Journal of Applied Geophysics 88 (2013) 8393

Contents lists available at SciVerse ScienceDirect

Journal of Applied Geophysics

journal homepage: www.elsevier.com/locate/jappgeo

The potential of audiomagnetotellurics in the study of geothermal elds: A case study from the northern segment of the La Candelaria Range, northwestern Argentina
Hernan Barcelona a,, Alicia Favetto a, Veronica Gisel Peri a, Cristina Pomposiello a, Carlo Ungarelli b
a b

Instituto de Geocronologa y Geologa Isotpica, Universidad de Buenos Aires-Consejo Nacional de Investigaciones Cientcas y Tecnolgicas, Buenos Aires, Argentina WesternGeco Geosolutions Integrated EM Center of Excellence, Milan, Italy

a r t i c l e

i n f o

a b s t r a c t
Despite its reduced penetration depth, audiomagnetotelluric (AMT) studies can be used to determine a broad range of features related to little studied geothermal elds. This technique requires a stepwise interpretation of results taking into consideration diverse information (e.g. topographic, hydrological, geological and/or structural data) to constrain the characteristics of the study area. In this work, an AMT study was performed at the hot springs in the northern segment of the La Candelaria Range in order to characterize the area at depth. Geometric aspects of the shallow subsurface were determined based on the dimensional and distortion analysis of the impedance tensors. Also, the correlation between structural features and regional strikes allowed us to dene two geoelectric domains, useful to determine the controls on uid circulation. The subsurface resistivity distribution was determined through 1D and 2D models. The patterns of the 1D models were compared with the morpho-structure of the range. Shallow and deep conductive zones were dened and a possible shallow geothermal system scheme proposed. A strong correlation was found between the AMT results and the geological framework of the region, showing the relevance of using AMT in geothermal areas during the early stages of subsurface prospecting. 2012 Elsevier B.V. All rights reserved.

Article history: Received 5 May 2012 Accepted 6 October 2012 Available online 13 October 2012 Keywords: Audiomagnetotellurics Geothermal eld Geoelectric strike La Candelaria Range Northwestern Argentina

1. Introduction The magnetotelluric (MT) method is widely applied in geothermal exploration due to its manifested sensitivity to detect the presence of uids. The main target is to determine the geometry of the geothermal system, including its extent in depth and the location of the source. The audiomagnetotelluric (AMT) method is usually used to improve the resolution of MT at shallow subsurface levels because of the higher frequencies detected. The seminal works of Hoover et al. (1976), Sandberg and Hohmann (1982) and Berktold (1983) adapted AMT for use to study geothermal areas. The main goal of AMT in geothermal systems is to dene the distribution of conductive and resistive zones associated with uids and cap rocks in the rst few hundred meters (Arango et al., 2009; Monteiro Santos et al., 1996). However, issues such as dimensionality, distortion and geoelectric strike are usually not considered to infer any geothermal eld characteristics. Although geothermal studies normally require a deeper penetration, a detailed analysis of the signal properties and data inversion can convert AMT into an efcient and relatively inexpensive method during the early stages of exploration, prospection or subsequent geothermal characterization.

Corresponding author at: Pabelln INGEIS, Ciudad Universitaria, Ciudad Autnoma de Buenos Aires, Argentina. Tel.: +54 11 4783 3021/3; fax: +54 11 4783 3024. E-mail address: h.lidenbrock@gmail.com (H. Barcelona). 0926-9851/$ see front matter 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jappgeo.2012.10.004

The La Candelaria mountain range (LCR) is located between morphotectonic units of the Cordillera Oriental and Santa Brbara system, in Salta province, northwest Argentina. Geothermal activity manifestations at the northern end including several hyperthermal hot springs have been described by Pesce and Miranda (2003). Seggiaro et al. (1997) proposed a geothermal water circulation model based on percolation and inltration of meteoric water in the upper highland areas where permeable rocks are exposed. Several discontinuities allow the conduction of water in depth where its temperature rises and emerges at lower topographic levels in the northern apex range. However, in the hot springs zone there are no geophysical studies to dene the shallow structure features, the structureuid relation, the local geometry of the geothermal system and the nature and extent of the occurrence. Here, we present AMT results from the northern segment of the LCR in order to dene the resistivity distribution at depth and infer some aspects related to the hot springs. We characterized the regional and local geology to determine the possible geoelectric response of the stratigraphic column. Also, structural analysis was helpful to dene structures that could have potentially strong effects on the local electromagnetic eld. Dimensional and distortion analyses were performed to focus on aspects that could contribute to detecting the geometry of the geothermal eld. The data were processed to obtain the resistivity distribution at depth in one and two dimensional models. The geoelectric structures dened were contrasted with the local geology, main structures and hydrogeological, geomorphological

84

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

and hydrological variations in order to constrain the interpretation. Finally, we classied the conductive anomalies based on their deduced sources.

