04.08.2013 Views

P. Schmoldt, PhD - MTNet - DIAS

P. Schmoldt, PhD - MTNet - DIAS

P. Schmoldt, PhD - MTNet - DIAS

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Recovering a synthetic 3D subsurface model using<br />

lower-dimensional inversion schemes<br />

8.1. Motivation<br />

8<br />

Two-dimensional (2D) inversion of magnetotelluric (MT) data is at present far more commonly<br />

used than three-dimensional (3D) inversion since 2D inversion significantly outperforms<br />

3D inversion in terms of speed, thus allowing for much better resolution of<br />

the subsurface through a larger feasible number of grid cells (cf. Sec. 6.3). Moreover,<br />

due to the shorter computation time of 2D inversion, investigators can study various aspects<br />

of subsurface regions through hypothesis testing, e.g. by using different a priori<br />

models or by removing features of an inversion model and examining whether they are<br />

re-introduced in subsequent inversion steps. Models of 3D inversion are often a “like it<br />

or lump it”, because computational cost prohibit calculation of additional inversion steps.<br />

However, validity of 2D inversion needs to be tested for cases where the electric resistivity<br />

structure of the subsurface is potentially 3D to some extent. Not taking into account the<br />

effects of 3D structures can severely corrupt resulting models; see Chapter 4 for a detailed<br />

discussion of distortion and dimensionality aspects of MT investigation.<br />

Different 2D inversion approaches have been applied to 3D subsurface cases before, but<br />

respective models have to be regarded with suspicion. For example, during their approach,<br />

using interpolation of Zyx-only 2D inversions 1 to image 3D structures of the Pannonian<br />

Basin (Hungary), Tournerie and Chouteau [2005] note that responses from 3D forward<br />

modelling of their interpolated model are significantly different for Zxy periods related to<br />

deeper regions (> 20 s). The authors relate the discrepancy to unaccounted for anisotropic<br />

structures in the subsurface. However, Zxy data, with the electric component orthogonal<br />

to the profile (i.e. TE in 2D MT inversion), is commonly assumed to be more affected by<br />

1 For a 3D subsurface no alignment of Zxy and Zyx data with a 2D electric resistivity interface can be made<br />

in the classical sense. Thus, TE and TM mode only denote ‘transverse’ in terms of ‘transverse to the<br />

profile direction’ and are not necessarily related to the orientation of a resistivity interface. With the<br />

profile along the y-axis, Zyx is associated with the TM mode.<br />

169

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!