Case History Audiomagnetotelluric studies to trace the hydrological system of thermal uid ow of Bakreswar Hot Spring, Eastern India:

A case history

Rajib K. Sinharay1, Shalivahan Srivastava1, and Bimalendu B. Bhattacharya2

ABSTRACT
An audiomagnetotelluric AMT study has been carried out in the Bakreswar Hot Spring BHS area of eastern India to locate the geothermal source in the vicinity of BHS. Phasetensor analysis of the AMT data shows that the region is broadly 2D. Rapid relaxation inversion RRI for both transverse-electric TE and transverse-magnetic TM modes has been carried out to obtain resistivity images of the subsurface. AMT results show that the north-south fault close to Bakreswar is a shallow feature, not deeper than 300 m, and thus cannot act as a heat source. The subsurface formation below the fault zone is highly resistive up to a great depth, indicating the absence of a heat source and geothermal reservoir in the vicinity of the BHS. AMT results indicate that the location of the geothermal reservoir is deep and lies beyond the proles of measurement in the northwestern side of the Bakreswar Hot Spring.

INTRODUCTION
The Bakreswar geothermal region is located in Birbhum district, West Bengal, a state in eastern India. The main hot spring is located at Bakreswar 23 5248 N; 87 2240 E Figure 1. The heat ow of the area is very high, with values varying from 145 mW / m2 to 180 mW / m2 Shanker et al., 1991. Chandrasekharam 2000 shows that it is more than twice the average global heat ow and similar to the young oceanic ridges. The geothermal gradient near the hot spring is 90 C / km. An important feature of this group of hot springs is high helium content in water with values up to 1.2% Nagar et al., 1996, and up to 2% in air at 1 m above the ground, which

is far above the atmospheric background level, and the gas is released in periodic bursts Ghose et al., 2002. Helium is extracted at a few hot springs for commercial purposes. Although Bakreswar Hot Spring BHS is located over the Chotanagpur Gneissic Complex CGC, the mineral composition of BHS water does not resemble the water of granitic terrain. In contrast, the composition of Bakreswar spring water is considered to be of volcanic origin. Chowdhury et al. 1964 show that, geochemically, Bakreswar spring water differs sharply from the local groundwater and other water sources. Furthermore, they suggest that the high concentration of nitrogen and helium is derived from the primordial sources. Murty and Sinha 1991 carried out helium and neon isotope studies in the Bakreswar Hot Spring and found that the presence of helium in the BHS is caused by radiogenic sources. Another important observation is that the chemical composition of the spring water collected in premonsoon and postmonsoon periods is fairly similar, suggesting very little seasonal variation Murty and Sinha, 1991. Shanker et al. 1991 also report the marginal seasonal variation of the mineral constituents in thermal water. In addition to geologic surveys Mukhopadhyay, 1996; Nagar et al., 1996; Mukherjee and Majumdar, 1999, various geophysical studies, viz., self-potential SP, gravity, magnetic, resistivity, and audiomagnetotelluric Roy et al., 1985; Mukhopadhyay et al., 1986; Bhattacharya et al., 1992; Majumdar et al., 2000; Bhattacharya et al. 2002 surveys, and geochemical Chowdhury et al., 1964; Murty and Sinha, 1991; Shanker et al., 1991 and isotropic studies Murty and Sinha, 1991; Shanker et al., 1991 have been carried out over a very limited area close to the BHS. These studies could not identify any geothermal reservoir or heat source associated with the system. Providing electrical resistivity images of subsurface structures that control the geothermal uid ow is a critical component in understanding the hydrothermal properties of complex geothermal reservoirs. The resistivity in thermal areas is affected by the vertically ascending, hot mineralized waters or vapors that originate from the

1 Indian School of Mines University, Department of Applied Geophysics, Dhanbad, India. E-mail: rsrism@yahoo.com; svismagp@yahoo.co.in. 2

S. N. Bose National Centre for Basic Sciences, Salt Lake, Kolkata, India. E-mail: bimalendu@bose.res.in.

contact between groundwater and high-temperature intrusive magmas. The intrusions themselves have a very low intrinsic resistivity at temperatures above approximately 800 C Bartel and Jacobson, 1987. Resistivity in geothermal areas is also governed, to a great extent, by the presence of hydrothermal alteration products such as clays, silica, or travertine. Pellerin et al. 1996 show that the main electromagnetic EM anomaly related to the reservoir is caused by the presence of electric charge at resistivity boundaries rather than EM induction. Thus, magnetotelluric MT is more effective in dening a conductive reservoir overlain by a larger and more conductive clay cap than other EM methods that measure only magnetic elds. Electric resistivity is a primary physical property of the earth that is strongly inuenced by hydrothermal processes present in geothermal reservoirs. If mapped, resistivity can be used to infer untapped fracture systems and regions of increased permeability and uid content, as well as conductive alteration minerals clays, etc. created by intensive hydrothermal water circulation in the reservoir. The natural-eld MT method has proved very useful for mapping near-surface low-resistivity zones caused by rock alteration and saline pore uids in geothermal areas e.g., Hoover et al., 1976; Hoover et al., 1978; Long and Kaufmann, 1980; Singh and Nabetani, 1995; Uchida, 1995; Romo et al., 1997; Lagios et al., 1998; Bai et al., 2001; Correia and Safanda, 2002; de Lugao et al., 2002; PerezFlores and Schultz, 2002; Risk et al., 2002; Wannamaker et al., 2002; Volpi et al., 2003; Harinarayana et al., 2004, 2006; Oskooi et al., 2005. Wannamaker 1997a, 1997b shows the advantages of having long-period MT data combined with controlled-source audiomagne8720E 8725E