2. Geological settings The LCR is part of an NS fault and thrust system evolved in response to the stress transference from the Western Cordillera to the foreland during the nal stages of Andean deformation (Mon and Gutierrez, 2007; Ramos, 1999). In particular, it represents a thick-skinned deformation front with a MiocenePliocene main event (Mon, 2001), but gently folded Pleistocene deposits reveal possible minor stages of subsequent compressive deformation. The stratigraphic column shows the basement deformation and describes a sequence of tectonic events associated with the evolution of the cretaceous rift system until the thermal subsidence stage interrupted by Andean orogeny deposits (Abascal, 2005; Iaffa et al., 2011). The basement exposed at the LCR is represented by the highly fractured Upper PrecambrianLower Cambrian metasedimentary rocks of the Medina Formation (Bossi, 1969; Ramos, 2008). Overlying this unit the partially cemented and fractured lower Cretaceous red sandstones of the Pirgua Subgroup lie unconformably (Galliski and Viramonte, 1988; Reyes and Salty, 1973; Viramonte et al., 1999). These continental deposits are related to the cretaceous synrift associated with crustal stretching and Gondwana break up. The Late Cretaceous Balbuena Subgroup overlies the Pirgua Subgroup (Moreno, 1970). It is constituted by conglomeratic sandstones, lutites and mudstones from the Lecho Formation and limestones from the Yacoraite Formation (Reyes and Salty, 1973; Turner, 1959). These units, together with the Paleocene continental deposits of the Santa Brbara Subgroup, correspond to the thermic subsidence rifting stage (Iaffa et al., 2011; Viramonte et al., 1999). Eocene to Oligocene continental sedimentation events related to different pulses during the rise of the Andes Cordillera after the Incaic tectonic phase are grouped as the Metn subgroup. Fluvial and uvio-lacustrine deposits of Ro Seco and Anta formations are exposed at the northern extreme of the LCR (Gebhard et al., 1974). Finally, the sequence is crowned by Miocene to Pliocene deposits of the Jujuy Subgroup (Gebhard et al., 1974). At the top, quaternary deposits do not exceed 50 m thickness (Berchei, 2003). The complete sedimentary sequence has a maximum thickness of 4500 m in adjacent basins and towards the LCR presents a signicant decrease in thickness due to the stacking basement (Cristallini et al., 1997). The main structural feature of LCR is an east verging high angle thrust exposed at the eastern margin and its associated fold (Moreno Espelta et al., 1975). LCR also presents a strong segmentation indicated by a NNE to NNW rotation of the longitudinal axis from south to north. Four segments separated by EW and ENEWSW lineaments can be recognized and the northernmost, probably the most modern one, correspond to the study area (Fig. 1). The southern margin of the northern segment of LCR is delimited by an EW inverted high angle fault dipping to the north, responsible for the northward stepwise thickness increase of the permeable Pirgua Formation (Salty and Monaldi, 2006). This segment is formed by two anticlines with different strikes, both dipping to the north. According to Seggiaro et al. (1997) these anticlines and the associated faults system are the structural framework of the geothermal system. The NS Balboa anticline is characterized by the cretaceous sandstone outcropping at its core. Also, the fold is dissected to the east by the main thrust or the intersection of two younger faults. The gentle NNWSSE Termas anticline was developed partially overlapping the preexisting structures and its formation is related to a high angle thrust with west trend (Seggiaro et al., 1997). The stratication at both sides of the anticline adopts an NS strike and presents a strong inection, probably controlled by a regional EW lineaments system dening a rearrangement or interference zone. Furthermore, NS,

Fig. 1. Geological map of the northern segment of La Candelaria Range and the location of the study area. Based on Salty and Monaldi (2006).

NESW and EW lineaments are present and some direct faults delimit the eastern anticline margin (Moreno Espelta et al., 1975).

3. Data acquisition and signal processing Thirty AMT sites were collected using Geometrics STRATAGEM with two BF6 sensors to measure the magnetic horizontal eld and two 60 m long dipoles with steel electrodes oriented NS and EW to measure the electric eld. The frequency used during the study ranged from 10 Hz to 1000 Hz, but depending on the curve features obtained, it eventually increased up to 7000 Hz to obtain good detail in the rst few tens of meters. The site locations were organized according to the structure and geological characteristics of the area. Consequently, the geometry of the grid was designed to be approximately rectangular (Fig. 2). In general, the nodes of the grid are approximately 1 km apart, but on the summit of the range the sampling distances were shorter to better describe the hot springs. On the anks of the range there is a lower density of sampling due to access difculties. The Geometrics Image software was used to control the registration and the data pre-processing and processing. These steps included ltering of the signal, manual gain setting for each channel, calculation of the complex tensor impedance (Z), resolution of transfer functions and acquisition of values for apparent resistivity and phase of the signal. More than thirty time series segments from each frequency band were stacked automatically for robust analysis in order to enhance the signal-noise ratio. The value of coherence threshold selected for electric and magnetic elds was 0.8, which allowed us to obtain high quality data.

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

85

Fig. 2. Location of the audio-magnetotelluric sites and the main proles selected.

