respectively in degrees to form the coordinate system {x‴, y‴, z‴}. This coordinate system defines the resistivity tensor principal axes. Adopted from Pek and Verner (1997).

those in Fig. 7(a). However as the dip increases, large variations in both the magnitude of the misfit and the orientation of the maximum misfit occur. In contrast to changes due to the fault orientation, altering the resistivity within the fault plane (ρx,z) has very little effect on the misfit (Fig. 8a) while ρy is equal to 200 Ω m. This is due to the thin nature of the anisotropic block which limits the MT methods sensitivity to the block and restricts its ability to resolve any observable difference between a ρx,z of 1 Ω m or of 180 Ω m. However, when we decrease the resistivity perpendicular to the fault plane to 180 Ω m, the magnitude of minimum misfit increases significantly. This is because we are introducing a variation in the orientation of ρy with respect to the base model which causes the magnitude of the minimum misfit to vary. Furthermore, while the depth to fracture network has little effect on the geometry and magnitude of the misfit, significant variations regarding the

period band attributed to the plotted PT misfit ellipses occur as the depth is varied (Fig. 8b). We then split the modelled fracture network into two independent fracture zones to investigate how their individual responses interact. Fig. 9(a) investigates how the response changes when varying the strike angle of the second domain while the strike angle of the first domain remains constant. When the anisotropic strike defining the deeper fracture network is equal to that of the first (αS2 = αS1), the two layers produce identical responses and therefore appear as though they are a single block. However, as the strike angle is found to be related to the orientation of misfit, a difference in strike (αS2 ≠ αS1) causes the misfit ellipses to represent an average of the misfit produced by each fracture zone. This averaging effect decreases for period bands away from the interface between the two fractured zones. Furthermore, this averaging is reduced further when the two fracture zones are separated. This causes the individual responses to plot over different periods and subsequently reduce the averaging effect present in Fig. 9(a). Application to field data

a)

From these results, we observe that electrical anisotropy is capable of varying the orientation and magnitude of maximum misfit and minimum misfit. Phase tensor misfit ellipses are then calculated for the anisotropic forward model of the PGS (Fig. 10) and then compared to ellipses calculated for the pre- and post-injection MT data collected by Peacock et al. (2012).

b) a)

Fig. 7. Pseudosections of PT misfit ellipses calculated for an anisotropic forward model calculated with respect to a base model which is defined by an isotropic fracture network with a resistivity of 200 Ω m. (a) PT misfit ellipses plotted for varying angles of strike and (b) PT misfit ellipses for varying angles of dip when a strike of 30° is assumed.

b)

Fig. 8. Pseudosections of PT misfit ellipses calculated for an anisotropic forward model calculated with respect to a base model which is defined by an isotropic fracture network with a resistivity of 200 Ω m. (a) PT misfit ellipses for varying values of minimum (ρx,z) and maximum (ρy) resistivities defining the modelled fracture network and (b) PT misfit ellipses for varying depths to the top of the modelled fracture network.

6

J. MacFarlane et al. / Journal of Volcanology and Geothermal Research 288 (2014) 1–7

a)

b)

interpreted that this characteristic response is caused by the orientated micro-fractures induced by fluid injection, we believe that these results provide evidence supporting the hypothesis that anisotropic forward modelling is able to characterise the major permeable pathways within the PGS. Conclusions

Fig. 9. Pseudosections of PT misfit ellipses calculated for an anisotropic forward model calculated with respect to a base model which is defined by an isotropic fracture network with a resistivity of 200 Ω m. (a) PT misfit ellipses plotted for a scenario where strike varies with depth while maintaining an anisotropic ratio of ρx,z = 1 Ω m and ρy = 200 Ω. (b) PT misfit ellipses plotted for a scenario when a separation between two independent fracture networks is introduced.

Comparison between observed and modelled data For both data sets (Figs. 5 and 10), the main effects are between 2 and 0 s and are represented by a zone of highly elliptical responses. Observable similarities such as the ellipse orientation and period suggest that these responses are the result of similar features. It is important to note that discrepancies between the two data sets including the magnitude of the misfit exist, especially when compared to the very large ellipses plotted between 5 and 8 s. We believe that these differences are caused by errors within the measured data and limitations within the modelling algorithm restricting the amount of complexity we can introduce to the model. The most significant limitation is the 2-dimensionality of the code. Further limitations result from the mesh definition which restricts the amount of cells we can concentrate around the anisotropic block. These limitations restrict our ability to adequately model the significant fracturing due to high injection pressures around the injection well causing the synthetic data to inadequately reproduce the previously mentioned ellipses between 5 and 8 s. Furthermore, we believe that the limitations within the modelling algorithm restrict our ability to model the non-mechanical porosities generated by the fluid and host rock interactions and as such we are unable to accommodate for variabilities other than those which occur within the fault plane. However when the aforementioned similarities are compared to results published by Peacock et al. (2012), who

Fig. 10. Pseudosection of PT misfit ellipses calculated for synthetic models of the pre- and post-injection MT data measured along the east–west profile at the Paralana geothermal system, as measured by Peacock et al. (2012), when a host rock resistivity (ρm) of 200 Ω m is assumed.

