A broadband tensorial magnetotelluric study in the Travale geothermal field

0375- 6505/85 $3.00 + 0.00 Pergamon Press Ltd. © 1985 CNR.

Geothermics, Vol. 14, No. 5/6, pp. 645- 652, 1985. Printed in Great Britain.

A BROADBAND TENSORIAL THE TRAVALE

MAGNETOTELLURIC GEOTHERMAL FIELD

STUDY

IN

V. R. S. H U T T O N , G. J, K. D A W E S , T. D E V L I N a n d R. R O B E R T S

Department of Geophysics, University of Edinburgh, Edinburgh, U.K. Abstract--As a contribution to the EEC study of the potential contribution of electric and electromagnetic techniques to geothermal exploration, magnetotelluric studies have been undertaken with a sounding bandwidth ranging from 2 to 7 decades of period at more than 30 sites within the chosen test area of Travale. This area must be one of the most unfavourable for the application of electrical techniques on account both of the thickness (up to 2 kin) of conducting (< 1 ohm.m in some locations) cover formations and of the intensity of the artificial disturbances from local power stations and distribution lines. Nevertheless it has been possible to obtain good quality data over part of the sounding band employing an automatic in-field analysis system and rigorous data analysis and to penetrate to reservoir depths at the centre of the graben by undertaking broadband soundings (up to 10' s) at some sites. For interpretation of the data for periods up to about 100 s, 2-D modelling is both satisfactory and essential (I-D modelling provides correct layer resistivitiesbut underestimates interface depths) and good agreement has been obtained for an electrical structure model and the relevant geological section. The 2-D models, which best fit the long period data, are characterised both by zones of highly conducting flysch cover formations and by an anomalously conducting basement. Restriction of the study to a test area within the Travale graben inhibits the unequivocal association of these conducting zones with the thermal anomaly. INTRODUCTION T h e T r a v a l e g e o t h e r m a l field, in which m o r e t h a n 50 boreholes have been drilled ,And a variety o f geophysical a n d geochemical studies have previously been u n d e r t a k e n , was chosen by the E E C as a suitable area for testing the p o t e n t i a l c o n t r i b u t i o n o f electric a n d electromagnetic i n d u c t i o n techniques to g e o t h e r m a l e x p l o r a t i o n . T h e E d i n b u r g h university g r o u p p a r t i c i p a t e d in this p r o g r a m m e by u n d e r t a k i n g m a g n e t o t e l l u r i c studies at m o r e t h a n 30 sites a n d g e o m a g n e t i c deep s o u n d i n g at 25 sites within the test area. T h e s o u n d i n g b a n d w i d t h ranged f r o m 2 to 7 decades o f period, with the long period o b s e r v a t i o n s being u n d e r t a k e n at eight o f the sites in c o l l a b o r a t i o n with the U n i v e r s i t y o f M u n i c h . T h e site locations are s h o w n in Fig. 1, together with those o f the m a j o r faults which separate the shallower v a p o u r - d o m i n a t e d reservoirs to the SW f r o m the deeper w a t e r - d o m i n a t e d g r a b e n system (Batini et al., 1982).

FIELDWORK AND IN-FIELD DATA ANALYSIS T h e o b s e r v a t i o n s were m a d e d u r i n g two field c a m p a i g n s in 1980 a n d 1981 respectively, using m a g n e t i c sensors a n d recording systems which were d e p e n d e n t o n s o u n d i n g b a n d w i d t h ( T a b l e 1). Table. 1. Field instrumentation and data acquisition times Period range (s) (a) l0-3 - l0 (b) 10- i03 (c) l02 - 104

Magnetometers

Recording

Induction coils Automatic in-field data analysis Torque magnetometers (Marianiuk, 1977) Digital on magnetic tape cassettes Fluxgate magnetometers Digital on magnetic tape cassettes 645

Acquisition time per site 2 - 7 hrs 2 - 7 days 4 - 7 wks

646

V. R. S. Hutton, G. J. K. Dawes, f. Devlin and R. Roberts

•

1kin

%% ~lkSR A

~k

A

•

%%%%%

BELFORTE(~

"~

AJ

"-..

----

MT

\~ --%.%%_~

45"10

FAULTS

~

C

-....

"q,_ ~

10 - I0 •

~

/

/%-

.....

•

•

Q

".. -A AMT_ 10- ~- 10 S • MT 10 - 10as

•

/6

q..- - %6v ~R A B / ,

4~~ . _

/

e

•

•

MONTALCINELLO

%.~II%~. ~..

1 "?' / 9 7 8"A , , ~ A c'

I1~

"~

Fig. I. Location of sites, faults and traverses AB and CD discussed in the text. SRA, ISI, NRA, CER, RAB and RAC are the locations of the broadband MT soundings.

