Identifying and removing micro-drift in ground-based electromagnetic induction data

• Stability of electromagnetic induction data was studied after baseline levelling.
• Both in-phase and quadrature phase FDEM data display remnant micro-drift.
• Micro-levelling is needed to ascertain accurate data representation.
• A wavelet-based procedure is proposed to identify and remove micro-drift.
• The procedure is evaluated and proven suited for different setups and datasets.

As the application of ground-based frequency domain electromagnetic induction (FDEM) surveys is on the rise, so increases the need for processing strategies that allow exploiting the full potential of these often large survey datasets. While a common issue is the detection of baseline drift affecting FDEM measurements, the impact of residual corrugations present after initial drift removal is less documented. Comparable to the influence of baseline drift, this ‘micro-drift’ introduces aberrant data fluctuations through time, independent of the true subsurface variability. Here, we present a method to detect micro-drift in drift-corrected FDEM survey data, therefore allowing its removal. The core of the procedure lies in approaching survey datasets as a time series. Hereby, discrete multi-level wavelet decomposition is used to isolate micro-drift in FDEM data. Detected micro-drift is then excluded in subsequent signal reconstruction to produce a more accurate FDEM dataset. While independently executed from ancillary information, tie-line measurements are used to evaluate the reliability and pitfalls of the procedure. This demonstrates how data levelling without evaluation data can increase subjectivity of the procedure, and shows the flexibility and efficiency of the approach in detecting minute drift effects. We corroborated the method through its application on three experimental field datasets, consisting of both quadrature and in-phase measurements gathered with different FDEM instruments. Through a 1D assessment of micro-drift, we show how it impacts FDEM survey data, and how it can be identified and accounted for in straightforward processing steps.

Figure optionsDownload as PowerPoint slide