P. Schmoldt, PhD - MTNet - DIAS

10.2. Inversion for mantle structures
can introduce artefacts in the resulting subsurface model, as shown in the synthetic model
study (Sec. 8.3.1). Therefore, the investigation of the Tajo Basin subsurface is extended
in the following Sections using anisotropic 1D and 2D approaches as well as isotropic 3D
inversion.
10.2.2. Anisotropic 1D inversion
Motivated by the successful application of anisotropic 1D inversion for the case of a
synthetic 3D model with oblique strike direction (Sec. 8.3.2) the approach is used for
the PICASSO Phase I data from the Tajo Basin. For anisotropic inversion, impedance
tensor data are not decomposed (cf. Sec. 9.6.2) in order to preserve information in the
off-diagonal elements of the impedance tensor. Effects of electric resistivity interfaces are
imaged by anisotropic structures with anisotropy directions (directions of maximum and
minimum resistivity) aligned with the 2D geoelectric strike direction. For 1D inversion
no alignment of the coordinate axes with a specific strike direction is required and the
anisotropy direction is determined during the inversion process. During the PICASSO
Phase I fieldwork campaign MT recording systems are aligned in respect to magnetic
North, which is within a range of 2° of true North for the PICASSO Phase I stations.
Hence, a rotation to true North (yielding a common orientation of all recording stations)
causes only a minor mixing of errors from different channels. Anisotropic 1D inversion
of Tajo Basin data is carried out with the algorithm ai1D [Pek and Santos, 2006] and the
same parameters used during the synthetic 3D model study (Tab. 8.4).
The inversion yields no systematic direction of maximum anisotropy and no straightforward
relation with the mantle strike direction (N29.4E) that can be made (cf. top-left
plot in Figure 10.17). Therefore, anisotropic resistivity values are assigned to the mantle
strike-parallel model (ρ) if they are inside a ±90° range of N29.4E and otherwise
assigned to the orthogonal model (ρ⊥); i.e. the bottom-right and bottom-left plot in Figure
10.17, respectively. Resulting models exhibit a diverse subsurface structure with the
ρ⊥ model containing mostly values in the range 10 2.5 – 10 4 Ωm, whereas the ρ model
contains mostly values between 10 1 and 10 3 Ωm (the anomaly below stations pic013 and
pic015 is due to rejection of long-period data at these station, cf. Section 9.4). The
anomaly beneath stations pic013 and pic015 is also reflected in the anisotropy magnitude
(top-right plot in Figure 10.17), exhibiting for the respective region a value of 1 (i.e.
no anisotropy). Unlike for the synthetic model (Sec. 8.3.2), the anisotropic inversion
1D model for the Tajo Basin yields anisotropy magnitude values greater than one for the
crustal depth range, thereby indicating that the Tajo Basin crust is more complex and cannot
approximated by an isotropic 1D structure. Further, in the anisotropic 1D inversion
models no change of electric resistivity can be found that could be associated with the
LAB.
Since anisotropic 1D inversion of the undecomposed datasets fails to provide a plausible
model of the Tajo Basin subsurface, anisotropic 1D inversion is also conducted for
a dataset decomposed according to the strike direction of the mantle. Decomposition of
255