The automatic in-field analysis system, S.P.A.M., used for the short period A M T range, was designed and developed by Dawes. From the experience gained in this project an updated 7channel system has now been developed. Preliminary analysis of the longer period data was also undertaken during the field operation using a DEC P D P mini-computer at a field base station. The use of these in-field analysis systems helped ensure that the quality of the data obtained was considerably better than would otherwise have been possible in an area of anomalously intense disturbance. Moreover, it was possible to adjust the location of subsequent soundings during the field p r o g r a m m e in an attempt to ensure that the station spacing would be adequate for interpretation--an objective not fully realised in the complex, " n o i s y " Travale field. DATA ANALYSIS AND MODELLING TECHNIQUES The data have all been re-examined and analysed rigorously on the main university computers, with careful corrections for instrumental response functions and with the application of new p r o g r a m m e s to aid the selection of acceptable events for further analysis. These provided estimates of the period dependence of the following parameters: (a) magnetic and telluric field polarizations, (b) magnetic vertical field transfer functions, (c) unrotated, invariant, average

Broadband Tensorial Magnetotelluric Studies in Travale

647

and rotated apparent resistivities and phases (to principal and graben directions), (d) azimuths of principal impedances and (e) skews--dimensionality indicator. A modified Monte Carlo inversion procedure was then used to provide I-D models, which best fit the apparent resistivity and phase data at all sites. Two-dimensional modelling was also undertaken for a traverse perpendicular to the Travale graben, using the finite-difference p r o g r a m m e of Brewitt-Taylor and Johns (1980). M A I N RESULTS OF A N A L Y S I S The main results of the magnetotelluric data analysis are demonstrated by presentation of major and minor response estimates for one site RAB (Fig. 1) for which observations in the three overlapping ranges (a) - (c) of Table 1 have been collected and of p and + estimates from all the b r o a d b a n d data sets; in this case, a fixed azimuth parallel to the graben direction has been used. These results are given in Fig. 2, from which the following features can be noted: (a) There is a mis-match in the estimates of major apparent resistivity at RAB at periods about 100 s between those obtained with the torque and those with the fluxgate magnetometer RAB

t

~

,.,)~

~

e

g

81

o

~

r

I=

~.jo, ' b

~

,

i

i

Ii

s E

N4o°~

g

~8,'W

: ! IUO,

Ira.

II.

/d t,

I1.1

FREGUENC¥

IN

0.81

H7

8.081

qD.~l

i

18fle,

rh tM.

18.

1.

O.t

0.81

O.OOl

O.NO$

FI~EQUENC.¥ IN HZ

Fig. 2. Frequency dependence of: (a) apparent resistivity, major and minor, (b) phase-major, (c) azimuth-major and (d) skew for station RAB. Frequency dependence of: (e) apparent resistivity, (f) phase and (h) skew for all broadband MT sites for the strike N40°W of the Travale graben. The dashed curves in (a),(b),(e) and (f) represent the preferred interpolation between the well-estimated AMT and long period MT data.

648

V. R. S. H u t t o n , (5. J. K. Dawes, 7". Devlin a n d R. Roberts

systems. Corresponding discrepancies were observed at all locations where observations werc made in the 10 - 104 s period range. Intensive study of the noise bias in the period range (c) data sets suggested that the level of the apparent resistivity values could be increased to that of period range (b) as a result of noise contamination. As the total recording interval for this latter period range was normally only a few days, it was not possible to extract a sub-set of high signal to noise ratio as was the case for period range (c). The apparent resistivity curves thus preferred for interpretation are those in which the observational data from the ranges l0 -~ - 10 ° s and 10 ~'- 10 ~ s are connected by the interpolated broken curves of Fig. 2. The initial modelling studies were however undertaken to fit the actual estimates obtained in the period range 10- 3 - 10~2 S. (b) For periods greater than a few seconds, the values of apparent resistivity, the azimuth of the principal impedance and the skew factor at station RAB all exhibit values which are indicative of a change from one to two to three dimensionality in the electrical structure (Fig. 2a - d). (c) All broadband data sets exhibit a similar change in dimensionality as exhibited by the skew values (Fig. 2h). (d) The apparent resistivity and phase plots at all sites (Fig. 2e and f) [only the values in the graben direction (E-polarization) are plotted] have rather similar forms, being indicative of a three-layer model with the sequence r e s i s t o r - conductor - resistor, the intermediate layer being highly conducting ( < a few ohm-m). Variations in the values of the basement resistance appear to exist between the sites with the lowest values at RAC and NRA near " h i g h s " in the reservoir temperature map, but, bearing in mind the 3-D nature of the data at longer periods, great caution must be exercised in interpreting these in terms of true differences in the deeper electrical structure. The period dependence of the magnitude and direction of the magnetic induction vectors, not presented in this brief report, is compatible with the complex faulted structure in the region and with the dominating effect of the conducting sedimentary sequences of the Travale graben.