totelluric CSAMT data in a reconnaissance survey at the Sulphur Springs thermal area in the USA. The Puga geothermal eld in Ladakh Himalayas, India has been studied using the MT method Singh and Nabetani, 1995; Harinarayana et al., 2004, 2006. We carried out an MT survey to determine the subsurface resistivity structure of the Bakreswar geothermal province to establish and delineate the location of the geothermal reservoir within the area.

GEOLOGY
The BHS, located in Precambrian CGC, is surrounded by alluvial and laterite soil with sporadic exposures of basement rocks Figure 1. The geothermal province lies in the West Bengal Basin. The basin has been developed along the margin of eastern Indian shield due to Gondwana rifting. This underlies a Gondwana trough covered by Rajmahal basalts of Jurassic age and younger sediments Ghose et al., 2002. Granite gneiss is the dominant basement rock over the area. The gneiss is regionally folded and has formed a northwestsoutheast-trending major anticlinal structure. Dolerite dikes appear mostly parallel to the axial plane of the fold. Four sets of fractures trending northwest-southeast, northeast-southwest, north-south, and east-northeastwest-southwest intersect each other near the hot spring Majumdar et al., 2000. The hot spring occurs along the fractures associated with faulted contacts of the Gondwana sediments in the coal belt of Birbhum which is aligned north-south and the Raniganj Jharia coaleld aligned east-west. The north-south-trending fault associated with the hot spring is the most important structural feature in this area. This fault zone is marked by 1.2-km-long and 50 -m-wide brecciated and mylonitic quartzo-felspathic rocks exposed at 1.5 km north of the hot spring, but it disappears under the alluvial terrain in the vicinity of the BHS.

23 55 N

PREVIOUS GEOPHYSICAL STUDIES

AEW2 AEW1 101 002 001 102 AEW4 109 110 011 108 008
23 55 N

003 103

BAKRESWAR 007 000 107 005 105 104

Delhi Study Area

004

23 50 N 8720E

Scale

2 km
8725E

23 50 N
Lithological boundary (Inferred) Granite gneiss with minor inclusion of amphibolite AMT profiles AMT sounding Hot spring

Alluvial/laterite soil (Quaternary) Brecciated sillicified zone Dolerite/metadolerite Pegmatite

Fault (Buried)

Figure 1. Location and geological map of Bakreswar geothermal province after Majumdar et al., 2000 and AMT proles.

Roy et al. 1985 rst carried out self-potential SP, gravity, magnetic, and electrical resistivity surveys in the close vicinity of the BHS. The self-potential survey did not show any signicant anomaly over the area. The gravity survey revealed that the BHS is located over the axis of a north-northwestsouth-southeast trending gravity low. In the west part of the BHS, a gravity high of about 1 mGal is observed. Bhattacharya et al. 1992 conducted detailed gravity and electrical resistivity surveys, as well as vertical magnetic-eld surveys by uxgate magnetometers, over the BHS area. Both the gravity and magnetic lows over the hot spring have been interpreted as a fractured zone. Vertical electrical sounding VES over the area showed that this fractured zone is saturated with hot water down to a depth of about 120 m from the surface Bhattacharya et al., 1992. Nagar et al. 1996 obtain similar results. They interpret gravity and magnetic lows in terms of a fault buried under the alluvial cover. Bhattacharya et al. 1992 conclude that, except for a localized low over the fault, the magnetic anomaly of the area is not signicant as reported earlier by Roy et al. 1985. Majumdar et al. 2000 carried out Wenner resistivity proles and VES using the Schlumberger array. The results indicate southward continuation of the north-south striking fault under the alluvial cover up to BHS. The fault is providing passage of the thermal uid at Bakreswar Figure 1. Another important outcome of their work is the presence of intrusive dolerite encountered at a depth of even 5 m in granite gneissic terrain in two out of four shallow boreholes drilled to a depth of 40 m.

The geophysical surveys carried out earlier delineated the northsouth fault and suggested that this fault is acting as a conduit. However, they could not identify the location of the geothermal reservoir.