4. Distortion and dimensional analysis of the impedance tensor The electromagnetic signal is inuenced by several factors such as the dimensionality of the propagation medium and the physical anomalies, which can distort the electromagnetic eld both locally and regionally. The determination of these effects allowed us to discriminate between the inherent properties of the medium from artifacts in the signal. Many types of geological distortions (Bahr, 1991) that may disturb the signal and mask regional information are common in geothermal environments. Therefore, the dimensional and distortion mutually dependent analysis needs to be carefully carried out to obtain reliable results from data inversions. Dimensional analysis consists in determining whether the features in the conductivity of the propagation medium present one, two or three dimensional variations. Mathematically, these variations are shown in Table 1. The distortion of Z was determined from the quantication of its asymmetry and the deviation from the conditions that dene its dimensionality. The parameters used for this purpose are all rotational invariant because the Z components involved in its denition are independent of the orientation system used. The conventional asymmetry parameter based on the Z magnitude is the skew dened by Swift (1967) as follows:   Z Z   xx yy   :  Zxy Zyx 

greater than 0.2, we assume 3D local anomaly (Bahr, 1991; Reddy et al., 1977). Also, Bahr (1988) proposed the phase sensitive skew which calculates the skew taking into account the distortions produced in Z over 2D structures by shallow conductive anomalies and is dened as follows: skewBahr where, S1 Zxx Zyy ; S2 Zxy Zyx D1 Zxx Zyy ; D2 Zxy Zyx : SkewBahr measures the deviation from the symmetry condition through the phase differences between each pair of tensor elements, considering that phases are less sensitive to surface distortions (i.e. galvanic distortion). The skewBahr threshold is set at 0.3 and higher values mean 3D structures (Bahr, 1991). We used both skews because they are complementary and are required to be zero in order to suppose a two-dimensional medium (Berdichevsky and Dmitriev, 2008). For each prole, Fig. 3 shows the values of skewSwift and skewBahr for each site at all frequencies. A few skew values are slightly above the threshold but without any preferential frequency distribution. Therefore, the skew parameters values suggest no signicant distortion and the general assumption of a 2D medium may be valid. Some authors have reported signicant variations in the skew parameters in the presence of noise in the signal (e.g. Jones and Groom, 1993; Lezaeta, 2002; Mart et al., 2004; Simpson and Bahr, 2005). In these cases the calculated skews are not reliable to quantify the distortion. An alternative and complementary method based on the WALDIM code developed by Mart et al. (2009) was used to determine the distortion and dimensionality of the study area. The code uses algorithms and statistical tests applied to a set of rotational invariants of Z dened by Weaver et al. (2000) and classied according to 1D, 2D, 3D and four superimposition models (3D/2D with a twist of
1 jD1 ; S2 S1 ; D2 j =2 jD2 j

skewswift

When the skewSwift is close to zero, we assume a 1D or 2D model when the skewSwift is close to zero, while when the skewSwift is
Table 1 Dimensionality criteria according to the relation of the magnetotelluric tensor components. Impedance tensor   Dimensionality 1D Zxx Zyy 0 Zxy Zyx 0 2D Zxx Zyy 0 Zxy Zyx 3D Zxx Zxy Zyx Zyy

Zxx Z Zyx

Zxy Zyy

86

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

Fig. 3. Frequency distributions of skewSwift (a) and skewBahr (b) site by site for the main proles. Dimensionality thresholds are represented by a gray layer.

Z, 3D/1D2D structure, 3D/1D2D structure resulting in a diagonal Z and a general 3D/2D). One of the main benets is that this procedure does not assume a dimension in advance. Also, the analysis weights the data error and its propagation to the invariants. The study area was analyzed using WALDIM dimensional models for all the frequency band and the results are shown in Fig. 4. The data revealed that the northern prole is mainly 1D. Particularly, at lower frequencies the trend is more diffuse and some 2D and 3D characteristics may be present. At high frequencies, the central prole behave as 3D and 2D, but at lower frequencies the medium is mainly 2D. The southern prole had a 1D, 2D and superimposed 3D/2D from the margins to the center. Also, mean dimensionality for each frequency band (1000 Hz 100 Hz and 100 Hz10 Hz) was performed. Fig. 5 shows the results as contour maps imposed over a digital elevation model. There is good agreement between a theoretical dimensionality dened by dominant geological features and the results obtained. Distal sites and the northern margin behave as 1D according to the subhorizontal stratication dominant structure. The two-dimensionality observed

near the range margins can be related to NS structures from the anticline. Also, the mountain range is 2D, but with sporadic local and supercial 3D distortion. It's interesting to point out that 3D/2D inuence in the signal showed at the northern segment of the studied area by the 10010 Hz frequency band is not present at higher frequencies. This change suggests that the hot spring water ow discharge at depth should be controlled by the roughly WE direct fault. However, the study area may not be treated as a 3D structure at this work scale. The 2D assumption during the inversion was accepted. The dimensional features of the studied area do not correspond with the classic 3D domes present in geothermal systems (e.g. Bibby et al., 2009; Jousset et al., 2011; Newman et al., 2008; Volpi et al., 2003). The two-dimensionality can be explained by the NS trend of LCR, but also by the inuence of fractures in the shallow uids circulation. In addition, the 3D shallow distortions may be interpreted as zones where fractures lost their longitudinal continuity. However, these inferences are dependent on the area covered and the penetration of the signal obtained.