ONE- A N D T W O - D I M E N S I O N A L M O D E L S Since the data at all sites have been found to be either 1- or 2-D for periods up to about 100 s, a 2-D modelling study has been undertaken for traverse CD of Fig. 1. The starting model was constructed on the basis both of the geological section [Fig. 3a and resistivity information provided by Batini et aL (1982)1 and of the resistivity values obtained from I-D models fitting the Edinburgh data sets. This starting model was then modified to provide the model of Fig. 3b. This gave a good fit to the E- and H-polarization apparent resistivity and phase plots for all stations along the profile. An example of the fit of the model and observed responses is given in Fig. 5b for station 11. The geoelectric model clearly defines the contrasting highly conducting Neogene and flysch cover sequence and the more resistive carbonate reservoir formation and underlying basement rocks, although the data do not necessarily resolve the two resistive layers. As an aid to interpreting the l-D models which had been obtained for other observational sites, I-D models were obtained for the computed E-polarization and average E- and Hpolarization apparent resistivity and phase plots for a series o f locations along the traverse CD using the 2-D electrical structure model of Fig. 3b. Comparison of the resulting set of 1-D models and the 2-D model showed that: (i) 1-D modelling of the E-polarization estimates provided correct values of layer resistivities, but underestimated values of the depth to the 200 o h m - m layer, as shown by the lower dotted curve (Fig. 3b) and (ii) I-D modelling of the average E- and H-polarization responses provided a much improved estimate of the correct depth to the carbonate layer, as indicated by the dashed curve in Fig. 3b.

B r o a d b a n d Tensorial Magnetotelluric Studies in Travale

649 N -

AS

Neogone

AS-F lysch

R

cover

Ikm R - C a r b o n a t e

series

Pa

Ikm

~P

I

C

10

6

V

.X . . . . . . . . . . . . . . .

11

12

Y

v

Paleozolc basement

,

lkm t

D

,0

200

2O

t

~lllD~m m °ooeeooooeoeee e .. ,oo

3

2OO

1kin

|kin b $oo Fig. 3. (a) Geological section for part of traverse DC of Fig. 1, from Batini et al. (1982). (b) Electrical model for traverse CD with resistivity values in ohm.m and locations of stations for which the model provides a good fit between observed and computed p and ~ estimates. Dotted lines: layer boundaries from I-D inversion of E-polarization model p and estimates; Dashed lines: depth to 200 ohm-m layer from 1-D inversion of average E and H-polarization model 0 and estimates. Figures in ohm.m.

Following these synthetic studies, data from sites along the traverse AB of Fig. l were remodelled to fit both average major and minor apparent resistivity and phase estimates and the invariant tensor parameter IZxxZyy - ZxyZy~1½. Unlike the synthetic data, however, neither of the resulting sets of 1-D models yielded realistic depths to the top of the reservoir formation. The best, but underestimated, depth to the reservoir was obtained from 1-D models fitting the major p and + data as shown in Fig. 4a, in which R1 is the reservoir depth from the 1-D models at the sites 1 - 9 and R2 the depths obtained from the contour map provided by Batini et a/. (1982). Figure 4b shows examples of their fit to the data at two of the sites, 4 and 8. The failure of the field data to provide good estimates of the reservoir depth from 1-D models of Pav and P~Nvis compatible with the comments above regarding noise contamination of the p estimates in the period range 1 - 100 s. Further 2-D modelling studies were undertaken to explain the broadband data sets of Fig. 2e and f up to 100 s when they become predominantly 3-D. To satisfy both the observed amplitude and lateral variation of the 100 s Pa values, it was necessary to modify the original electrical model as shown in Fig. 5a. The computed p and phase plots for these models 1 and 2 for station i 1 (Fig. 5b) now show reasonably good agreement with the AMT data and the well-estimated responses at 100 s. These models differ from the model of Fig. 3b in the presence of the block of 0.5 ohm.m flysch cover and in the anomalously low resistivity ( ~ 50 ohm.m) of the basement and are preferred to it. The main difference between the two models of Fig. 5b is the exact location of the 0.5 ohm.m block. Changing the reservoir resistivity from 200 ohm.m to 50 ohm.m produces negligible change in the computed responses.

