ArticlePDF Available

Usage of UAVs for Surveying and Monitoring Icebergs (Essay)

Authors:

Abstract and Figures

Icebergs pose a hazard to shipping and to offshore oil and gas activities. To mitigate and better manage the risks, it is essential to measure, survey, and monitor them. Collecting in-situ measurements and data from icebergs is challenging. Icebergs frequently roll and/ or break-up. A vessel must maintain a safe separation distance from an iceberg. The margin of safety is calculated to be either one iceberg length or twice the height (whichever is estimated to be the larger number) from the iceberg, thus the bigger the iceberg the greater the separation distance. Given this, it is not feasible to place personnel onto icebergs to make direct measurements. All surveying and measuring must be performed from a distance. Remotely operated unmanned aerial vehicles, commonly know as UAVs, can bridge that gap. In this essay, we summarize our investigations over the past four years to develop both the technology and the operational protocols to effectively use UAVs to survey and monitor icebergs in the harsh environment of the North Atlantic.
Content may be subject to copyright.
46
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
Iceberg 3033 from the 2019 field program. It is
calculated to have a length of 142 m (along the
longest axis), a height of 45 m, and a mass of
around 710,000 tonnes (as a comparison, a cube of
water with 50 m sides – a typical Olympic swimming
pool – would have a mass 500,000 tonnes).
Copyright Journal of Ocean Technology 2020
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
47
C-CORE
Copyright Journal of Ocean Technology 2020
48
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
Copyright Journal of Ocean Technology 2020
Icebergs pose a hazard to shipping and to
offshore oil and gas activities. To mitigate
and better manage the risks, it is essential to
measure, survey, and monitor them. Collecting
in-situ measurements and data from icebergs
is challenging. Icebergs frequently roll and/
or break-up. A vessel must maintain a safe
separation distance from an iceberg. The
margin of safety is calculated to be either one
iceberg length or twice the height (whichever
is estimated to be the larger number) from the
iceberg, thus the bigger the iceberg the greater
the separation distance. Given this, it is not
feasible to place personnel onto icebergs to
make direct measurements. All surveying and
measuring must be performed from a distance.
Remotely operated unmanned aerial vehicles,
commonly know as UAVs, can bridge that gap.
In this essay, we summarize our investigations
over the past four years to develop both the
technology and the operational protocols to
effectively use UAVs to survey and monitor
icebergs in the harsh environment of the
North Atlantic.
Evolution of the UAV for Icebergs Program
Three key requirements were initially defined
when the program began in the summer of
2016. The first was the development of a
UAV and associated mechanism to deliver
GPS tracking units (GTU) onto icebergs
to gather iceberg position data remotely.
Positional data is important for developing
models that can predict the drift of an iceberg
and for improving the detection algorithms
used, for example, by the International Ice
Patrol to produce iceberg population products.
The second requirement was to explore
the feasibility of using UAVs to acquire
photographic datasets that, through application
of photogrammetric techniques, estimate
important physical characteristic such as
length, height, and mass of an iceberg. The
final requirement was the development of a
UAV mounted Ice Penetrating Radar (IPR)
system to estimate the draft of an iceberg – the
draft being the distance from the waterline to
the base of iceberg. Characteristics such as
length, height, draft, and mass are essential to
populate and develop probabilistic models of
iceberg distributions to refine offshore structure
design and evaluate risk to subsea assets.
The UAVs, supporting technologies, and
operational methodologies have been
developed through frequent overland testing
and refined in offshore deployments. We
have operated from four different vessels
(Figure 1). The first deployment was over
the course of two days in June 2017, and
took place from a 9 m research vessel. We
tested the GTU delivery system and the
photogrammetry technologies. In July 2018,
we operated from a 75 m supply vessel
and spent 10 days offshore. We refined the
delivery system and the photogrammetric
methods, and undertook proof of concept
flights with the UAV mounted IPR. During
the 2019 iceberg season, we performed a
campaign that was focused on the operational
delivery of GTUs and the collection of
photogrammetric datasets. In total, the
campaign spent 52 days at sea, split over four
voyages. We operated from a 62 m Coast
Guard cutter and a 32 m crab fishing vessel.