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

87

Fig. 4. Dimensional variations for each prole based on WALDIM code.

5. Geoelectric strike and structural constrains In a two-dimensional case, the strike, , is dened as the rotation angle of Z to obtain the minimum values on its diagonal components related to the M parameter as follows:  2   2 jZxx j Zyy  M  2  2 :     Zxy  Zyz  This approach was developed by Swift (1967) and is based on the modules of Z feature that may be signicantly distorted by noise on the signal, local supercial inhomogeneities and/or static distortion. In these cases the Swift method is inappropriate (Bahr, 1988). One way to solve the problem is to decompose Z and separately obtain the distortion and regional contribution (Smith, 1995). This decomposition requires adopting a previous distortion model. The Groom and Bailey (1989) method (GB) is based on Z decomposition into a set of distortion tensors and a regional 2D model expressed as: h i h i S R Z e Z where Z S is the superimposed Z tensor, e is the electric distortion tensor and Z R is the anti-diagonal regional tensor. In addition, e is product of a scalar and the twist, shear and anisotropy tensors. These tensors rotates all vectors by a twist angle, shears the pattern by a shear angle and stretches the vectors by different factors, respectively. All together are called the distortion parameters. The regional strikes were obtained through the STRIKE code from McNeice and Jones (2001), based on the GB method with a 3D/2D superimposed galvanic distortion model. We carried out a single-site multi-frequency analysis. Besides, 2D Z R and distortion parameters were obtained after running the code. The data set had low distortions (shear and twist angles between 5 and +5 degrees) and parameter stability was obtained at all the frequency bands. The calculated strikes, site by site, at all frequencies are shown in Fig. 6. All the strikes show coincident directions through the frequency band (i.e. in depth) analyzed and spatially dene two domains. The northern and the occidental margin of the range present NS to NNE strikes while in the southeastern zone and partially over the edge of the range the strikes are clearly NE. The strikes obtained are consistent with the geological structures and their relationship is schematized in Fig. 7. The NNW to NNE

strikes from the western, middle and northern part of the study area are related to the strata, which adopt different orientations and inclinations according to the structural variations of the anticline. The N35E geoelectric strikes of the SE sector are associated with the NE structural lineaments that cross the Termas anticline and acquire their maximum expression dening the morphology of the Balboa anticline. The abrupt change of the strikes over the eastern margin of LCR can be explained by the EW cretaceous structures controlling the segmentation of the mountain range and they are expressed in the area by the strong inection shown by the Termas anticline.

6. Inversion of AMT data 6.1. 1D inversion 1D inversion of data was performed using the invariant curves for resistivity and phase. The resistivity and phase invariants were calculated as the geometric mean of the apparent resistivity and the arithmetic mean of the phases, respectively, using both electric and magnetic transverse modes. The Bostick algorithm was used as a rst qualitative approach and also it was inverted with the Occam algorithm to determine the minimum number of layers to obtain good t to the data (Constable et al., 1987). Commonly, a ve layer model (including the lower half-space) was enough to achieve a good t comparing the model response with the data curves measured at each site. Models are shown in Fig. 8 arranged by eastwest proles. The resistivity distribution revealed strong conductive anomalies at the northern prole and at the middle and margins of LCR over the central prole. In addition, another conductive anomaly was found at the southeastern part of the study area. The slightly higher conductivity at the east of the range coincides with a lower topographical level in this zone (e.g., compare between 160-350-450 and E2-310-410 groups of sites). The resistivity differences between deep layers at both sides of the range can be controlled by topography differences associated to hydrogeological aspects or unmapped lithological variations. A large number of models with conductiveresistive layers alternations in the rst 100 m of depth followed by a very conductive medium (b 2 m) were observed. They are located at higher gradients of slope, towards the foot of the gentle slopes and may be related to the lithological features and the geomorphologic intervention over the local hydrology, through the water runoff control.

88

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

Fig. 5. Contour maps representing mean WALDIM dimensional values for shallow (1000100 Hz band) and deeper (10010 Hz band) medium. Black triangles represent audio-magnetotelluric site location. Note that the north location is inverted.