AUDIOMAGNETOTELLURIC DATA ACQUISITION AND ANALYSIS

The region around Bakreswar is populated and some places are not available for AMT measurements because of permanent structures, cultural noise, and other disturbances. The station spacing in this area varied from 250 m to 750 m Figure 1. Twenty-two frequencies were used for AMT data acquisition ranging from 11 12 7.5 Hz to 10,000 Hz. The time-series data were recorded for 5 to 7 hours at each AMT site using the Phoenix V5 system. Eigh 21 22 teen AMT soundings were conducted in an area of about 6 km2. HorX22Y 11 X12Y 21 X22Y 12 X12Y 22 1 izontal electric- and magnetic-eld components were recorded si , 1 multaneously, along with the vertical component of the magnetic detX X11Y 21 X21Y 11 X11Y 22 X21Y 12 eld. Data processing Phoenix, 1996 was performed using singlesite processing. The time series data was recorded with sampling rates of 25,000 Hz, 12,880 Hz, 3072 Hz, 100,000 90 and 384 Hz at 2 frequency/octave with antialias Line AEW1, site 001 Line AEW1, site 001 75 lters at 12.3 KHz, 5 KHz, 840 Hz, and 62 Hz, 10,000 60 respectively. The number of stacks was xed at 1000 45 20. The data were saved to records of 1024 sam30 100 ple points per channel with a 64 byte time tag 15 stamp. The recorded time series data is converted 10 0 1 10,000 1000 100 10 10,000 1000 100 10 1 to the frequency domain by applying a 32-point discrete Fourier transform DFT. Gain factors were applied to normalize the data back to the input values in terms of mV/km and nT. Larger peri10,000 90 Line AEW1, site 004 Line AEW1, site 004 75 ods are attained through a process of decimation 60 1000 which involves subsampling and low-pass lter45 ing the data to remove periods shorter than the 100 30 Nyquist period of the decimated data Simpson 15 and Bahr, 2005. The calculated DFT values for 0 10 each channel at each frequency are used to create 1 10,000 1000 100 10 1 10,000 1000 100 10 a cross-power matrix. This matrix is then used to accumulate the sum of the data along with the sum of the weights which together are used to cal10,000 90 Line AEW2, site 005 Line AEW2, site 005 culate the weighted average. From the weighted 75 60 1000 average data, the AMT parameters are calculated 45 using standard formulas to get apparent resistivi100 30 ty and phase as well as other geophysical charac15 teristics. 0 10 The apparent resistivity and phase plots for 10,000 1000 100 10 1 10,000 1000 100 10 1 some of the sites are shown in Figure 2. Some of the curves have outliers in either apparent resistivity or phase or both. These outliers are mainly 10,000 90 Line AEW4, site 107 at 60 Hz or its harmonics. This could be a result of 75 Line AEW4, site 107 unstabilized power supply. These outliers were 60 1000 45 not included in the modeling. For some of the 100 30 sites, in general close to the BHS, the phase data 15 lies outside the range of 090; i.e., phase data are 0 10 out-of-quadrant. However, these are extreme cas10,000 1000 100 10 1 10,000 1000 100 10 1 es of resistivity distributions Egbert, 1990; LazaFrequency (Hz) Frequency (Hz) eta and Haak, 2003. Out-of-quadrant phase data have been modeled using anisotropic layering in Figure 2. Unrotated apparent resistivity and phase data for representative sites of the the presence of large resistivity variations Heise Lines AEW1, AEW2, and AEW4; site 001, site 004, site 005, and site 107 open symbols and Pous, 2003. However, for our analysis, we xy, solid symbols yx; x and y correspond to directions north and east, respectively.

have not considered those data sets. The apparent resistivity for most sites indicates the problem of galvanic distortion as most site data are static-shifted. Geothermal environments are well-known to have static shift problems. We tackle the regional undistorted impedance tensor following Caldwell et al. 2004 and Bibby et al. 2005. They have shown that the measured and regional phase tensors are identical, independent of galvanic distortion, and simply expressed in terms of the observed impedance tensor components. The phase tensor in terms of real and imaginary components of Z X iY in a Cartesian coordinate system x1,x2 Figure 3a can be written as Caldwell et al., 2004; Bibby et al., 2005

Apparent Resistivity (ohm-m)

Apparent Resistivity (ohm-m)

Apparent Resistivity (ohm-m)

Apparent Resistivity (ohm-m)

Phase (degrees)

Phase (degrees)

Phase (degrees)

Phase (degrees)

where

detX X11X22 X21X12 .

and the coordinate axes, but the shape and geographic position of the ellipse will be unchanged. Equation 1 can be further written as

