عنوان مقاله [English]
نویسندگان [English]چکیده [English]
The magnetotelluric (MT) method is one of the several electrical techniques used in geophysical explorations. This natural-source electromagnetic technique is used to obtain electrical resistivity information from the subsurface structure. The data used in this method consists of a simultaneous measurement of naturally-occurring, time-dependent magnetic field fluctuations and of the electric fields induced in the earth by the magnetic fields. At a single site on the earth’s surface, a number of recording time segments of the horizontal (Hx, Hy) and vertical (Hz) magnetic and horizontal (Ex, Ey) electric fields are recorded to obtain the spatial resistivity variation within the earth. One of the major problems in the interpretation of MT data is the effects caused by topographical features or near-surface inhomogeneities. Such small local features may have a galvanic response (due to boundary charges) that is essentially independent of the frequency within the range of an MT sounding. These small 2D or 3D inhomogeneities cause the measured electric fields to be perturbed from their regional values and a shift of the apparent resistivity curves take place vertically in a log-log scale of the apparent resistivity sounding curves. This is called the MT static shift and needs to be minimized somehow to allow an accurate interpretation of data. Due to the importance of this phenomenon, a number of studies have been devoted to understanding and correcting this problem. The electromagnetic array profiling (EMAP) is a special form of MT data acquisition to resolve the spatial aliasing effects as well as the static shifts generated in MT measurements.
In the EMAP method, surface electric fields are sampled tangentially to a continuous survey traverse. Tensor impedance components along the EMAP traverse are either computed from the primary surface magnetic fields sampled at a fixed reference station or estimated from an array of magnetic stations (Figure 1). The Earth's resistivity distribution beneath the EMAP traverse is determined from these in-line tensor impedance components.
For both qualitative and quantitative interpretations, the importance of MT data inversion has increased in the past few decades. Among the developed two-dimensional (2D) inversion algorithms, the Occam’s inversion and the non-linear conjugate gradient (NLCG) methods are useful algorithms for MT data modeling. The original Occam method was provided for a 1D inversion of MT data and it was then developed for the 2D case by deGroot-Hedlin and Constable in 1990. This algorithm seeks the minimum possible structure model subject to an appropriate fit to the data. The NLCG algorithm employs the non-linear conjugate gradient method directly to minimize the objective functional of the MT inverse problem. In this algorithm, the computation of the full sensitivity matrix and the complete solution for the normal equation system in the model space are avoided.
The main objective of this study was to investigate the ability of Occam and NLCG inversion algorithms in an interpretation as well as 2D inverse modeling of Oklahoma EMAP data. It was also attempted to compare the results of these new inversions with those already acquired by the other different inversions. To achieve these goals, EMAP data were processed considering the noise amount and the bad data were removed and suitable data sets were defined for the inversion. For the inverse modeling, first an appropriate mesh grid for both resistivity blocks and forward computations was defined for each inversion algorithm. Then, the inversion parameters such as regularization parameter, the starting model and the target misfit were set to achieve acceptable results. Finally, the 2D NLCG and Occam inversion algorithms were applied to EMAP data to obtain geoelectrical models for the subsurface geological structures in the studied area. In addition, a set of MT data from four stations with five components of electric and magnetic field data along a line perpendicular to the EMAP profile were investigated to integrate with the results of the EMAP data inversions.
The results of this study indicate that data processing and controlling of the inversion parameters such as: an appropriate grid, regularization parameters, the starting model and the target misfit are very important for obtaining a suitable model. Therefore, by properly setting these parameters, the capabilities of the NLCG and the Occam algorithms for inverting a large volume of data have been tested and compared to those of other inversion algorithms. Furthermore, the capabilities of EMAP and MT data to map the subsurface resistivity distribution of the geological structures have been shown.
مرادزاده، ع؛ طهماسبی، ف و فاتح، م.، 1386، بررسی امکان بهکارگیری شبکههای عصبی مصنوعی در مدلسازی وارون دوبُعدی ترکیبی دادههای دو مد قطبش روش مگنتوتلوریک، فصلنامه علوم زمین، 64، 88-101.