Professional Documents
Culture Documents
Geophysical Applications of Magnetic Sensors in Smartphones
Geophysical Applications of Magnetic Sensors in Smartphones
in smartphones
Nathan Campbell1, Estella Atekwana2, Adam J. Mathews3, and Ahmed Ismail1
https://doi.org/10.1190/tle39050312.1
Downloaded 06/26/20 to 165.225.34.131. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
1
Oklahoma State University, Stillwater, Oklahoma, USA. E-mail: nathanocampbell@gmail.com; ahmed.ismail@okstate.edu.
2
University of Delaware, Newark, Delaware, USA. E-mail: atekwana@udel.edu.
3
Western Michigan University, Kalamazoo, Michigan, USA. E-mail: adam.mathews@wmich.edu.
312 The Leading Edge May 2020 Special Section: Near-surface imaging and modeling
Table 1. Comparison of the magnetic field sensors in the smartphone and G858 according to sensitivity in The smartphone data were noisier
nanotesla and nanotesla per least significant bit, price, and maneuverability. compared to the G858. As such, the data
were transferred to MATLAB for appli-
Unit Sensitivity Price Maneuverability
cation of the redundant lifting scheme
Samsung Galaxy Note 4 Phone 600 nT/LSB US$350 High filter (Aghayan et al., 2017). The filter
G858 magnetometer 0.01 nT US$30,000 Low combines a Wiener filter for noise reduc-
Downloaded 06/26/20 to 165.225.34.131. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Special Section: Near-surface imaging and modeling May 2020 The Leading Edge 313
on the surface, it is assumed that the
unknown object was a buried piece
of metal.
The dipolar nature of the pipes is
also clearly seen in Figure 3a. For
example, pipe 3 has a magnetic high at
Downloaded 06/26/20 to 165.225.34.131. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
314 The Leading Edge May 2020 Special Section: Near-surface imaging and modeling
and much less expensive than the G858. The AndroSensor that are lined with plastic and 0.6 m of sand. It rises 30 m high,
application required little battery power, allowing the phone to and its depth extends 9 m below the surrounding ground level.
record data for several hours without needing to be charged. On See Figure 4 for details.
the contrary, the smartphone data involved more processing A walking survey was performed at the landfill in order to
time to achieve the desired results. compare the G858 and Android magnetometers. The first
The G858 is better at sensing subsurface magnetic targets. profile (line 1) began at the northern extent of the landfill,
Downloaded 06/26/20 to 165.225.34.131. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
This is to be expected as the sensitivity rating of the smartphone went south directly over the peak, and ended outside of the
is much less than the G858. The significance of the smartphone landfill to the southeast. The second profile (line 2) began at
results is that it was still able to image the pipes. Future work the western side and headed east directly over the peak and
involving the reduction of noise and elongation effects would past the landfill on the road. Because the smartphone was in
be beneficial. airplane mode to avoid magnetic interference from messages
or updates, a Bad Elf brand GPS receiver was carried during
Case study the survey.
The Stillwater landfill is an active municipal solid waste The G858 sensor was mounted on the front of an aluminum
landfill. At the end of 2014, the facility was estimated to have rod, while the smartphone was strapped face down with Velcro
accepted total waste of 3.25 million megagrams since 1968 to the back on a PVC pipe extension. The aluminum rod with
(Chen, 2016). It spans an area of 800 × 400 m, with the southern PVC extension measured approximately 2.5 m in length. The
half as the active section and the northern half as the completely shoulder strap was adjusted to let the sensors hover approximately
buried section. Internally, it is made up of eight rectangular cells 0.3 m above the ground. The frequency of acquisition was set at
2 Hz for both sensors.
The data sets were combined and graphed in Excel. To correct
for heading error in line 1, the smartphone data set was adjusted
vertically by uniformly adding 1200 nT. The same correction
was applied to line 2, except 1000 nT was added. These values
were determined by averaging the noise from the smartphone
when outside the landfill (Figures 5 and 6) and subtracting that
number from the G858 reading at the same location. The data
from both lines were corrected to show the TMI by removing
the IGRF.
Visual analysis shows that when on foot, the smartphone
performed remarkably well in imaging magnetic anomalies over
the landfill. Noise from the smartphone ranging from ±500 nT
dominates the end of the survey, where the G858 values flatten
out. These points, located approximately 550 m for Figure 5 and
375 m for Figure 6, mark the end of the landfill and a transition
into land without buried material below. Figure 7 provides a closer
look at line 1 between 250 and 400 m. This segment, starting at
the topographical high and moving southeast, contains the largest
anomalies on the line. When subjected to an intense magnetic
anomaly like a landfill, the smartphone sensor can reproduce the
results of the G858, but once outside the boundary, magnetic
noise from the phone overwhelms the data.
Quantitative analysis was used to determine how well the
Android performed compared to the G858. A scatter plot of the
two data sets (Figure 8) and the root-mean-square deviation
(RMSD) were calculated in Excel. RMSD is a measure of how
observations match a model. In this case, the observations would
be the Android, and the G858 would be the model. Line 1 had
1345 values and an RMSD of 480. Line 2 contained 1151 values
with an RMSD of 654. As seen in Figure 8, a strong linear trend
shows that the Android and G858 data were comparable over
the landfill.