The presence of highly fractured zones within EGS is well documented. We have performed anisotropic forward modelling which successfully reproduces the directional component of MT responses which are associated with permeability formed during fracture stimulation experiments within an EGS. Despite geophysical ambiguity, synthetic responses were able to reproduce subtle differences in MT response which can be correlated to differences between pre- and postinjection MT data. The results of this study provide a case supporting the use of anisotropic modelling as a method of modelling changes to the subsurface caused by the flow of a conductive material through the crust with significant implications for the geothermal and coal seam gas industries. Acknowledgments The authors would like to thank Petratherm and their joint venture partners along with aboriginal land owners for land access. Furthermore, we would like to thank Petratherm and their joint venture partners for supporting this project. For financial support, we thank the School of Earth and Environmental Sciences at the University of Adelaide, the South Australian Department for Manufacturing, Innovation, Trade, Resources and Energy, the South Australian Centre for Geothermal Energy Research and the Institute for Mineral and Energy Resources. We would also like to thank AuScope and Goran Boren for the equipment used when collecting the data. In addition, I would like to thank everyone involved in the collection and processing of the data. References Aizawa, K., Yoshimura, R., Oshiman, N., Yamazaki, K., Uto, T., Ogawa, Y., Tank, S., Kanda, W., Sakanaka, S., Furukawa, Y., Hashimoto, T., Uyeshima, M., Ogawa, T., Shiozaki, I., Hurst, A., 2005. Hydrothermal system beneath Mt. Fuji volcano inferred from magnetotellurics and electric self-potential. Earth Planet. Sci. Lett. 235, 343–355. http://dx.doi.org/10. 1016/j.epsl.2005.03.023. Aizawa, K., Kanda, W., Ogawa, Y., Iguchi, M., Yokoo, A., Yakiwara, H., Sugano, T., 2011. Temporal changes in electrical resistivity at Sakurajima volcano from continuous magnetotelluric observations. J. Volcanol. Geotherm. Res. 199, 165–175. Albaric, J., Oye, V., Langet, N., Hasting, M., Lecomte, I., Iranpour, K., Messeiller, M., Reid, P., 2013. Monitoring of induced seismicity during the first geothermal reservoir stimulation at Paralana, Australia. Geothermics http://dx.doi.org/10.1016/j.geothermics.2013.10.013. Arango, C., Marcuello, A., Ledo, J., Queralt, P., 2009. 3D magnetotelluric characterization of the geothermal anomaly in the Llucmajor aquifer system (Majorca, Spain). J. Appl. Geophys. 68, 479–488. Athy, L.F., 1930. Density, porosity, and compaction of sedimentary rocks. AAPG Bull. 14, 1–24. Brugger, J., Long, N., McPhail, D., Plimer, I., 2005. An active amagmatic hydrothermal system: the Paralana hot springs, Northern Flinders Ranges, South Australia. Chem. Geol. 222, 35–64. Caldwell, T.G., Bibby, H.M., Brown, C., 2004. The magnetotelluric phase tensor. Geophys. J. Int. 158, 457–469. http://dx.doi.org/10.1111/j.1365-246X.2004.02281.x. Elburg, M.A., Bons, P.D., Foden, J., Brugger, J., 2003. A newly defined Late Ordovician magmatic-thermal event in the Mt Painter Province, northern Flinders Ranges, South Australia. Aust. J. Earth Sci. 50, 611–631. Foden, J., Elburg, M.A., Dougherty-Page, J., Burtt, A., 2006. The timing and duration of the Delamerian Orogeny: correlation with the Ross Orogen and implications for Gondwana assembly. J. Geol. 114, 189–210. Geiermann, J., Schill, E., 2010. 2-D magnetotellurics at the geothermal site at Soultz-sousForêts: resistivity distribution to about 3000 m depth. Compt. Rendus Geosci. 342, 587–599. Geodynamics Ltd, 2013. June 2013 quarterly activities report. Technical Report. Geodynamics Ltd. Hashin, Z., Shtrikman, S., 1962. A variational approach to the theory of the effective magnetic permeability of multiphase materials. J. Appl. Phys. 33, 3125–3131. Heise, W., Bibby, H.M., Caldwell, T.G., Bannister, S.C., Ogawa, Y., Takakura, S., Uchida, T., 2007. Melt distribution beneath a young continental rift: the Taupo Volcanic Zone, New Zealand. Geophys. Res. Lett. 34, L14313.