Fig. 9 shows 1D models from the AMT sites located at the edge of LCR (roughly north to south prole). The geoelectric structure of the range is characterized by a resistivity medium of 810 m and the rst 100150 m is dominated by more resistive layers (>10 m). Also, there is an abrupt change in the resistivity distribution related to deep and highly conductive layers at sites 330 and 430. This anomaly is located near the hot springs zone and the occurrence is related to a change of the range slope, probably associated to cross EW to NE faults of the Termas anticline. The 1D geoelectric structures of the range correlate to their stratigraphic features. The resistivity of shallow levels is probably related to interstratication of shale and sandstones at the basal level of the Anta or Ro Seco formations. Thus, the conductiveresistive layers sequence could indicate a lithological

control over the supercial water ow. Also, the 10 m deep structure may be associated with cretaceous sandstones. 6.2. 2D inversion According to the results obtained from the dimensional analysis, three proles were chosen to perform 2D models using the WinGlink of Geometrics package. The inversion algorithm used was developed by Rodi and Mackie (2001) and is based on the non-linear conjugated gradient approach. The grid, including topography, was built following the recommendations for the maximization of the algorithm efciency. The cell dimensions between the stations were increased by a factor of 1.2, by 1.4 away from the stations and 1.2 at depth. The total

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

89

Fig. 6. GB regional strikes site by site at all frequencies are shown over a Google Earth satellite image. The length of the petals represents the number of measurements in each bin (10 width). Also, detailed structural map was added (same references as Fig. 1).

depth of the model was obtained as twice the skin depth for the higher value of apparent resistivity at the lower frequency obtained. The value of the half-space starting model was determined by the maximum resistivity found in the 1D models (b 40 m).

The model smoothness and the data mists were analyzed to set up the regularization parameter () for each prole. A regular trade off curve estimation, testing values from 1000 to 0.3, gave the best t for = 10.

Fig. 7. Relationship between the observed strikes and the regional structural environment. Two geoelectric domains are consistent with differential control of the geological structures.

90

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

Fig. 8. 1D resistivity models are shown for each site for the three main proles. The topography of the proles is shown by means of a digital elevation model (based on SRTM).

Both data modes (i.e. electric and magnetic transverse) were used during the inversion process. The apparent resistivity and phase error oor values decreased stepwise until nal values of 10% and 2.9, respectively, to allow the modeling of higher to lower order of geoelectric anomalies.

6.2.1. Results The geoelectric structures, their main characteristics and spatial variations can be represented through three 2D models. Fig. 10 shows the three EW sections in the northern, central and southern part of the study area of the LCR. The adjustment of all the models

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

91

Fig. 9. 1D models for the corresponding sites over the edge of La Candelaria Range. Site location is shown by means of a digital elevation model and constitutes a roughly NS prole.

are good (normalized root mean square between 1 and 1.5) and with homogeneous distribution of the mists among sites. The northern prole is located at the end of the range, where no relevant topographic changes are present, and the resulting model is represented in Fig. 10a. The resistivity distribution is relatively homogenous, characterized by the presence of a conductive anomaly (b 3 m) conned along the prole. The conductive anomaly top is placed at 150 m deep with a thickness of 300 m. The surrounding medium is less conductive (57 m) and the highest values of resistivity (30 m) are observed closer to the surface. The central prole, which partially crosses the hot springs area, is shown in Fig. 10b. The resistivity distribution is characterized by roughly vertical geoelectric structures. A shallow conductive anomaly that continues at depth is highlighted in the model and is placed below the hot springs area. This anomaly is truncated to the E about the 430 and 630 AMT sites. In addition, a shallow conductivity anomaly (2 m) that continues at depth is located to the eastern at topography sector of the prole. All these geoelectric structures are consistent with the corresponding 1D models. The prole at the southernmost part of the study area covers the surrounding plains and passes through the mountains with a topographic difference of 150 m between the midpoint and the ends. The resulting model is characterized by three anomalies (Fig. 10c). A central resistive anomaly that continues up to 500 m deep denes the low conductivity of LCR in this sector (>10 m) but real values might be overestimated by galvanic distortion. In addition, two different conductive structures are located at the foot of the slopes. The western conductive anomaly appears to be more supercial and is discontinued at depth. However, the anomaly of the eastern margin has its top at 200 m deep and continues downward. 6.2.2. Skin depth and sensitivity The geoelectric structures dened in 2D models were assessed according to the penetration of the signals. This feature is mainly related to the frequency and the resistivity distribution at depth. The maximum possible skin depth was estimated as a function of the

