Coherency-based sorting of magnetotelluric data

Document Type

Conference Proceeding

Publication Date



© 1996-2018 Society of Exploration Geophysicists. All Rights Reserved. Random noise reduces the quality of authentic magnetotelluric (MT) signals and reduces the coherency between measured and predicted data. MT impedance tensor estimates and subsequent apparent resistivities are expected to be less susceptible to noise-induced biases if the most coherent data segments are used in the calculations (Stodt, 1986). We modified our Phoenix Geophysics real time MT software to sort data based on E-predicted coherency. The best 30 data segments are sorted and ranked; apparent resistivities are calculated for the best 10, 20, and 30 data segments and compared with resistivities resulting from all available data. In the 1.0 to 10 second period range (prone to noise contamination) significant bias reductions in apparent resistivities are obtained by using only the best-quality data segments.

Publication Title

1987 SEG Annual Meeting