LOTEM data processing for areas with high cultural noise levels

Abstract

Long Offset Transient Electromagnetic (LOTEM) soundings conducted in industrialized areas demand complex approaches to data processing for the removal of unwanted cultural noise. When sporadic, aperiodic noise (from pumps, electric fences, etc.) or periodic noise from power lines is present, as is the case in the F.R.G., standard data processing commonly outputs averaged transients which are contaminated by the noise influence(s), and any processing done after stacking usually fails to improve the data resolution. A new approach to the problem of periodic noise is to process each individual transient before averaging, which greatly improves the signal-to-noise ratios and provides a more detailed geoelectric image of the subsurface. An improved prestack, digital-recursive, true-amplitude notch filter is applied to synthetic and real data. Sporadic noise is best handled with either a symmetric or an area-defined rejection selective stack, whereby the signal amplitudes are selectively kept and the noise is statistically rejected. Data show that for LOTEM soundings the symmetric rejection algorithm improves the signal-to-noise ratios significantly. After selective stacking, the system response is deconvolved and the data smoothed with digital recursive filters. The output transients are then converted to apparent resistivities and inverted to a horizontally layered Earth model. A comparison of prestack and poststack data processing for one profile shows that with prestack data processing the section correlates well with the known geology of the area, whereas poststack processing outputs an interpreted geoelectric section that is inconsistent with the geology.