Operating from these platforms in a range of
nearshore and offshore environments, and the
associated meteorological and oceanographic
conditions, greatly enabled us to refine our
technologies and the operating procedures.
Deploying Sensors onto Icebergs
Timestamped locations of icebergs are a
critical data requirement for developing iceberg
drift forecasting models and for refining
iceberg detection techniques. Small (smaller
than a typical smartphone) battery powered
standalone GTUs that transmit location data
via satellite to a web site are common and
cheap (about $100). By placing a GTU onto
an iceberg, one can theoretically collect a
sequence of positions, a so-called iceberg track.
The GTU design we have developed to obtain
iceberg tracks is shown in Figure 2. It is made
from a commercially available GPS asset
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
49
Copyright Journal of Ocean Technology 2020
tracking device secured to the top of a rubber
mat. The mat has toothed metal plates riveted
onto the base which bite into the ice surface.
In addition, the rubber sheet is flexible so that
it conforms to the surface it lies upon, further
helping to maximize surface friction on the,
potentially sloping, ice. It is programmed to
transmit its position every 30 minutes and can
last over 30 days.
Placing a GTU onto an iceberg is a non-trivial
task. As stated in the introduction, a vessel
must stay a safe separation distance from an
iceberg. Using a UAV to cross this divide
requires a number of logistical and technical
issues to be overcome. Most importantly,
to land (or touch down) a UAV requires a
level surface of sufficient size that is clear of
obstructions or walls or overhangs (comparable
to the requirements for helicopters, but of
dimensions appropriate to the smaller UAV).
Icebergs frequently do not have level surfaces
and frequently have walls and obstructions.
If there is a suitable area that is level and
free from obstructions, given that the iceberg
pitches and rolls, it will likely not be level
for long. To address this, at the beginning of
the project we decided that we did not want
to attempt to land the UAV onto the iceberg
to place the GTU; rather, we would lower the
GTU on a tether, then disconnect the tether
from the UAV leaving the GTU on the iceberg.
Having a physical separation between the
iceberg and UAV greatly reduced the risk of
the UAV colliding with the iceberg (however,
the risk is still present).
The mechanism to implement the tether system
has evolved over the course of program.
This was primarily due to our experiences
in the field that improved our understanding
of which design requirements had a stronger
impact on successfully deploying the GTUs.
To a lesser degree, the design also evolved
as commercially available components for
modifying UAVs have become available,
meaning we could rely less on custom payload
design and the associated manufacturing costs.
Figure 1: Vessels from which we have performed flight operations (top left) Knave I ; (top right) Atlantic Eagle ; (bottom left)
Juniper, and (bottom right) Patrick and William.
KEN KEEPING, MARITIME SURVEY SERVICES LTD.
DEAN PORTER
C-CORE
C-CORE
50
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
Copyright Journal of Ocean Technology 2020
The latest iteration of the two-stage release
sequence is shown in Figure 3. The system is
two-stage because it takes two discrete actions
on the part of the flight crew to release the
GTU. The GTU is attached by a short and a
long tether. The short tether secures the GTU
snug under the UAV during flight. Stage one
triggers a servo that releases the short tether.
The GTU drops until it is hanging freely under
the UAV by the long tether. The pilot then
reduces altitude to lower the GTU onto the
surface of the iceberg. Once the GTU is on the
iceberg and the line is de-weighted, the rubber
mat of the GTU unfurls. Stage two activates a
second servo to release the long tether leaving
the GTU on the iceberg.
The Delivery UAV is a custom build that
follows the typical quadcopter design (shown
in Figure 3 top, left). It is designed around
a “stick” frame that is in the shape of an
X. The motor and propellers are mounted
at the end of the arms and the rest of the
system (batteries, flight controller, telemetry,
cameras, release mechanisms, etc.) are in
the centre. Each year, as newer improved
components have become available, we have
integrated them into the design to improve
the performance. The UAV has an endurance
of around 15 minutes (reduced under strong
wind conditions and is payload dependent),
which is more than ample to deploy a sensor.
In 2019 we developed both a floating variant
that was built inside a waterproof hull as well
as the standard “stick” frame. The objective
with the floating frame was that, in case of a
technical failure, we could land the unit in the
ocean and then recover it. The stick frames
are faster and cheaper to produce and have
proven to be a reliable design.
