D020 Statics in Magnetotellurics - Shift or Model?

Transcription

1 D020 Statics in Magnetotellurics - Shift or Model? W. Soyer* (WesternGeco-Geosystem), S. Hallinan (WesternGeco- Geosystem), R.L. Mackie (WesternGeco-Geosystem) & W. Cumming (Cumming Geoscience) SUMMARY Magnetotelluric data sets often show signs of significant local galvanic distortion - static shifts in its simplest form - complicating their quantitative analysis. Our accounting for this problem is two-fold. Colocated magnetic TDEM soundings provide representative resistivity estimates of an approximate 1 D near-surface structure, allowing for calibration of the MT apparent resistivity at the highest frequency limit. Remaining distortion that is unaccounted for can be addressed directly and automatically via 3D inversion: as galvanic effects occur equally in the synthetic simulation of the induction process whenever currents cross resistivity interfaces, the inversion algorithm generates shallow near-surface structure that can reproduce the observed statics. Today's increasing computer resources allow for the detailed meshes and small near surface cell sizes necessary for this approach. Constraints on model roughness prevent local statics solutions from affecting the deeper resistivity image in the minimum structure inversion. For some data sets and geological environments, direct inversion for static shift parameters in addition to the above structural approach may be preferred. Applicability of this two-fold approach is demonstrated on a combined MT data set from the geothermal prospect of Glass Mountain, California.

2 Introduction In spite of much progress in magnetotelluric data acquisition and processing, local galvanic distortion can still pose a challenge to the analysis of magnetotelluric data sets. Caused by built up charge at small-scale heterogeneities of similar dimensions as the sensor dipole length, the electric field vector may be strongly distorted. The phenomenon can be properly described mathematically (Habashy et al. 1993), but despite the development of various sophisticated decomposition algorithms a recovery of the desired undistorted impedances is essentially impossible. The issue can be addressed a number of ways: manual adjustment tensor decomposition focus on impedance phases calibration (TDEM) minimum structure 3D inversion Manual shifting of absolute values from educated guesswork and comparison with other MT soundings cannot be considered a general solution to the problem. Decomposition algorithms all depend on assumptions of a regional 2D, undistorted impedance tensor that are not valid for the general 3D case (e.g., Broom and Bailey 1989, Chave and Smith 1994). Although the phase tensor formulation by Caldwell at al (2004) is of appealing simplicity, absolute impedance values provide most critical information on subsurface resistivity, and cannot be deduced from phase alone. Our accounting for electric galvanic distortion is two-fold: co-located time domain measurements (e.g. Pellerin and Hohmann, 1990) can allow for reliable estimation of shallow resistivity and thus a calibration of high frequency MT apparent resistivities (examples are shown in Figure 1). The availability of computer resources to work on large, detailed 3D meshes means that now the remaining statics in MT data can be reproduced by minimum structure 3D inversion (Mackie and Madden 1993, Rodi and Mackie 2001) by automatic generation of shallow near surface structure. A B Figure 1: Two examples of (un-shifted) MT soundings with synthetic 1-D model responses from co-located central loop TDEM soundings. Both MT soundings on average underestimate subsurface resistivity, but only (B) represents the classical case of galvanic distortion. Careful judgment on each sounding is necessary in calibrating MT from TDEM. The applicability of these methods is demonstrated on a combined 165-site MT sounding data set from the geothermal prospect of Glass Mountain, California, analyzed during a research project for the Geothermal Resources Development Account (GRDA), California Energy Commission.

3 Profile C Figure 2: Glass Mountain study area with shaded topography (left), and MT site distribution (right). The location of profile C used for display of resistivity cross-sections is also shown. TDEM: a magnetic-only estimate on shallow resistivity Time domain data were acquired alongside the MT data acquisition in single and central loop configuration, and the magnetic transients were modeled for 1-D resistivity structure. Synthetic MT forward responses of these models were calculated, overlapping with the observed MT data set in the 1-10 khz frequency band. Measured MT phases typically agree well with synthetic TDEM phases, and the synthetic apparent resistivities can provide a realistic, representative estimate on near-surface resistivity in ideal conditions. The 100x100m single loop TDEM response is reliable in conductive sequences, but not where near-surface resistivities exceed 100 Ωm (see Buselli, 1982 for discussion). Large moment, 300x300m central (in-) loop soundings may be more reliable, but this is an expensive solution. Figure 3: TDEM resistivity estimation at MT station n083, from single loop sounding n083 and central loop sounding r015. Galvanic distortion and 3D modelling The very local static distortion is outside the scope of structural MT modelling, i.e. we do not intend to appropriately resolve the structure causing the distortion if its scale is smaller than the electrode-spread. In practice, however, the scale of contrasts relating to static effects in the data overlaps with the range of dimensions used in modelling, and whether removal of galvanic effects from MT soundings prior to modelling is desirable is not always a trivial question. This is especially true for survey areas with high topographic variation, where at least part of the galvanic effects may be attributed to topography (Jiracek, 1990). On the other hand, even if the scale of the structural contrast causing the shift is smaller than the minimum mesh cell, the shifts may be reproduced accurately in 3D inversions by finding very shallow near-surface structures, limited to a small number of cells (and not interpreted). Varying the constraints on the model roughness with depth prevents local statics solutions 70th EAGE Conference & Exhibition Rome, Italy, 9-12 June 2008