650

V. R. S. Hutton, G. J. K. Dawes, T. Devlin and R. Roberts

A

4

6

6!

o

-500

9

5

151

4

B

7 8

t:7

'1I

17

- 1000

Cl

-1500 Height

0

2

8km

4

in m e t r e s

STATION

4

STATION

1004:

100

if" ~

8

J

10

J

0 10

.( L

.[

I 9C

<~) 4-" z

b

0

1000

100

10

FREQUENCY

0"1

!000

HZ

100

10

FREQUENCY

1

0-1

HZ

Fig. 4. (a) The depths R2 of the top of the reservoir formation for sites on traverse AB of Fig. 1 (1) and the depths R I to a resistive substratum as deduced from the best-fitting l-D models of observed major p and ~ estimates at sites along AB. (b) Examples of fit of computed (-) and observed (o) major 0 and + estimates obtained for stations 4 and 8 from operation of system (a) of Table I.

CONCLUSIONS (a) For a traverse across the Travale graben a 2-D electrical model which fits the observed magnetotelluric responses has been obtained. This model is compatible with the known geological structure and in particular it identifies the interface between the highly conducting sedimentary sequences and the more resistive carbonate formation which constitutes the reservoir in this field.

Broadband Tensorial Magnetotelluric Studies in Travale I

a I0

C

v 200

b

II

12

v

V

v

10

S

L

651

I km

D

0

I

i : (3) ',

lkm

200

500

2km

SO

b STATION E

il

POL.

1000 I

H

100(

100

POL.

10~

10

1

0,'[ 0,001

i 0.01

0.1

1

10

100

0.'

0,001

0.01

0,1

1

10

100

10

100

period in seconds

period in seconds 9O

7 8O

=_zi~

•

451 30 15 0 0,001

,5

3o

T 't [1gtr._'l~~ 0,01

0.1

1

10

it111 ITI l i ~ it± I

t

15,

100

0

0,001

0.01

0.1

1

Fig. 5 (a) Electrical models for traverse CD of Fig. 1 with resistivity values in ohm'm and locations of stations for which the models provide a good fit between observed AMT and 100 s period data and computed p and + estimates. The solid lines provide the block boundaries in model 1 and the dashed lines an alternative location for the 0.5 ohm'm block, model 2. (b) Example of E and H polarization fit of computed and observed 13 and ~ estimates at station 1I. The observed estimates at 100 s are indicated by squares. Dashed line: computed values for model 1 of Fig. 5a; Heavy unbroken line: computed values for model 2 of Fig. 5a; Thin unbroken line: computed values for model Fig. 3b.

(b) M o d e l s w h i c h satisfy the lateral variation o f apparent resistivity at 100 s require the presence o f a z o n e o f resistivity as l o w as 0.5 o h m . m near the centre o f the thermal a n o m a l y and a b a s e m e n t resistivity o f the order o f 50 o h m . m in the Travale graben. O b s e r v a t i o n s f r o m the region a r o u n d the test area are, h o w e v e r , essential if the potential association o f such a n o m a l o u s l y l o w resistivities and a g e o t h e r m a l source is to be c o n f i r m e d . (c) It has b e e n s h o w n that 1-D m o d e l l i n g o f the E - p o l a r i z a t i o n M T responses provides g o o d estimates o f the true resistivities but u n d e r e s t i m a t e s the interface depths. W h i l e 1-D m o d e l l i n g

652

V. R. S. H u t t o n , G. J. fix. Dawes, 7-. Devlin and R. Roberts

of the average MT responses of synthetic data provides reasonably accurate interface depths, application of this procedure to the Travale data is less successful probably due to noise con+ lamination in the period range I .... 100 s. (d) The advantages of: (i) using an automatic in-field analysis system to ensure the acquisition of good quality data, (ii) making AMT measurements to obtain information about the electrical structure within the sedimentary sequences rather than the total conductances only and (iii) undertaking long period MT soundings at some sites over several weeks for penetration to reservoir depths in the centre of the graben have all been demonstrated. (e) The effectiveness of electromagnetic induction techniques for geothermal exploration would appear from this study to be limited primarily to delineating reservoir depths. As it now seems clear (Berktold, 1983) that many different types of geothermal field exist, the application of intensive E.M. techniques to most potential fields is likely to be much more rewarding. Moreover in the normal exploration situation, the observations would be made without the artifical disturbances of large power stations and distribution lines which were a major feature of this test area. REFERENCES Balini, F., Duprat, A. and Ungemach, P. (1982) Derivation of a preliminary conceptual model of the Travale geothermal field. (Personal communication).

Berktold, A. (1983) Electromagnetic studies in geothermal regions. Geophys. Surveys 6, 173- 200. Brewitt-Taylor, C. R. and Johns, P. B. (1980) Diakoptic solution of induction problems. J. Geomag. Geoelectr. Suppl. .']2, 7 3 - 7 8 .

Marianiuk, J. (1977) Photoelectric converter for recording the geomagnetic field elements: construction and principles of operation. Publ. Inst. Geophys. Polish Acad. Sci. V, 114-157.