In the 2019 field campaign, we successfully
deployed 119 tracking units. The map
presented in Figure 4 highlights the extent
and coverage of this data collection effort. It
shows the voyage trajectories, the icebergs
that were logged via observation, and the
GPS tracks of the icebergs that had a GTU
deployed onto them. The maximum track
duration was 356 hours (nearly 15 days), the
mean and median duration was 46.3 hours,
and 11.7 hours, respectively.
Photogrammetry
Photogrammetry is the process of
extracting measurements from photographs.
Figure 2: The design of the GPS tracking unit (GTU): (left) The top of the GTU. The GPS tracker is the small white unit labelled 6385. It is
wrapped in foam which is attached to the mat. The foam provides protection from hard deployments. (right) The underside of the tracker
showing the teeth that “bite” into the ice to provide friction.
C-CORE
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
51
Copyright Journal of Ocean Technology 2020
Figure 3: Sequence of images to visualize the delivery of a GTU onto an iceberg: (top, left) the underside of the UAV showing the dual-servo
system; (top, right) mission crew loading the GTU onto the UAV; (middle, left) UAV and GTU waiting to take-off; (middle, right) prototype UAV
during overland based test flying with GTU affixed underneath; (bottom, left) test flying after stage one release; (bottom, right) UAV after stage
two release leaving the GTU deployed on an iceberg.
C-CORE
52
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
Copyright Journal of Ocean Technology 2020
Figure 4: Map showing the data gathered from the 2019 iceberg tagging campaign.
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
53
Copyright Journal of Ocean Technology 2020
In the context of surveying icebergs,
photogrammetry is used to generate 3D
reconstructions from which the length,
width, height, and the volume of the topside
(or sail) of the iceberg can be computed.
Using the topside volume and a simple
buoyancy calculation, the mass of the
iceberg can be estimated.
To acquire the photographs for the 3D
iceberg reconstructions, we use high-end
“prosumer” UAV systems. During the 2017
and 2018 field programs, we used a DJI
Mavic UAV. The UAV was modified by
adding floats, which would allow us to land
on water, if required. The modified system
worked well under low wind conditions (<15
km/hr), but once wind velocities increased,
the floats acted as a sail and negatively
impacted the UAV’s ability to operate in the
higher wind velocities. As our operational
experience increased, more time was spent
flying in stronger wind conditions and the
floats were removed. During the 2019 field
program, we used the Autel EVO UAV, which
had very similar performance characteristics
(both flight behaviour and camera) to the DJI
Mavic and a slightly lower price point.
To acquire the iceberg photograph dataset,
the UAV operator circles the iceberg,
keeping it centred in the camera image.
An image is acquired every three to five
seconds. Usually the operator would acquire
a high angle set of photographs and a low
angle dataset. This improves the coverage of
the internal and external faces of the iceberg.
Acquiring good datasets requires a certain
degree of skill from the operator, especially
in higher wind conditions.
The size of an iceberg determines the duration
of the flight. A dataset for a small iceberg can
be acquired in a few minutes. For a larger
iceberg, it can be close to the endurance limit
of the UAV (usually around 20-25 minutes
depending on wind speed). For some of the
largest icebergs, we had to perform two flights
to collect the complete dataset.
We use a commercial photogrammetry
software package to compile the photographs
into a 3D point cloud. The 3D point cloud
is post-processed to clean the optical
artifacts (such as removing the vessel in
the background) and to turn the points into
a closed 3D surface, from which physical
characteristics can be computed. Samples of
the 3D point cloud and the closed surface can
be seen in Figure 5.
In 2019, we collected nearly 86
photogrammetric datasets from which
we were able to yield 74 useful 3D
reconstructions. The visibility at the time
of acquisition can impact the quality of the
photographs. This can make it difficult to
create the 3D reconstructions, especially with
icebergs that have limited features.
Measuring the Draft of an Iceberg
Ice Penetrating Radar (IPR) is a geophysical
remote sensing tool and technique that is
used to determine the thickness of ice. It is
traditionally used for measuring the thickness
of glaciers, lake and river ice, and ice roads.