4 affecting the deeper resistivity image in the minimum structure inversion. Ever increasing computing resources and parallelization of modelling algorithms allow for very detailed discretization (1 million cells and more), such that these shallow structural adjustments typically do not interfere with neighboring MT soundings. Our 3D algorithm also allows for direct inversion of statics parameters. MT data sets with prevalent very high statics can require this direct approach in order to save deeper model parts from influence of static effects. Constraints on total variance of predicted shifts and an iterative dumping mechanism encourage undistorted data structure to be reflected in robust model structure. Not surprisingly, inversions for static shift factors tend to result in smoother models (Figure 4) and the choice whether or not shift factors are directly to be inverted depends on the data set and the geological environment. RMS=1.59 RMS=1.45 Figure 4 Resistivity cross sections through two different 3D inversion models along profile C, without (top) and with inversion for static shifts (bottom). Both inversions were run with a higher structural smoothing constraint than the final model shown in Figure 5. A lower RMS and a slightly smoother model is achieved for the inversion including statics prediction. Outlook Smaller dipole lengths may reduce statics in standard band MT soundings, but at the cost of a strong decrease in signal level. Use of higher frequency techniques (RMT or VLF-R) in combination with short dipole lengths may result in smaller statics at the highest frequency limit, but this extended HF range is outside the frequency band (and spatial dimensions) currently used in 3D inversions, thus not constituting a full solution to the problem. Besides TDEM instrumentation, magnetic frequency domain EM soundings can be applied for statics calibration, requiring a layout of less spatial extent than TDEM loops, and therefore faster to deploy in difficult terrain. In areas of very high resistivity contrasts at shallow depth (current channeling), the problem is of magnetic galvanic distortion, where minimum structure MT inversion is perhaps unsuitable.

5 Figure 5: Interpreted resistivity cross-section through final 3D MT inversion model. The lower smoothing constraint in this inversion allowed for better resolution of the shallower high conductivity structure. Note the good agreement between isotherms from well temperatures and extrapolated gradients, and the resistivity structure. In summary, solutions are likely to be found through a combination of a) independent estimations of near-surface resistivity, at a suitable scale, and b) inversion modeling of residual statics in 3D in the near-surface, including topography, again using suitably dimensioned mesh, frequency range and regularization operator. Above all, a correct understanding of the causes of static shift is required in each survey case rather than blind application of any one correction approach. References: Buselli, G. [1982] The effect of near-surface superparamagnetic material on electromagnetic measurements, Geophysics, 47, Caldwell, T.D., Bibby, H.M. and Brown, C. [2004] The magnetotelluric phase tensor. Geophysical Journal International, 158, Chave, A.D. and Smith, J.T. [1994] On electric and magnetic galvanic distortion tensor decompositions, Journal of Geophysical Research, 99, Groom, R.W. and Bailey, R.C. [1989] Decomposition of magnetotelluric impedance tensors in presence of local three-dimensional galvanic distortion, Journal of Geophysical Research, 94, Habashy, T.M., Groom, R.W. and Spies, B.R. [1993] Beyond the Born and Rytov approximations: a nonlinear approach to electromagnetic scattering, Journal of Geophysical Research, 98, Jiracek, G.R. [1990] Near-surface topographic distortions in electromagnetic induction, Surveys in Geophysics, 11, Mackie, R.L. and Madden, T.R. [1993] Three-dimensional magnetotelluric inversion using conjugate gradients, Geophysical Journal International, 115, Pellerin, L. and Hohmann, G.W. [1990] Transient electromagnetic inversion: a remedy for magnetotelluric static shifts, Geophysics, 55, Rodi, W. and Mackie, R.L. [2001] Nonlinear conjugate gradients algorithm for 2-D magnetotelluric inversions, Geophysics, 66,