The phase tensor is graphically represented as an ellipse. The semimajor and semiminor axes of the ellipse are given by max,, and min,, respectively, and the orientation of the major axis is given by , which is the strike. The angle denes the position of a reference axis relative to the coordinate axes and denes the position of the principal axis of the ellipse relative to the reference axis. Rotating the coordinate axes will change the angle between the reference axis
xis

cos 2

sin 2

sin 2 cos 2

cos 2

sin 2

sin 2 cos 2

a)
fer

1 1 11 222 12 2121/2, 2 1 2 11 222 12 2121/2, 2 1 12 21 tan1 , 2 11 22

x1

x2

en

ce a

Re

m
in

max

and

1 12 21 tan1 . 2 11 22
Bibby et al. 2005 dene a nondimensional quantity:

b)

3 2 1 0 10,000 1000 100 10 1

max min . max min

Frequency (Hz)

c)
(degrees)

20 0 20 40 10,000

1000

100

10

Frequency (Hz)

d)
Strike (degrees)

80 40 0

40 80 10,000 1000 100 10 1

Frequency (Hz)

Figure 3. a Graphical representation of a phase tensor as an ellipse. b The phase tensor ellipticity, for site 001 of the AMT sounding shown in Figure 2. c The phase tensor skew angle for site 001 of theAMT sounding shown in Figure 2. d The strike of the major axis of the phase tensor ellipse for site 001 of the AMT sounding shown in Figure 2.

Caldwell et al. 2004 and Bibby et al. 2005 dene the dimensionality of the feature with the parameters and , respectively. Ideally, for 1D and 2D structures, 0 and 0; 0, respectively. A nonzero denes the feature as 3D. There are two conditions for a 2D earth and both should be satised simultaneously, i.e., 0 and 0 for ideal cases. We show a phase tensor analysis of the sounding 001 Figure 2. Considering that the noise is present in the site, we have considered 0.15 and lying within 3.0 to be equivalent to zero for a 2D indicator. Ideally, should be positive and less than one. In extreme situations, this could be greater than one or negative. This happen when either max or min is negative or, equivalently, when one of the phase angles, tan1 max or tan1 min, lies outside the range of 090 Bibby et al., 2005. It is seen that is greater than one with our data sets at two frequencies, i.e., 60 Hz and 7500 Hz a multiple of 60 Hz Figure 3b. It could again be related to an unstabilized power supply in the area and/or other noise sources. On the basis of Figure 3b and Figure 3c, the 2D structure is for frequencies from 960 to 3750 Hz and from 15 to 80 Hz. A nonzero for the rest of the frequencies indicates that the subsurface feature is 3D for such frequencies. The near constant azimuth of the phase tensor strike direction 40 45, Figure 3d determines the strike direction of the 2D structure. This direction coincides with the major regional geologic strike, which is a northwestsoutheast trending anticlinal structure. The invariants det D and trace D where D is the distortion help in determining the site gain or static shift. Distortion is usually described in terms of the inuence of small localized conductivity heterogeneities, which alter the direction and magnitude of the electric eld at the measurement site Bibby et al., 2005. Bibby et al. 2005 express the principal component of the regional impedance tensor parallel X and perpendicu-

 T detX , X 12 and X 21 are the components of the real part in the rotated coordinate system, det D P, and trace D T. For the 2D structure, det D and trace D are chosen in such a manner that they ensure that S2 is positive at all the frequencies in the selected range. For the analysis, the values of the invariants are det D 1 and trace D 2.1, respectively. These values ensure that S2 is positive at all the frequencies in the selected range. Figure 4 shows elliptical representations of the phase tensor for every period for a representative sounding station 001. Apparent resistivities and phases can be computed from the along-strike electric currents transverse electric TE and from the across-strike electric currents transverse magnetic TM. Obviously, the obtained apparent resistivity and phase from the two modes are different for the 2D structure. Berdichevsky et al. 1998 study the sensitivity of these two modes on different aspects of the subsurface features. The TE mode is the most sensitive to along-strike conductors, whereas the TM mode is the most sensitive to shallow lateral-resistivity variations.
2

lar X to the strike as X TS and X TS , where S2

4 PX 12X 21

2X 12

2X 21

APPROXIMATE 2D INVERSION
The rapid relaxation inversion RRI scheme Smith and Booker, 1991 has been used for 2D modeling. RRI minimizes a combination of the t to the data and is a measure of lateral and vertical smoothness of the model. This is an iterative scheme that takes into account the apparent resistivity and phase of both the TE and TM modes of the data during inversion. The distortion correction was applied using phase tensor analysis and the impedance tensors were rotated in the obtained regional strike direction. Prior to inversion, the TE and TM data are weighted by their associated error so that undue weight is not given to poor quality data. The data are inverted using leastsquares iterations and subsequently Huber weights are applied. Redescending weights are then employed to down weight data which are more severely affected by the large outliers. In the RRI, one requires the value of the alpha parameter which scales the horizontal versus vertical derivatives in the norm. Horizontal smoothening increases with the increase in the value of alpha. It also depends upon the site spacing. Following Harinarayana et al. 2006, alpha has been decided by monitoring the convergence of the data mist and slowness of the model roughness. Smith and Booker 1991 dene the model roughness R as Cm 2, where C is a roughening matrix and m is logarithms of block resistivities m. These two criteria help in nding the optimal value of alpha by systematic trial. The optimal value of alpha used in the analysis is eight. Inversion involves nding a model that locally minimizes or maximizes a scalar cost function or tness that normally is a measure of discrepancy or correlation d between the observed dobs and the estimated dest data. The 2D models are parameterized on a grid consisting of 100 horizontal and 80 vertical nodes. RRI models have been constructed by 40 least-square iterations, three robust Huber weights, two redescending weights, and 45 smoothing iterations. The overall root mean square rms mist between observed data and predicted response of the 2D models are presented in Figures 5a, 6a, and 7a. To ascertain that the obtained models are not caused by inversion artifacts, several inversions were executed with different half-space models as well as pseudosection starting models. The features com-