lower frequency of the data and the integrated conductance in the models at each site. The results allowed us to determine the maximum skin depth about 300 m and 400 m. Also, sensitivity maps were performed in order to determine those parts of the 2D resistivity models that are more sensitive to the data. The estimated sensitivity was calculated with the normalized value of the diagonal part of the term ATR 1A, were A represent the model parameters and R is the covariance matrix (Mackie, 2002). The sensitive maps shown at Fig. 10 g,h,i revealed that the conductive anomalies are better dened until about 400 m depth. Therefore, the sensitivity results are consistent with the maximum skin depth calculated. 6.2.3. Interpretation and classication of the anomalies We identied conductive geoelectric structures that can be described, interpreted and classied using different interrelated features. The structural characteristics, the resistivity variations associated with the lithostratigraphic control, the topographic features and the 1D model interpretations allowed us to dene shallow and deep conductive zones in the 2D models. Shallow conductive zones are located to the north and towards the margins of LCR. They are interpreted as a product of the interaction of various aspects which leads to the presence and/or accumulation of water. The main factor is associated with stratigraphic control over the permeability of Neogene continental and underlying cretaceous sandstones. Also, continental lacustrine MiocenePliocene or marine Miocene deposits interbedded in the sequence may facilitate the conductive response but act as aquiclude/aquitard in the hydrogeological system. The meteoric water runoff and the inltration assisted by discontinuities also enhance the drainage and water accumulation at depth. The extensive shallow anomaly in the northern segment of the study area (Fig. 10a) is an example of the interaction between all these factors. In this case, the dominant EW, ENEWSW and NS discontinuities over the Termas anticline inection allow the inltration of water into deeper permeable stratigraphic levels. Berchei (2003) identied three aquifer systems in the city of Rosario de la Frontera, 4 km NW of the study area. According to the author, the systems

92

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

70

40

25

16

10

3 2

a
900 700 500

W
160 150 140 130

b
900

W
430 030 350 340

630 020 320

c
900

W
435 450 440

730

120 110 E2

420 310

410

700 500
1 km 1 km

700

500
1 km

300 2 1 0 RMS

300 2 1 0
(ohm.m)

300 RMS 2 1 0
(ohm.m)

RMS

d
(Hz)

e
(Hz)

f
(Hz)

(ohm.m)

TM T observed (ohm.m)

calculated (ohm.m)

(Hz)

(Hz)

TE 4 observed

calculated

g
W 900 600 300 0
1 km

h
E
150 140 130 120 110 E2 160

i
W
350 340 030 430 630 020 320 310

(Hz)

900 600 300 0 -300

900 600 300 0

W
450 440

730 435 420

E
410

1 km

-300

-300 3E-2 sensitivity 2E-4 6E-6 0

1 km

Fig. 10. 2D inversion models (a, b, c), transverse magnetic and electric data and t (d, e, f) and, sensitivity maps for each model (g, h, i) for the EW proles at the northern (a), central (b) and southern (c) extremes of the study area. Histograms are shown to represent the mist per site by MRS. Note the vertical exaggeration at the models and the sensitive maps.

have a north drain direction. The shallow conductivity anomaly may be associated with the conned/semiconned aquifer system or be static-bearing. The shallow anomaly manifestation splits to the south. The EW and ENEWSW structures associated with the location of the hot springs and the topography control establish a boundary between the northern conductive front and the marginal systems of shallow conductive anomalies at the foot of the range slopes. This system was characterized by a better control over the ow and ltration of meteoric water associated with the decreasing slope. The system is partially discontinuous on the western margin and has longitudinal continuity associated with a dendritic drainage basin over the eastern margin. Deep conductive zones were dened as deep conductive anomalies related to the presence of uids that may potentially be associated with geothermal uid sources. An example of one such zone is located in the central prole under the western slope of the range (Fig. 10b). The deep character of the anomaly was supported by both a 2D model and a 1D layered pattern. Its spatial relationship with the hot springs and the favorable structural and topographic features allow us to suggest its potential link with geothermal activity. Also, according to dimensional, geological and morpho-structural

constraints a 500 m thick uid system related to uid-lled fractures, sporadic pipes with longitudinal variable extension, connecting a potential deeper reservoir, is possible in these zones. The model in Fig. 10c shows a signicant conductive anomaly in the SE sector of the area. Topographic features and spatial characteristics share aspects of shallow conductive zones. However, its distinguishing feature is the continuity at depth. In addition, its development correlates with a NE strike zone associated with the presence of deep faulting structures near the Balboa anticline (main recharge area of the geothermal system). Therefore, this conductive anomaly has shallow geoelectric structure features that may be interacting with potentially deep structures related to geothermal uids in a less constrained environment than that at the edge of the range. 7. Conclusions The results obtained regarding dimensionality and distortion analysis suggest a variable dimensional medium strongly correlated with geological features. 1D, 2D and 3D behaviors were related to horizontal stratication from the surrounding plain, NS anticline structure and local discontinuity fractures at the foot of the slope and on highs, respectively. However, the skew method applied supports the