Figure 4. A digital elevation model of the landfill survey area at 30 m resolution
There are a few possible sources of error in this survey.
created in ArcMap using data downloaded from the Shuttle Radar Topography
Mission (OpenTopography, n.d.). Walking profiles are shown in black. Line 1 went Additional heading error can be seen throughout Figures 5 and 6,
from north to south and line 2 from west to east. The star indicates the unmanned where the smartphone signal is consistently above or below the
aerial vehicle staging area for takeoff and landing. G858 signal. Since the landfill had many obstacles such as berms
Special Section: Near-surface imaging and modeling May 2020 The Leading Edge 315
and gas vents, the direction was not always north/south or east/ in less-developed countries have a smartphone that can be used
west. Turns were required to avoid these places and created small for geophysical education. Smartphones are easy to use and most
frequency heading errors throughout. One place this can clearly people already own one.
be seen is in Figure 7 at 420 m, where the average reading from
the smartphone jumps up 1000 nT. This is due to a turn in the Data and materials availability
road, where the line was walked from an eastbound direction Data associated with this research are available and can be
Downloaded 06/26/20 to 165.225.34.131. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
to a northbound direction. Additionally, the aluminum rod was obtained by contacting the corresponding author.
not always aligned perpendicular to the walking direction.
Therefore, the G858 was sampling a spot on the landfill that Corresponding author: nathanocampbell@gmail.com
was at times up to 0.5 m either left or
right of the smartphone. This only
occurred in places that were very steep.
Also, sensor height became difficult to
control on the more challenging sec-
tions of the landfill.
Conclusion
The smartphone was able to detect
short steel pipes buried up to 0.75 m.
When compared to the G858, results
for the smartphone as an underground
pipe detector are mixed. The pipes were
delineated from the surrounding envi-
Figure 5. Line 1 from the walking survey of the landfill comparing the G858 with the smartphone. It begins at the
ronment, but because it was a magneti- northern extent of the landfill and continues south and southeast. Arrows indicate limits of the landfill. IGRF was
cally quiet area, noise dominated the removed to show TMI.
smartphone sensor outside of a 1 m
radius of the pipes.
When performing a walking survey
over a strong magnetic anomaly such
as a landfill, the smartphone magne-
tometer collects data that are compa-
rable to the G858 magnetometer. In
magnetically quiet conditions, where
there is little magnetic material to
influence the sensor, noise from other
components of the phone dominate the
signal, and the relatively low sensitivity
of the sensor becomes apparent.
Anomalies less than 600 nT in strength Figure 6. Line 2 from the walking survey of the landfill comparing the G858 with the smartphone. It begins
are likely to be lost in the noise unless at the western extent of the landfill and extends eastward. Arrows mark the limits of the landfill. IGRF was
the sensor is above a large magnetic removed to show TMI.
anomaly such as a landfill.
Technology enhances our under-
standing of the world. As sensors con-
tinue to shrink in size, require less power
to operate, and diminish in cost, it is
important to consider the advantages
and disadvantages of gaining portability
while losing sensitivity. Smartphones
can image anomalies in the subsurface
such as pipes, landfills, and dikes. They
also perform well enough to act as a
presurvey device before considering
investing money in a professional mag-
netic survey. Additionally, most students Figure 7. Segment of line 1 comparing the G858 and smartphone from 250 to 400 m. IGRF was removed to show TMI.
316 The Leading Edge May 2020 Special Section: Near-surface imaging and modeling
Downloaded 06/26/20 to 165.225.34.131. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Figure 8. Scatter plot for line 1 of the Android and G858 data over the landfill. The dotted line
and equation represent the linear line of regression.
References
Aghayan, A., P. Jaiswal, and E. Atekwana, 2017, Potential field data processing using
the redundant lifting scheme: 89th Annual International Meeting, SEG, Expanded
Abstracts, 1797–1802, https://doi.org/10.1190/segam2017-17792006.1.
Asahi Kasei, 2014, AK09911 3-axis electronic compass: Asahi Kasei Manual.
Chen, M., 2016, Evaluation of Title V permit renewal application no. 2014-1556-
TVR: Oklahoma Department of Environmental Quality, Air Quality Division.
Ferster, C. J., and N. C. Coops, 2013, A review of earth observation using mobile
personal communication devices: Computers & Geosciences, 51, 339–349, https://
doi.org/10.1016/j.cageo.2012.09.009.
Fiv Asim, 2015, AndroSensor, http://www.fivasim.com/androsensor.html, accessed
24 March 2020.
Geology.com, n.d., Oklahoma physical map, https://geology.com/topographic-
physical-map/oklahoma.shtml, accessed 26 March 2018.
Geometrics, 2001, G-858 MagMapper: Geometrics Manual.
Heran, W. D., G. N. Green, and D. B. Stoeser, 2003, A digital geologic map database
for the state of Oklahoma: United States Department of the Interior.
Ndiaye, M., and A. Diagne, 2014, Geomagnetic investigation method using iPhone
integrated magnetic sensor: International Journal of Geosciences, 5, 1–4, https://
doi.org/10.4236/ijg.2014.51001.
OpenTopography, n.d., https://opentopography.org.
Suksmono, A. B., D. Danudirdjo, A. D. Setiawan, and D. Rahmawati, 2017, Magnetic
subsurface imaging systems in a smartphone based on the built-in magnetometer:
IEEE Transactions on Magnetics, 53, no. 11, https://doi.org/10.1109/
TMAG.2017.2697001.
Special Section: Near-surface imaging and modeling May 2020 The Leading Edge 317