2D INVERSIONS OF MAGNETO TELLURIC DATA FROM HIMALAYAN REGION

Abstract

2 Dimensional inversion has been applied to the processed existing data of the Himalayan region to study the conductivity distribution of the subsurface. The profile I have used extends from Deoband to Gangotri of 200km length. I have used 33 sites which are earlier processed by MT group at IIT Roorkee in this profile. The 2D inversion over these sites provides valuable information regarding various geological, features of the Himalayan region such as HFT, MBT, and MCT. This information can be used for the purpose of the study of the earthquake in this region. In this work 2D inversion method has been applied over the same data for three different frequency ranges in order to study the conductivity distribution in various depths. The three different frequency ranges are 0.001 to 10s, 0.001 -100s and 0.001 -10000s. Here the inversion has been performed using the WinGlink software and the misfit between measured data and data synthesised from the model is performed by the Tikhnov's regularisation method. The starting model is a three layer homogeneous models of thicknesses I km, 1-6km and below 6km having resistivity values l 00ohm-m, 200ohm-m and 500ohm-m respectively In first level the tipper data was inverted over the starting model. After this in second level TE mode phase (i.e. no weightage is given to resistivity) only is inverted over the tipper model as TE mode resistivity is more distorted by inductive distortions. Then in level third TM mode resistivity and phase is inverted over the TE mode. But the conductivity structures obtained from TE mode and TM mode were not same. In order to finalise the inversions joint inversion has been applied to the inverted TM mode to obtain the accurate conductivity distribution of the subsurface. The main difference between TE and TM modes is that TM modes charges the structure and its anomalies are superimposed by galvanic effect, but the TE mode does not charge the structure and its anomalies are superimposed by inductive effect. But at near surface due to high frequency, inductive distortion is more as TE mode produces inductive distortions so we cannot apply TE mode to detect the near surface conductive structure. Hence TM model is suitable for near surface study. For deep study due to low frequency, galvanic distortion is high and inductive distortion is low. So in this case TE mode is suitable. As we find TE and TM mode are complementary to each other we prefer joint inversion to obtain good information of conductivity distribution of the subsurface. 3 The second part of my work includes the inversions of 36 sites data from the same profile but processed by MT group at GERMC. The purpose of this is to find the difference between the processing tools of IITR and GEMRC by comparing the results of the inversions. The Third part of work includes the averaging of the adjacent sites using four different methods. The four averaging methods are (i) Weighted average of resistivity and phase values taking spacing with respect to reference site as weight (ii) Weighted average of resistivity and phase values taking corresponding error as weight (iii) Median (iv) Simple average. After obtaining four averaged values, four sets of EDI files are created and inversion has been applied over each set of EDI files. The idea behind the averaging is to smooth the resistivity and phase values so that maximum data we can use in the inversion rather than masking noisy data. Although smoothing the resistivity and phase values restricts to give the accurate picture of the subsurface. But in this case the profile length is of 200km and the result obtained from the inversions shows accuracy of the model has not been affected by averaging methods. It is clear that smoothing does not affect the accuracy of the subsurface image in case of long profile length. As the geoelectric strike direction varies from site to site, an average value of strike direction of N60°W has been taken for the rotation purpose and the comparison between the non-rotated and rotated model has been studied in the final part of the work. 0