IEEE Geoscience and Remote Sensing Magazine - March 2021 - 181

potential applicability of 1D methods. These are all obvious advantages, owing to the advances made in the FDEM
survey and considering survey speed, precision, resolution,
and multiple array setups (i.e., simultaneously measuring
multiple survey lines).
CONCLUSIONS
Particularly when survey targets are well defined, considering their smallest instrument-specific response allows for efficiently designing survey strategies with optimal inline and
interline sampling intervals. One should be cautious when
assigning sampling densities, as a random component inherent to the actual measurement positioning should be accounted for. Inline sampling intervals should be determined
such that a desirable RRMSE is achieved, depending on the
application and data acquisition speed. Interline sampling
intervals should then be based on instrument noise levels,
response amplitude, and time efficiency.
Staggering due to time-lag or georeferencing corrections
can be easily resolved by repeating survey lines. Repeating one survey line, as opposed to multiple random lines,
should be effective when the time lag can be considered
stationary. Following the same practical premise, the
combined effects of survey heading and anisotropic responses can be mediated by predicting unidirectional
data using Bayesian ridge regression or other ML methods. Repeated survey lines are thus used as training data
for the algorithm. Although the presented methodology
has its obvious shortcomings, such as limitations imposed by survey area heterogeneity, issues related to staggering can be overcome reliably and in a practical manner. Especially with array-type instrumentation, further
automation of both data acquisition and data processing,
alongside the continual increase of survey areas and data
collection speed, will only add to the need for adaptive
survey designs and strategies.
ACKNOWLEDGMENTS
We would like to thank Valentijn Van Parys, Marthe Smetryns, Benjamin Saey, and Timothy Saey for their help in
the field as well as tenant farmer R. Van Berkel and land
owner Staatsbosbeheer for granting access to the study
area. We also gratefully acknowledge the Dutch Heritage Agency Rijksdienst voor het Cultureel Erfgoed for
supporting this article. W. Waegeman received funding
from the Flemish Government under the " Onderzoeksprogramma Artificielë Intelligentie (AI) Vlaanderen "
Programme. " Philippe De Smedt is a postdoctoral fellow
of the Research Foundation-Flanders; research grant
FWO13/PDO/046.
AUTHOR INFORMATION
Daan Hanssens (daan.hanssens@ugent.be) received his
B.Sc. and M.Sc. degrees in bioscience engineering from
Ghent University, Belgium, in 2014 and 2016, respectively,
where he is currently pursuing his Ph.D. degree with the
MARCH 2021

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

Research Group on Soil Spatial Inventory Techniques.
His research focuses on the collection and processing of
frequency-domain electromagnetic data for very near
surface exploration.
Willem Waegeman (willem.waegeman@ugent.be) received his M.Sc. degree in computer science and his Ph.D.
degree in engineering from Ghent University, Belgium,
in 2004 and 2008, respectively. Since 2013, he has been
a professor at the Faculty of Bioscience Engineering,
Ghent University. His research interests include theoretical research in machine learning, data mining, and
statistics as well as the applications of these disciplines
in the life sciences.
Ynse Declercq (ynse.declercq@ugent.be) received her
master's degree in bioscience engineering from Ghent University, Belgium, in 2016, where she is currently finishing
her Ph.D. degree in environmental magnetism. Her research focuses on air pollution impacts, mainly by investigating soil magnetic and chemical fingerprints.
Hans Dierckx (h.dierckx@kuleuven.be) graduated as an
applied physics engineer from Ghent University, Belgium,
in 2006 and received his Ph.D. in physics in 2010. Since
2019, he has been an assistant professor at KU Leuven,
Belgium. His research interests include the investigation
of nonlinear waves and electrical phenomena, building
further on techniques from mathematical physics. He has
codeveloped a geometric framework for the dynamics of
electrical sources in the heart and is also interested in geophysical applications.
Henri Verschelde (henri.verschelde@ugent.be) is
with the Department of Mathematical Physics and Astronomy, Ghent University, Belgium. His research interests include quantum mechanics, field theories, and
special relativity.
Philippe De Smedt (philippe.desmedt@ugent.be) received his M.S. degree in archaeology and his Ph.D. degree in bioscience engineering from Ghent University,
Belgium, in 2013. His research interests include the
application of frequency-domain electromagnetic instrumentation for large-area prospection and paleoenvironmental reconstruction.
REFERENCES
[1] D. Butler, Near-Surface Geophysics. Tulsa, OK: Society of Exploration Geophysicists, 2005.
[2] I. Trinks et al., " Large-area high-resolution ground-penetrating radar measurements for archaeological prospection, " Archaeol. Prospect., vol. 25, no. 3, pp. 171-195, 2018. doi: 10.1002/arp.1599.
[3] M. E. Everett, " Theoretical developments in electromagnetic
induction geophysics with selected applications in the near
surface, " Surveys Geophys., vol. 33, pp. 29-63, Jan. 2012. doi:
10.1007/s10712-011-9138-y.
[4] L. Pellerin, " Applications of electrical and electromagnetic
methods for environmental and geotechnical investigations, "
Surveys Geophys., vol. 23, no. 2/3, pp. 101-132, 2002. doi:
10.1023/A:1015044200567.

181

IEEE Geoscience and Remote Sensing Magazine - March 2021

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2021