Estimation of Data Errors for Small-loop EM Induction Systems and how Considering them Improves 1D Inversion Results

Small-loop, multi-frequency electromagnetic induction systems (SLEM) are a usual tool for investigating the electrical conductivity and magnetic permeability of the subsurface, up to some tens of meters deep. The data, the in-phase (IP) and quadrature (Q) components of the induced magnetic field, are interpreted by direct observation of the plan views of the responses obtained at each frequency, or by applying 1D inversion methods. Although these systems are really fast, they do not provide data errors. We applied a methodology to estimate these errors in a particular field example, for different system orientations, finding that, regardless of the system orientation, the errors of IP rapidly increase as frequency decreases and are generally much greater than the errors of Q. Though the errors usually increase with decreasing frequency, there can be especially noisy intermediate frequencies. Another important result is that even repeating an acquisition line only three times, good error estimates can be obtained. This could be easily done in field surveys and would be very useful, considering that we also found that including some information about data errors in the 1D inversions, effectively improves the quality of the obtained subsurface images.