mon to all the models have nally been considered to t the observed data Bai et al., 2001. Different starting models brought the identical major features. Overall rms mist is presented in Figure 5a for traverse AEW1. TE phase shows a larger mist at all the stations except site 002. Site 103 also shows a larger mist for TM apparent resistivity. Normalized residuals of apparent resistivity and phase are shown in Figure 5b for both the TE and TM modes for the frequencies over which 2D inversion has been carried out. The 2D inverted model along AEW1 Figure 5c shows a zone of resistivity less than 500 ohm-m down to a depth of 200 m above a resistive basement and does not provide any conduit for thermal uid ow. Figure 6a shows a good t for apparent resistivity and phase in both the modes for prole AEW2. Normalized residuals of apparent resistivity and phase are shown in Figure 6b for both the TE and TM modes for the frequencies over which 2D inversion has been carried out. The 2D inverted model for prole AEW2 Figure 6c shows a shallow conducting zone down to a depth of 100 m in the northwestern part of the prole which becomes deeper in the southeastern part, extending down to a depth of 300 m. The resistive basement is sheared along this prole and a relatively low resistivity zone is mapped in the northwestern part of it below the depth of about 2 km. The compact rock strata looks faulted in the east of the BHS and the eastern part of the fault is downthrown side. Thick alluvial cover is observed in the eastern part which is downthrown side of the fault, whereas in the western part, the rocks are exposed but weathered. Overall rms mist is reasonably good Figure 7a for traverse AEW4 for all the modes except for TE phase. Normalized residuals of apparent resistivity and phase are shown in Figure 7b for both the TE and TM modes for the frequencies over which 2D inversion has been carried out. The 2D inverted model for prole AEW4 Figure 7c maps a near-surface low-resistive stratum of resistivity less than 500 ohm-m down to a depth of 300 m. This zone overlies a broadly resistive basement. The deeper resistive part in the 2D model is not providing any passage of geothermal uids. However, in the northwestern part of this prole, decreased resistivity is obtained in deeper zones. This result makes the northwestern part more interesting because of the possibility of existence of a low-resistivity zone possibly serving as a conduit of geothermal uid. Proles AEW1 and AEW2 are mostly over granite gneiss exposure, whereas prole AEW4 is over alluvium/laterite soil Figure 1. The top of the relatively compact granite gneiss 2500 ohm-m is

min
1 4

Log10 Frequency (Hz)

Figure 4. Phase tensor ellipses plotted for every data point for site 001 of the AMT sounding shown in Figure 2. A radius of unity corresponds to a phase of 45 indicative of 1D structure.

300 400 m for AEW1 and AEW2 and approximately 600 m below AEW4. Site 101 AEW1 is over relatively unweathered granite gneiss exposure. The signature of the vertical fault is seen in AEW2 but not in AEW4 because it is over the alluvium/laterite soil cover. Berdichevsky et al. 1998 discuss the relative advantages and disadvantages of the TM and the TE mode data to map geologic structures at different depths with different resistivity. They conclude that the TM mode is more sensitive to near-surface structures while the TE mode may be more sensitive to deep structures, and that the TM mode is more susceptible to static shifts whereas the TE mode may be almost undistorted. They also showed that the TM mode is more robust to the 3D effects caused by conductive struc-

tures and the TE mode is more robust to the 3D effects caused by resistive structures. For lines AEW1 and AEW4, a large mist is seen for the TE mode as compared to the TM mode. It implies that the deeper resistivity structures in these two proles are not mapped reasonably well. Supposedly, the 2D structure differs even for the sites only a few hundred meters apart e.g., 103 and 000. This may be attributed to the fact that there are 3D conducting structures. The hydrothermal conceptual model could be explained by the vapors originating from the contact between groundwater and high temperature intrusive magmas or by hydrothermal alteration such as silica. Given the generally high resistivity detected by the survey, there is apparently little smectite clay alteration. For example, at MT

a)
Overall rms

8 6 4 2 0 101 002 001 003 103 004

Site number

b)
Normalized residuals for TEa
8 4 0 4 8 12 1000 10 1000

c)
0

101

002

001

003

103

004

Resistivity (Ohm-m)
200 500 1000

Frequency (Hz) Normalized residuals for TE

30 20 10 0 10
1000 10

Depth (m)