H. Barcelona et al. / Journal of Applied Geophysics 88 (2013) 8393

93

2D assumption. The NS and NE strike domains were dened according to the geoelectric strikes and the dominant structural features. Both domains mark zones with possibly different structural control over the shallow uid circulation of the geothermal eld and particularly the NE domain can be related to deep-to-shallow uid transference. According to the dimensionality results this system can be related to uid-lled fractures and sporadic pipes with longitudinal variable extension, connecting a possible deeper reservoir. 1D and 2D models were used to characterize the entire resistivity structure of the mountain range. Also, stratigraphic control over the shallow uid circulation at the hot springs zone was found. Furthermore, we dened shallow and deep conductive zones based on their interpreted origin. The shallow conductive zones surrounding the range can be more than 300 m thick and are related to the inltration of meteoric water, channeled by fractures and accumulated at depth. The deep conductive zones can be potentially associated with a deeper geothermal reservoir, and are conned to the hot springs area and to the southeastern sector, probably related to the depth expression of the Balboa anticline. The multiple features related to the AMT method can be correlated with geothermal-bearing areas. The results from dimensionaldistortion analysis need to be carefully assessed to infer geometric properties of the geothermal eld at depth. Also, structural analysis and geoelectric strikes allow identifying the fractures with favorable geological conditions to transfer uids from depth. Looking for 1D models was helpful to nd the best interpretation for the resistivity structures observed, always focused on shallow characteristics but, in some cases, useful to constrain deeper anomalies. Furthermore, the integral AMT analysis allowed a better interpretation of 2D models based on a geologically constrained framework. According to these results, the relevance of using AMT to determine a broad range of features related to geothermal elds during the early stages of prospecting subsurface is evident. Acknowledgments Our special acknowledgment to Geotermia Andina for providing nancial support to improve the logistics. We kindly thank E. Llambias, for his invaluable help in the eld. Also, we thank Dr. Viramonte and his group from GEONORTE-INENCO Universidad de Salta CONICET for their contribution and discussions during the eld work. References
Abascal, L., 2005. Deformacin andina de la cuenca de Choromoro, NO de Tucumn: estilo estructural combinado. Revista de la Asociacin Geolgica Argentina 60 (2), 268277. Arango, C., Marcuello, A., Ledo, J., Queralt, P., 2009. 3D magnetotelluric characterization of the geothermal anomaly in the Llucmajor aquifer system (Majorca, Spain). Journal of Applied Geophysics 68, 479488. Bahr, K., 1988. Interpretation of the magnetotelluric impedance tensor: regional induction and local telluric distortion. Journal of Geophysics 62, 119127. Bahr, K., 1991. Geological noise in magnetotelluric data: a classication of distortion types. Physics of the Earth and Planetary Interiors 66 (12), 2438. Berchei, V., 2003. Hidrogeolga en la ciudad de Rosario de la Frontera y su relacin con la Cuenca del Ro Rosario. Facultad de Ciencias Naturales, Universidad de Salta. Berdichevsky, M.N., Dmitriev, V.I., 2008. Models and Methods of Magnetotellurics, rst ed. Springer, Berlin. Berktold, A., 1983. Electromagnetic studies in geothermal regions. Surveys in Geophysics 6, 173200. Bibby, H.M., Risk, G.F., Caldwell, T.G., Heise, W., 2009. Investigations of deep resistivity structures at the Wairakei geothermal eld. Geothermics 38 (1), 98107. Bossi, G., 1969. Geologa y estratigrafa del sector sur del Valle del Choromoro. Acta Geolgica Lilloana 10 (2), 1764. Constable, S.C., Parker, R.L., Constable, C.G., 1987. Occam's inversion: a practical algorithm for generating smooth models from electromagnetic sounding data. Geophysics 52, 289300. Cristallini, E., Comnguez, A.H., Ramos, V.A., 1997. Deep structure of the Metn-Guachipas region: tectonic inversion in Northwestern Argentina. Journal of South American Earth Sciences 10 (5), 403421. Galliski, M.A., Viramonte, J.G., 1988. The cretaceous paleorift in north-western Argentina: a petrological approach. South American Earth Science Bulletin 1 (4), 329342.

