Use of magnetovariational results to improve the distortion diagnostics in deep magnetotelluric soundings

E. BORZOTTA

Moscow - Zvenigorod, Russia

Workshop; International Workshop on Electromagnetic Soundings; 2010

Moscow State University, Faculty of Geology

USE OF MAGNETOVARIATIONAL RESULTS TO IMPROVE THE DISTORTION DIAGNOSTIC IN DEEP MAGNETOTELLURIC SOUNDINGS   E. Borzotta Unidad de Geof¨ªsica, IANIGLA, CCT CONICET, (5500) Mendoza, Argentina e-mail: eborzota@lab.cricyt.edu.ar   An interpretative experience from nine deep magnetotelluric soundings was carried out in the central region of Argentina (Fig.1). As shown in this figure, the soundings are away each other so much as 50-100 km along 500 km from the Andean region in the West to the cratonic zone in the East. Soundings were done from 1985 to 1993. The main interest was focused in the determination of deep conductive layers in the crust and upper mantle. These soundings do not have interpenetrated their effective volumes; therefore, each one of them have to be interpreted isolate within a particular geological context. After field data processing, the first step in interpretation process ¨Cas it is known (Berdichevsky et al 1989)- is to make a distortion diagnostic from the principal curves at each sounding location, using the distortion theory and the available tectonic knowledge. The target is to build in mind the most reliable tectonic structure causing distortions in each site, and its subsequent removing or attenuation to estimate a normal curve; i.e. a curve without distortion, previous 1D modelling. A correct estimate of each normal curve ¨Cespecially its position- is crucial to obtain correct depth estimates of deep layers. To perform this operation, the distortion theory (Berdichevsky & Dmitriev 1976, Berdichevsky et al 1989) was used, but it is very important to have a priori information to arrive to satisfactory results. In this experience, the magnetotelluric soundings are isolated each others, which sometimes makes it more difficult to obtain good diagnostics. Therefore, a magnetovariational study was carried out using the horizontal geomagnetic variation field. Taking into account the near presence of the Pilar Geomagnetic Fig.1: Map of studied region in South America. Squared areas indicate orogenic zones. Dashed areas are sedimentary basins. Black points mean MT sounding sites, and PGO means Pilar Geomagnetic Observatory, used as a reference site in MV study.    Observatory (PGO) (Fig.1), giving synchronous magnetograms to those in the field, comparisons between horizontal magnetic events in field and PGO were conducted and hodographs were built for all MT sites. This study was of great importance to reveal large conductive structures causing distortions in some soundings; and thus, to obtain better normalizations of soundings, detecting unsuspected structures and mistakes in diagnostics. As a first step, using the distortion theory and rather scarce available tectonic information from the surrounding of each MT location, an approximate normal curve was estimated at each site. Then, from the ¦ÑN curves, the integrate conductivity at T= 1000 s of period were also estimated at all locations. As these conductivities were then compared with magnetic transfer functions (magnetovariational study), the period of 1000 s was selected because it is approximately the central period between 10¡¯ and 30¡¯ where the majority of magnetic events are present.    The integrate conductivities were estimated from each MT sounding, considering the estimate normal curve at 1000 s, using the tetra-logarithmic abacus proposed by Fournier (Fournier 1965). These values were normalized (Sr) using the integrate conductivity at 1000 s of period at PGO (730 Siemens); the PGO is seated on a cratonic region (Fig.1). On the other hands, using the hodographs from the horizontal magnetic field, the transfer functions (amplitude ratios) corresponding to H and D between 10¡¯ to 30¡¯ of periods was also estimated at each location, using PGO as a reference site. Then, the average transfer functions between 10¡¯ to 30¡¯ of period were estimated at each location. To do this estimate, H and D transfer functions, for a given period, were averaged only in the case they had similar values; otherwise, when H and D transfer functions differ, the lower value was used (H or D), considering that the higher value is suggesting a conductive structure. All values thus obtained for all magnetic events between 10¡¯and 30¡¯ of period were then averaged. Now, the graphic expression of Sr = f(Ar) on a bi-logarithmic space clearly suggests a linear law (Fig.2). Therefore, if this were correct,   Fig.2: Graphic expression of Sr =f(Ar), where: Sr is the integrate conductivity corresponding to the normal curve, at each site, at T= 1000 s, normalized to PGO at the same period. Ar is the average horizontal magnetic transfer function between 10¡¯ and 30¡¯ of period. Black points indicate results obtained from the first estimates of normal curves. Only at two sites the estimates were correct according to this methodology; in the other locations, according to the Ar values, the Sr values are not correct (points are outside the line). Then new values of Sr are estimated shifting the points outside the line on it, and therefore new conductivities and correct positions of normal curves were obtained.        