Page 85 - Geothermal Energy Systems Exploration, Development, and Utilization

2.4 Geophysics 61
results both in mineral exploration (Farquharson and Craven, 2009) and in the
characterization of geothermal reservoirs (Uchida and Sasaki, 2006; Heise et al.,
2008; Newman et al., 2008).
Most 3D inversion codes are based on finite difference approaches (Mackie,
Smith, and Madden, 1994; Newman and Alumbaugh, 2000; Sasaki, 2004; Siripun-
varaporn et al., 2005), some other approaches such as the edge finite-element
method (Farquharson and Craven, 2009; Han et al., 2009) and the IE (Zhdanov
et al., 2000) have been tried to develop fast and reliable codes. While the differences
in the results they yield are not as big as the step from 2D to 3D inversion, Han
et al., (2009) show that methods may be somewhat faster than others. Nonetheless,
the main obstacle for their widespread application has been available computing
power, as calculation of a 3D MT structure is very challenging. This development
is likely to be accelerated with the further improvement of computing speed and
power, such that 3D inversion may well become the standard interpretation in the
near future.

Limitations, Problems, and Shortcomings of the MT Method The MT method has
been refined considerably over the last years but problems still exist, primarily with
noise in the measurements and lack of an adequate interpretation. Improvements
in data collection, data processing, and three-dimensional numerical modeling
continue to reduce such problems. Artifacts are inherent in every inversion
algorithm due to noise, undersampling, and three dimensionality, and so inverse
modeling results that provide a good data fit should not be regarded as the only
possible answer. A geologically reasonable model that fits the data is still the best
assurance that a model is credible.
Cultural noise may be considered as a main limitation when no filtering is
possible. An alternative approach for noise removal was proposed by Weckmann
et al. (2005), which uses a combination of frequency domain editing with subse-
quent single site robust processing. Nonetheless, even with a remote reference and
sophisticated processing noise remains a major problem especially in industrialized
areas. This problem may even occur once a geothermal field has been developed.
Subsequent MT exploration and monitoring is more difficult because pipes and
pumps generate a lot of electromagnetic noise that will contaminate the natural
signals.
As far as resolution with depth is concerned, the deeper the unit is, the thicker it
has to be in order to be mappable by MT. The MT data can be interpreted to give
an estimate of resistivity variations with depth. And, because MT needs a resistivity
contrast to be present in order to map a boundary, and because these units need
to be fairly thick to be mapped, the sections will not have the resolution of seismic
sections. A conductor below a massive salt layer – a setting that presents a challenge
to seismic imaging because of the high velocity contrast between the salt and the
underlying sedimentary rocks – can be detected quite successfully with MT, as salt
is usually highly resistive. The opposite is the case if a weak conductor is below by a
good conductor. Such a situation is difficult to resolve for MT. In high temperature
reservoirs, the overlying clay cap presents such a good conductor, which may make