Gebhard, J.A., Gudici, A.R., Oliver Gascn, J., 1974. Geologa de la comarca entre el Ro Juramento y Arroyo Las Tortugas, provincias de Salta y Jujuy, Repblica Argentina. Revista de la Asociacin Geolgica Argentina 19 (3), 359375. Groom, R.W., Bailey, R.C., 1989. Decomposition of magnetotelluric impedance tensors in presence of local three-dimensional galvanic distortion. Journal of Geophysical Research 94 (B2), 19131925. Hoover, D.B., Frischknecht, F.C., Tippens, C., 1976. Audio-magnetotelluric soundings as a reconnaissance exploration technique in Long Valley, California. Journal of Geophysical Research 81, 801809. Iaffa, D.N., Sbat, F., Muoz, J.A., Mon, R., Gutierrez, A.A., 2011. The role of inherited structures in a foreland basin evolution. The Metn Basin in NW Argentina. Journal of Structural Geology 33, 18161828. Jones, A.G., Groom, R.W., 1993. Strike angle determination from the magnetotelluric tensor in the presence of noise and local distortion: rotate at your peril! Geophysical Journal International 113 (2), 524534. Jousset, P., Haberland, C., Bauer, K., Arnason, K., 2011. Hengill geothermal complex (Iceland) characterized by integrated geophysical observations. Geothermics 40 (1), 124. Lezaeta, P., 2002. Condence limit of the magnetotelluric phase sensitive skew. Earth Planets and Space 54 (5), 451457. Mackie, R., 2002. Users Manual and Software Documentation for Two-Dimensional Inversion of Magnetotelluric Data. GSY-USA, Inc., San Francisco. Mart, A., Queralt, P., Roca, E., 2004. Geoelectric dimensionality in complex geologic areas: application to the Spanish Betic Chain. Geophysical Journal International 157 (3), 961974. Mart, A., Queralt, P., Ledo, J., 2009. WALDIM: a code for the dimensionality analysis of magnetotelluric data using the rotational invariants of the magnetotelluric tensor. Computers and Geosciences 35 (12), 22952303. McNeice, G.W., Jones, A.G., 2001. Multisite, multifrequency tensor decomposition of magnetotelluric data. Geophysics 66, 158173. Mon, R., 2001. Estructuras curvadas y levantamientos verticales en la Cordillera Oriental (provincias de Salta y Tucumn). Revista de la Asociacin Geolgica Argentina 56 (3), 367376. Mon, R., Gutierrez, A., 2007. Estructura del extremo sur del sistema subandino (provincias de Salta, Santiago del Estero y Tucumn). Revista de la Asociacin Geolgica Argentina 62 (1), 6268. Monteiro Santos, F.A., Dupis, A., Andrade Afonso, A.R., Mendes-Victor, L.A., 1996. An audiomagnetotelluric survey over the Chaves geothermal eld (NE Portugal). Geothermics 25 (3), 389406. Moreno, J.A., 1970. Estratigrafa y paleografa del Cretcico superior en la cuenca del noroeste argentino, con especial mencin de los subgrupos Balbuena y Santa Brbara. Revista de la Asociacin Geolgica Argentina 25 (1), 944. Moreno Espelta, C., Viramonte, J., Arias, J., 1975. Geologa de rea termal de Rosario de la Frontera y sus posibilidades geotrmicas. I Congreso Iberoamericano de Geologa Econmica 4, 543560. Newman, G.A., Gasperikova, E., Hoversten, G.M., Wannamaker, P.E., 2008. Three-dimensional magnetotelluric characterization of the Coso geothermal eld. Geothermics 37 (4), 369399. Pesce, A., Miranda, F., 2003. Catlogo de manifestaciones termales de la Repblica Argentina. Servicio Geolgico Minero ArgentinoInstituto de Geologa y Recursos Minerales, Buenos Aires. Ramos, V.A., 1999. Rasgos estructurales del territorio argentino, Evolucin tectnica de la Argentina. In: Caminos, R. (Ed.), Geologa Argentina. Instituto de Geologa y Recursos Minerales, Buenos Aires, pp. 715784. Ramos, V.A., 2008. The basement of the Central Andes: the Arequipa and related terranes. Annual Review of Earth and Planetary Sciences 36, 289324. Reddy, I.K., Rankin, D., Phillips, R.J., 1977. Three dimensional modelling in magnetotelluric and magnetic variation sounding. Geophysical Journal of the Royal Astronomical Society 5 (3), 13325. Reyes, F.C., Salty, J.A., 1973. Consideraciones sobre la estratigrafa (subgrupo Pirgua) del noroeste argentino. V Congreso Geolgico Argentino 1, 1536. Rodi, W., Mackie, R.L., 2001. Nonlinear conjugate gradients algorithm for 2D magnetotelluric inversion. Geophysics 66 (1), 174187. Salty, J.A., Monaldi, C.R., 2006. Hoja Geolgica 2566-IV Metn, Escala 1:250.000, Provincia de Salta. Servicio Geolgico Minero Argentino 319, Buenos Aires. Sandberg, S.K., Hohmann, G.W., 1982. Controlled-source audio-magnetotellurics in geothermal exploration. Geophysics 47, 100116. Seggiaro, R., Aguilera, N., Ferretti, J., Gallardo, E., 1997. Estructura del rea Geotrmica de Rosario de la Frontera, Salta, Argentina. VIII Congreso Geolgico Chileno 1, 390394. Simpson, F., Bahr, K., 2005. Practical Magnetotellurics, rst ed. University press, Cambridge. Smith, J.T., 1995. Understanding telluric distortion matrices. Geophysical Journal International 122, 219226. Swift, C., 1967. A magnetotelluric investigation of an electrical conductivity anomaly in the southwestern United States. Ph.D. Thesis, MIT Press. Cambridge. Turner, J.C., 1959. Estratigrafa del cordn de Escaya y de la Sierra de Rinconada (Jujuy). Revista de la Asociacin Geolgica Argentina 13 (1), 1539. Viramonte, J.G., Kay, S.M., Becchio, R., Escayola, M., Novitski, I., 1999. Cretaceous rift related magmatism in Central-Western South America. Journal of South American Earth Science 12, 109121. Volpi, G., Manzella, A., Fiordelisi, A., 2003. Investigation of geothermal structures by magnetotellurics (MT): an example from the Mt. Amiata area, Italy. Geothermics 32 (2), 131145. Weaver, J.T., Agarwal, A.K., Lilley, F.E.M., 2000. Characterization of the magnetotelluric tensor in terms of its invariants. Geophysical Journal International 141 (2), 321336.