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.

RMS−misfit<br />

3.0<br />

2.5<br />

2.0<br />

1.5<br />

1.0<br />

0.1 1 10 100<br />

Roughness(1/τ)<br />

10.2. Inversion for mantle structures<br />

Iteration 0<br />

Iteration 1<br />

Iteration 2<br />

Iteration 3<br />

Iteration 4<br />

Iteration 5<br />

Iteration 6<br />

Iteration 7<br />

Iteration 8<br />

Iteration 9<br />

Iteration 10<br />

Fig. 10.25.: RMS misfit for isotropic 3D inversion models at different iteration steps of the second inversion sequence using the lowest<br />

misfit model of the initial inversion sequence as starting model; i.e. the model obtained from the fifth iteration with a halfspace starting<br />

model and a model roughness of 10 (cf. Fig. 10.23). The RMS misfits for the fifth iteration of the initial sequence are indicated by<br />

markers connected with the dashed black line. The wsinv3d algorithm [Siripunvaraporn et al., 2005a] determined up to four models<br />

at each iteration step and uses the lowest misfit model as starting model for the following inversions (cf. Sec. 6.3). The overall lowest<br />

RMS misfit is obtained during the second inversion sequence after eight iterations (purple markers). Results are plotted in terms of<br />

RMS misfit versus roughness 6 for all models at each iteration step.<br />

displaying individual misfits for each station are moved to the Appendix (Sec. A.4.2).<br />

It is possible that the model of the eighth iteration step represents a local, rather than the<br />

global, minimum (cf. Sec. 6.3), but owing to the considerably low misfit of this model<br />

the second inversion sequence is terminated after ten iterations (resulting in a total of 15<br />

iteration steps).<br />

Results of the second 3D inversion sequence are presented in terms of the lowest misfit<br />

model with the highest model roughness (RMS = 1.50, 1/τ = 10), the model with the<br />

lowest misfit for a slightly lower roughness (RMS = 1.59, 1/τ = 3.16), and the lowest<br />

roughness model (RMS = 1.75, 1/τ = 0.32); labelled ‘a’, ‘b’, and ‘c’ in Figure 10.27,<br />

respectively. The three models exhibit a strong mutual similarity, i.e. all models comprise<br />

a relatively conductive region ≤100 Ωm in the centre of the PICASSO Phase I profile at<br />

a depth range associated with the lithospheric-mantle beneath the Tajo Basin (i.e. 30 –<br />

110 km). For the Earth’s lithospheric-mantle usually electric resistivities in the range of<br />

1000 Ωm or greater are determined (cf. Sec. 5.2.2). A reduction in the order of one magnitude<br />

in electric resistivity implies an unusual geological setting of the lithospheric-mantle,<br />

e.g. significantly increased temperatures or well-connected networks of conducting constituents<br />

(cf. Sec. 5.1).<br />

Ten additional iterations are carried out to ensure that the relatively low electric resistivity<br />

of the lithospheric-mantle in the centre of the profile is not the result of smoothing<br />

constraints used during the inversion process in combination with low sensitivity of the<br />

PICASSO Phase I dataset to the lithospheric-mantle. Therein, a layered a priori model<br />

containing a 100 Ωm crust (≤31 km), a 1000 Ωm lithospheric-mantle (31 – 110 km), and<br />

100 Ωm sublithospheric mantle is used, which enforces a more resistive lithosphericmantle.<br />

Inversion models obtained using the layered a priori model exhibit a more re-<br />

265

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

Saved successfully!

Ooh no, something went wrong!