J. MacFarlane et al. / Journal of Volcanology and Geothermal Research 288 (2014) 1–7 Heise, W., Caldwell, T.G., Bibby, H.M., Bannister, S.C., 2008. Three-dimensional modelling of magnetotelluric data from the Rotokawa geothermal field, Taupo Volcanic Zone, New Zealand. Geophys. J. Int. 173, 740–750. House, L., 1987. Locating microearthquakes induced by hydraulic fracturing in crystalline rock. Geophys. Res. Lett. 14, 919–921. http://dx.doi.org/10.1029/GL014i009p00919. Kromkhun, K., 2010. Petrogenesis of High Heat Producing Granite: Implication for Mt Painter Province, South AustraliaPh.D. Thesis University of Adelaide. Martí, A., 2014. The role of electrical anisotropy in magnetotelluric responses: from modelling and dimensionality analysis to inversion and interpretation. Surv. Geophys. 35, 179–218. http://dx.doi.org/10.1007/s10712-013-9233-3. McLaren, S., Dunlap, W.J., Sandiford, M., McDougall, I., 2002. Thermochronology of high heat-producing crust at Mount Painter, South Australia: implications for tectonic reactivation of continental interiors. Tectonics 21, 1–2. McLaren, S., Sandiford, M., Powell, R., Neumann, N., Woodhead, J., 2006. Palaeozoic intraplate crustal anatexis in the Mount Painter Province, South Australia: timing, thermal budgets and the role of crustal heat production. J. Petrol. 47, 2281–2302. Nesbitt, B.E., 1993. Electrical resistivities of crustal fluids. J. Geophys. Res. Solid Earth 98, 4301–4310. Neumann, N., Sandiford, M., Foden, J., 2000. Regional geochemistry and continental heat flow: implications for the origin of the South Australian heat flow anomaly. Earth Planet. Sci. Lett. 183, 107–120. Newman, G.A., Gasperikova, E., Hoversten, G.M., Wannamaker, P.E., 2008. Threedimensional magnetotelluric characterization of the Coso Geothermal field. Geothermics 37, 369–399. Paul, E., Flöttmann, T., Sandiford, M., 1999. Structural geometry and controls on basement-involved deformation in the northern Flinders Ranges, Adelaide Fold Belt, South Australia. Aust. J. Earth Sci. 46, 343–354.

7

Peacock, J.R., Thiel, S., Reid, P., Heinson, G., 2012. Magnetotelluric monitoring of a fluid injection: example from an enhanced geothermal system. Geophys. Res. Lett. 39, L18403. Peacock, J.R., Thiel, S., Heinson, G., Reid, P., 2013. Time-lapse magnetotelluric monitoring of an enhanced geothermal system. Geophysics 78, B121–B130. Pek, J., Verner, T., 1997. Finite-difference modelling of magnetotelluric fields in two-dimensional anisotropic media. Geophys. J. Int. 128, 505–521. Rikitake, T., 1948. 1. Notes on electromagnetic induction within the earth. Earthquake Res. Inst. 24, 1–9. Spichak, V., Manzella, A., 2009. Electromagnetic sounding of geothermal zones. J. Appl. Geophys. 68, 459–478. Tester, J.W., Anderson, B.J., Batchelor, A., Blackwell, D., DiPippo, R., Drake, E., Garnish, J., Livesay, B., Moore, M., Nichols, K., et al., 2006. The Future of Geothermal Energy. Massachusetts Institute of Technology, p. 358. Wannamaker, P., 2005. Anisotropy versus heterogeneity in continental solid Earth electromagnetic studies: fundamental response characteristics and implications for physicochemical state. Surv. Geophys. 26, 733–765. Wülser, P.A., 2009. Uranium Metallogeny in the North Flinders Ranges Region of South AustraliaPh.D. Thesis University of Adelaide. Yasukawa, K., Ishido, T., Suzuki, I., 2005. Geothermal reservoir monitoring by continuous self-potential measurements, Mori geothermal field, Japan. Geothermics 34, 551–567. http://dx.doi.org/10.1016/j.geothermics.2005.04.005. Zlotnicki, J., Sasai, Y., Yvetot, P., Nishida, Y., Uyeshima, M., Fauquet, F., Utada, H., Takahashi, Y., Donnadieu, G., 2003. Resistivity and self-potential changes associated with volcanic activity: the July 8, 2000 Miyake-jima eruption (Japan). Earth Planet. Sci. Lett. 205, 139–154. http://dx.doi.org/10.1016/S0012-821X(02)01025-7.