2500 5000 7500

2000

Frequency (Hz) Normalized residuals for TMa

4 0 4 8
1000 10 3000 NW 0 200 400 600 800 m SE

Frequency (Hz) Normalized residuals for TM

16 12 8 4 0 4 8 1000 10

Frequency (Hz)

Figure 5. a Overall rms mist for 2D model AEW1 for TM and TE resistivity and phases: b TE resistivity, TE phase, TM resistivity, TM phase. b Normalized residuals for TE and TM modes of apparent resistivity a and phase along traverseAEW1: b site 101, site 002, site 001, site 003, site 103, site 004. c 2D RRI along traverse AEW1.

sites 004 and 107, weak clay alteration might extend from the surface to a depth of about 20 m. However, the silica concentration ranges between 70 and 100 ppm Shanker et al., 1991. Silica sinter deposition implies subsurface temperature range of 160 to 180 C if it is still actively depositing silica. Sinter and silicied rocks tend to be resistive, even when hot. Certainly, the hydrothermal alteration such as clay could not explain the resistivity structure. A hydrothermal system would have to have unusually high fracture porosity over a large volume to have a relatively low resistivity zone below

1000-m depth. Such a system would necessarily create a large thermal anomaly. A very effective cap would be required to avoid having widespread thermal manifestations. The resistivity data indicate that the conductor is a thin zone that outcrops at the surface. The limited area of the hot springs and the thin and poorly developed clay cap detected by the shallow MT images is unlikely to explain the smectiteillite clay-oriented interpretation model. Very little surface geology or alteration analyses conrm the pattern of smectite clay alteration that might be associated with the shallow MT interpretation.

a)
Overall rms

2.8 2.4 2 1.6 1.2 110 011 102 000 005 105 104

Site number

c)
0

110

011

102

000 BHS

005

105 104

b)
Normalized residuals for TEa
8 4 0 4 1000 10 1000 200 500 1000 2500 5000 7500 2000 1000 10

Resistivity (ohm-m)

Frequency (Hz) Normalized residuals for TE Depth (m) Frequency (Hz) Normalized residuals for TMa
3000 4 1000 10 NW 0 200 400 600 800 1000 1200 1400 m SE

8 4 0 4

4 0

Frequency (Hz) Normalized residuals for TM

4 0 4 1000 10

Frequency (Hz)

Figure 6. a Overall rms mist for 2D model AEW2 for TM and TE resistivity and phases: b TE resistivity, TE phase, TM resistivity, TM phase. b Normalized residuals for TE and TM modes of apparent resistivity a and phase along traverseAEW2: b site 110, site 011, site 102, site 000, site 005, site 105, site 104. c 2D RRI along traverse AEW2.

a)
Overall rms

6 4 2 0 109 108 008 007 107

Site number

b)
Normalized residuals for TEa
4 2 0 2 4 6 8 1000 10

c)
0

109

108

008

007

107

Resistivity (Ohm-m)
1000 200 500 1000 2500 2000 5000 7500

Frequency (Hz)

Normalized residuals for TE

10 5 0 5 10 15 20 1000 10

Frequency (Hz)
2 0 2 4 1000 10 3000 NW 0 200 400 600 800 1000 m SE

Normalized residuals for TMa

Frequency (Hz)

Normalized residuals for TM

12 8 4 0 4 8 12 1000 10

Frequency (Hz)

Figure 7. a Overall rms mist for 2D model AEW4 for TM and TE resistivity and phases: b TE resistivity, TE phase, TM resistivity, TM phase. b Normalized residuals for TE and TM modes of apparent resistivity a and phase along traverseAEW4: b site 109, site 108, site 008, site 007, site 107. c 2D RRI along traverse AEW4.

Depth (m)

CONCLUSIONS
The AMT results demarcate the shallow nature of the north-south fault near Bakreswar, rejecting its possibility to act as a source of hydrothermal uid that is in agreement with earlier geophysical results. Thus, the exposed fault can act only as a conduit of thermal water. But interestingly, the AMT sections indicate the possible geothermal location that was hitherto not seen in the earlier geophysical results. The sections along both AEW2 and AEW4 indicate the possibility of existence of a deeper source or reservoir in the northwestern part of the survey area. The results further indicate that the geothermal uid is traveling a longer path from the northwestern side. Thus, the work requires a further extension on the western and northwestern side to

establish and delineate the location of the geothermal reservoir in the area. The hydrothermal conceptual model could be explained by the vapors originating from the contact between groundwater and hightemperature intrusive magmas or by hydrothermal alteration of silica. The models presented here are approximate 2D images of the otherwise complex 3D geothermal terrain and provide insight into the possible geothermal reservoir in the area which was not possible with the earlier geophysical work.