a procedure to check and correct estimated normal curves would be available. In the present study, points on figure 2 ¨Ccorresponding to field sites- located outside the estimate linear law (Fig.2) were shifted up or down in order to re-locate them on the line. As a consequence, new more accurate integrate conductivities at T= 1000 s are now available for those locations, thus indicating that the position of ¦ÑN curves, first estimated, in such locations were not correct. Observing figure 2 it becomes clear that, according to this methodology, the initial normalization were only correct for LP and BUL sites (Fig.1). In the other sites, diverse mistakes were produced in the first estimates of ¦ÑN, due to the scarce geological information used. Let us briefly see the arguments taken into account in the first estimates of ¦ÑN at some sites and the tectonic consequences arise later from the correct ¦ÑN (¦ÑNc). USPA sounding was carried out in a tectonic valley into de Andean Cordillera. Both principal curves (~ NS and WE) seem to be shifted down by edge effects (Rokityansky 1982), suggesting a 3D structure. The NS curve (presumably ¦Ñ¨U, parallel to Andean Cordillera) has a strong deformation probably produced by the coast effect from Pacific Ocean (230 km westward). The WE curve does not seem to present deformation, but according to suppose 3D context, this sounding is no interpretable because the correct position of ¦Ñ©Ø is unknown. However using information from MV study it is possible to shift up ¦Ñ©Ø at its correct position as ¦ÑN curve, becoming the interpretation now possible and also confirming the 3D character of the structure in this region.   TUPU sounding was carried out next to Frontal Cordillera, with the mountain region to the West and the plain to the East (Figs.1,3). Principal directions are roughly NS and WE. To estimate ¦ÑN, the NS curve ¨Cparallel to Cordillera- was considered as ¦Ñ¨U. Therefore, as ¦Ñ©Ø seems to have less deformation, this one was shifted down in order to its high period branch agree to the high period branch of ¦Ñ¨U. However, this ¦ÑN thus estimated, without enough tectonic information, revealed to be no correct. In fact, according to MV results, the integrate conductivity in the site at 1000 s of period would be a lot smaller (Fig.2). So, to obtain the correct ¦ÑN curve (¦ÑNc), the ¦Ñ©Ø had to be shifted up (Fig.3). If were correct, the ¦ÑNc would be giving evidence Fig.3: Heavy lines correspond to the MT ¦Ña principal directions obtained at TUPU sounding. At a first approximation the site was considered as bi-dimensional, according to the available tectonic information. The lower curve, approximately parallel to Andean Cordillera was considered as ¦Ñ¨U. To obtain the normal curve (¦ÑN), the upper curve (apparently less deformed) was shifted down in order to its large period branch agree with the large period branch of longitudinal curve. However, the MV study (Ar) indicates as wrong this normalization. The correct position of ¦ÑN would be the ¦ÑNc with less conductivity. If true, the sounding would have 3D tectonic with a probable 3D conductive effect on the longitudinal curve.   of a probable 3D conductive effect on ¦Ñ¨U (Berdichevsky et al 1998). As a consequence, ¦Ñ¨Uis descended, revealing a no suspected 3D structure. BEAZ sounding was carried out into a very saline basin (Fig.1). However, the correct normal curve (¦ÑNc) (obtain from MV results and using the procedure explained) is suggesting that both principal curves have excessive resistivity values, which is an unexpected result. Therefore, some resistive structures should be present into the basin, no visible in surface.                                           Fig.4: Amplitude transfer functions corresponding to TUPU site, referred to PGO. Only events between 10¡¯ to 30¡¯ of period were used to estimate Ar.     The procedure used in this interpretative experience present however a feeble aspect. As visible on figure 2 the horizontal magnetic field has small sensibility to detect large conductive structures compare to the electric field, which is a lot more sensible. Therefore, a small variation in the magnetic transfer functions implies large variation in the integrate conductivities. As a consequence, we should use many magnetic events in order to estimate correctly as the transfer functions, due to the natural dispersion of magnetic events, especially at large periods.    On the other hand, this experience gives evidence that a more acceptable distortion diagnostic ¨Ccloser to the true- can be obtain if we have: a)    a good skill in distortion theory. b)   a magnetovariational study, including horizontal magnetic field, and the vertical one if possible (induction vectors), conducted previously to MT study. c)    good knowledge of tectonic and geological context, with a priori information.   References   Berdichevsky, M. N., Dmitriev, V.I., 1976. Basic principles of interpretation of magnetotelluric sounding curves. In: Ad¨¢m A. (Ed.), Geoelectric and Geothermal Studies (East-Central Europe, Soviet-Asia), KAPG Geophysical Monograph. Akad¨¦miai Kiad¨®, Budapest,pp. 165-221. Berdichevsky, M.N., Vanyan, L.L., Dmitriev, V.I. 1989. Methods used in the U.S.S.R. to reduce near-surface inhomogeneity effects on deep magnetotelluric sounding. Phys. Earth Planet. Inter. 53, 194-206. Berdichevsky, M.N., Dmitriev, V.I., Pozdnjakova, E.E, 1998. On two-dimensional interpretation of magnetotelluric soundings. Geophys. J. Int., 133, 585-606. Fournier, H.G., 1965. Abaque des solutions du systeme (¨¦tablies en suivant la m¨¦thode L. Cagniard). Institute du Physique du Globe, Paris. Note N¡ã9. Paris. Rokityansky, I.I., 1982. Geoelectromagnetic Investigation of the Earth¡¯s Crust and Mantle. Springer ¨CVerlag, Berlin.