Noise properties of Fourier deconvolution for time-domain electromagnetic soundings

The transient electromagnetic method (TEM) is widely used for mapping the earth's subsurface resistivity structures, e.g., for groundwater or mineral exploration, using airborne and ground-based systems. Data from TEM surveys can be modified using Fourier deconvolution to correct for instrument drift, increase apparent bandwidth, or alter the apparent excitation waveform, e.g., to fuse or compare measurements from different systems, for visual presentation purposes, or to make the data compatible with a specific processing or inversion code. Although this method has been applied in several studies, little attention has been devoted to its properties with regard to noise. Using a generic analytical system model and examples featuring synthetic and field data, we perform a detailed analysis of the noise properties of Fourier deconvolution in the context of TEM. We find that although the effects from stationary noise are trivial, effects from nonstationary noise are less intuitive and more severe, e.g., causing on-time noise phenomena to degrade off-time data. In general, we observe that the method decreases the signal-to-noise ratio, and our recommendation is therefore that Fourier deconvolution should only be applied when it is desirable for presentation purposes or when strictly necessary for processing or inversion.