REFERENCES
Bai, D., M. A. Meju, and Z. Liao, 2001, Magnetotelluric images of deep crustal structure of the Rehai geothermal eld near Tengchong, southern China: Geophysical Journal International, 147, 677687. Bartel, L. C., and R. D. Jacobson, 1987, Results of a controlled-source audiofrequency magnetotelluric survey at the Puhimau thermal area, Kilauea Volcano, Hawaii: Geophysics, 52, 665677. Berdichevsky, M. N., V. I. Dmitriev, and E. E. Pozdnjakova, 1998, On twodimensional interpretation of magnetotelluric soundings: Geophysical Journal International, 133, 585606. Bhattacharya, B. B., R. K. Sinharay, and Shalivahan, 2002, Audiomagnetotelluric AMT investigations for 2D electrical conductivity modeling over geothermal province of Bakreswar, eastern India: 72nd Annual International Meeting, SEG, Expanded Abstracts, 488491. Bhattacharya, R. S., G. K. Dutta, and P. R. Rao, 1992, Report on geophysical mapping around Bakreswar-Tantloi group of hot springs, Birbhum district, West Bengal: Report unpublished, Geological Survey of India. Bibby, H. M., T. G. Caldwell, and C. Brown, 2005, Determinable and nondeterminable parameters of galvanic distortion in magnetotellurics: Geophysical Journal International, 163, 915930. Caldwell, T. G., H. M. Bibby, and C. Brown, 2004, The magnetotelluric phase tensor: Geophysical Journal International, 158, 457469. Chandrasekharam, D., 2000, Geothermal energy resources of India: Proceedings of World Geothermal Energy Congress 2000, 133145. Chowdhury, A. N., B. B. Bose, and G. Banerjee, 1964, Studies on geochemistry of thermal spring at Bakreswar: Report of the 22nd session, International Geological Congress, Part 12, 143160. Correia, A., and J. Safanda, 2002, Geothermal modeling along a two-dimensional crustal prole in southern Portugal: Journal of Geodynamics, 34, 4761. de Lugao, P. P., E. F. LaTerra, B. Kriegshauser, and S. L. Fontes, 2002, Magnetotelluric studies of the Caldas Novas geothermal reservoir, Brazil: Journal of Applied Geophysics, 49, 3346. Egbert, G. D., 1990, Comments on Concerning dispersion relations for the impedance tensor, E. Yee, and K. V. Paulson, authors: Geophysical Journal International, 102, 18. Ghose, D., D. P. Chowdhury, and B. Sinha, 2002, Large-scale helium escape from earth surface around Bakreswar-Tantloi geothermal area in Birbhum district, West Bengal, and Dumka district, Jharkhand, India: Current Science, 82, 993996. Harinarayana, T., K. K. Abdul Azeez, K. Naganjaneyulu, C. Manoj, K. Veeraswamy, D. N. Murthy, and S. P. E. Rao, 2004, Magnetotelluric studies in Puga valley geothermal eld, NW Himalaya, Jammu and Kashmir, India: Journal of Volcanology and Geothermal Research, 138, 405424. Harinarayana, T., K. K. Abdul Azeez, D. N. Murthy, K. Veeraswamy, S. P. E. Rao, C. Manoj, and K. Naganjaneyulu, 2006, Exploration of geothermal structure in Puga geothermal eld, Ladakh Himalayas, India by magnetotelluric studies: Journal of Applied Geophysics, 58, 280295. Heise, W., and J. Pous, 2003, Anomalous phases exceeding 90 in magnetotellurics: Anisotropic model studies and a eld example: Geophysical Journal International, 155, 308318. Hoover, D. B., F. C. Frischknecht, and C. Tippens, 1976, Audiomagnetotelluric sounding as reconnaissance exploration technique in Long Valley, California: Journal of Geophysical Research, 81, 801809. Hoover, D. B., C. L. Long, and R. M. Sentert, 1978, Some results from audiomagnetotelluric investigations in geothermal areas: Geophysics, 43, 15011514. Lagios, E., D. Galanopoulos, B. A. Hobbs, and G. J. K. Dawes, 1998, Twodimensional magnetotelluric modelling of the Kos Island geothermal region Greece: Tectonophysics, 287, 157172. Lezaeta, P., and V. Haak, 2003, Beyond magnetotelluric decomposition: Induction, current channeling, and magnetotelluric phases over 90: Journal of Geophysical Research, 108, 2305, doi: 10.1029/2001JB000990. Long, C. L., and H. E. Kaufmann, 1980, Reconnaissance geophysics of a known geothermal resource area, Weiser, Idaho, and Vale, Oregon: Geo-

physics, 45, 312322. Majumdar, R. K., N. Majumdar, and A. L. Mukherjee, 2000, Geoelectric investigations in Bakreswar geothermal area, West Bengal, India: Journal of Applied Geophysics, 45, 187202. Mukherjee, A. L., and R. K. Majumdar, 1999, Hydrology and convective heat ow of Bakreswar thermal spring area, Birbhum district, West Bengal in Proceedings of Golden Jubilee Seminar on Exploration Geophysics in India: Geological Survey of India, Special Publication 49, 219228. Mukhopadhyay, D. K., 1996, Geothermal parameters and thermal energy potential of Bakreswar, Tantloie group of hot springs in Birbhum district, West Bengal and Dumka district, Bihar, in U. L. Pitale and R. N. Padhi, eds., Geothermal energy in India: Geological Survey of India, Special Publication 45, 8798. Mukhopadhyay, M., R. K. Verma, and M. H. Ashraf, 1986, Gravity eld and structures of the Rajmahal Hills: Example of the Paleo-Mesozoic continental margin in eastern India: Tectonophysics, 131, 353367. Murty, S. V. S., and B. Sinha, 1991, Helium and neon isotopes in Bakreswar Hot Spring: Presented at the 5th National Symposium on Mass Spectrometry. Nagar, R. K., G. Vishwanathan, S. Sagar, and A. Sankaranarayanan, 1996, Geological, geophysical and geochemical investigations in BakreshwarTantloi thermal eld, Birbhum and Santhal Parganas districts, West Bengal and Bihar, India, in U. L. Pitale and R. N. Padhi, eds., Geothermal energy in India: Geological Survey of India, Special Publication 45, 349360. Oskooi, B., L. B. Pedersen, M. Smirnov, K. Arnason, H. Eysteinsson, A. Manzella, and the DGP Working Group, 2005, The deep geothermal structure of the Mid-Atlantic Ridge deduced from MT data in SW Iceland: Physics of the Earth and Planetary Interiors, 150, 183195. Pellerin, L., J. M. Johnston, and G. W. Hohmann, 1996, A numerical evaluation of electromagnetic methods in geothermal exploration: Geophysics, 61, 121130. Perez-Flores, M. A., and A. Schultz, 2002, Application of 2-D inversion with genetic algorithms to magnetotelluric data from geothermal areas: Earth, Planets and Space, 54, 607616. Phoenix, 1996, Operating manual of V5-16. Risk, G. F., H. M. Bibby, C. J. Bromley, T. G. Caldwell, and S. L. Bennie, 2002, Appraisal of the Tokaanu-Waihi geothermal eld and its relationship with the Tongariro geothermal eld, New Zealand: Geothermics, 31, 4568. Romo, J. M., C. Flores, R. Vega, R. Vazquez, M. A. Prez Flores, E. G. Trevio, F. J. Esparza, J. E. Quijano, and V. H. Garca, 1997, A closelyspaced magnetotelluric study of the Ahuachapn-Chipilapa geothermal eld, El Salvador: Geothermics, 26, 627656. Roy, A. K., A. Bagchi, P. Chatterjee, A. Mukherjee, and C. V. Brahman, 1985, Reconnaissance geophysical investigation for locating source of geothermal energy in Bakreswar area, Birbhum district, West Bengal: Unpublished report, Geological Survey of India. Shanker, R., S. K. Guha, N. N. Seth, A. Ghosh, S. Ghosh, D. R. Nandy, B. L. Jangi, and K. Mathuraman, eds., 1991, Geothermal Atlas of India: Geological Survey of India, Special Publication 19, 110113. Simpson, F., and K. Bahr, 2005, Practical magnetotellurics: Cambridge University Press. Singh, R. P., and S. Nabetani, 1995, Resistivity structure of Puga geothermal eld: Proceedings of the World Geothermal Congress, 2, 887892. Smith, J. T., and J. R. Booker, 1991, Rapid inversion of two- and three-dimensional magnetotelluric data: Journal of Geophysical Research, 96, 39053922. Uchida, T., 1995, Resistivity structure of Sumikawa geothermal eld, northeastern Japan, obtained from magnetotelluric data: Proceedings of the World Geothermal Congress 1995, 921925. Volpi, G., A. Manzella, and A. Fiordelisi, 2003, Investigation of geothermal structures by magnetotellurics MT: An example from the Mt. Amiata area, Italy: Geothermics, 32, 131145. Wannamaker, P. E., 1997a, Tensor CSAMT survey over the Sulphur Springs thermal area, Valles Caldera, New Mexico, United States of America, Part I: Implications for structure of the western caldera: Geophysics, 62, 451465. , 1997b, Tensor CSAMT survey over the Sulphur Springs thermal area, Valles Caldera, New Mexico, United States of America, Part II: Implications for CSAMT methodology: Geophysics, 62, 466476. Wannamaker, P. E., G. R. Jiracek, J. A. Stodt, T. G. Caldwell, V. M. Gonzales, J. D. McKnight, and A. D. Porter, 2002, Fluid generation and pathways beneath an active compressional orogen, the New Zealand Southern Alps, inferred from magnetotelluric data: Journal of Geophysical Research, 107, 2117, doi: 10.1029/2001JB000186.