The IPR transmitter emits a pulse of high-
frequency electromagnetic waves into the
ice below the IPR. At the interface between
different materials, such as ice to ocean where
the electromagnetic properties change abruptly,
the signals are reflected. A receiving sensor
records the reflected waves. The IPR system
then processes the transmitted and received
signal and computes the amplitudes and travel
times. By knowing the travel time and the
speed of the signal through ice, the distance
between the IPR and the reflecting surface can
be inferred and hence the thickness. The height
above sea level can be subtracted from the
thickness to compute draft.
The IPR used for this investigation was a
modified version of a small, man-portable ice
penetrating radar system used for glacier ice
thickness measurements. In its unmodified
configuration for glacier surveying, it
weighs about 10 kg with sufficient battery
capacity for around eight hours of surveying
Figure 5: Screenshots from the photogrammetry workflow (the same iceberg as shown in the opening photo of this essay): (left)
photogrammetry software showing the 3D iceberg and the individual photographs located in space at their time of acquisition and (right) the
reconstructed surface from which measurements such as length and height can be made.
54
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
Figure 6: Prototype UAV
and Ice Penetrating Radar
leaving the back deck of the
vessel to survey the iceberg
in the background. The slim
tubes underneath the UAV
are the radar antennas.
C-CORE
Copyright Journal of Ocean Technology 2020
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
55
Copyright Journal of Ocean Technology 2020
and requires two technicians to operate it.
The first technician hauls the transmitter
hardware and transmitting antennas
(mounted to a sled) along the surface, and
the receiver and receiving antennas are
dragged along behind in another sled by
a second person. The largest form-factor
components in the design are the transmitting
and receiving antennas and their separation
distance. Fundamentally, larger antennas
(and larger separation distances) allow lower
frequencies which allow deeper penetration
depths. For the glacier configuration (able to
measure as deep as 750-800 m), the antennas
are 8 m in total length.
Antenna length and the onboard computing
configuration were the primary components
that had to be modified so that the IPR could
be carried on a UAV. For the UAV IPR (seen
in Figure 6), we reduced the antenna lengths
to 86 cm which, in theory, gave us the ability
to measure icebergs up to 150 m thickness.
We had to modify the IPR computer so that it
could be controlled wirelessly using a tablet
computer. The IPR is powered from the UAV
power supply; in comparison to the battery
requirements to power the UAV, battery
consumption for the IPR is minimal. The UAV
was a heavy-lift version of the delivery UAV.
The output from the IPR is visualized as a
radargram. An annotated radargram from data
collected in 2018 is shown in Figure 7 (top). To
aid in an understanding of the radargram and
how the data was collected, the photograph
annotated with a flight planning sketch that
was used during the field program is shown
in Figure 7 (bottom, left). The flight goal was
that the UAV would fly from the vessel out
over and beyond the iceberg in as straight a
path as possible. The yellow/grey colouring
throughout the radargram represents the
amplitude of the signal. Strong reflections are
shown as brighter yellow/grey lines. The ocean
and metal of the vessel generate a lot of noise.
The horizontal x-axis represents the distance
flown. The vertical y-axis represents the travel
time, which is subsequently converted to a
thickness estimate by assuming a constant
velocity through the ice. Significant experience
with IPR systems in conjunction with ancillary
data (timing of leaving the vessel, arrival at
the iceberg, etc.) is required to interpret the
radargram. The analysis identified a single
reflection that is interpreted to be the base
of the iceberg. Applying a constant velocity
for the radar signal indicates a thickness of
about 60 m at this section of the iceberg. This
can be compared with the characteristics
shown in Figure 7 (bottom, right). This
dataset, of the same iceberg, was generated
by the Smart Iceberg Management System
(SIMS) described in the essay Development
of a Decision Support Tool to Aid Iceberg
Management Operations in this issue of the
JOT. The SIMS equipment – a vessel mounted
integrated LiDAR/sonar system from which
full 3D profiles of icebergs can be made – and
project team were frequently on the same
voyages enabling comparison dataset of the
same iceberg to be made. In this case, the
SIMS data indicate a thickness of about 60-65
m, which is comparable to the IPR derived
estimate of thickness.
We still consider the UAV IPR a prototype
technology. To acquire good data, a very
straight flight path must be flown, which
is challenging in windy conditions. The
smaller the iceberg or the more complex the
geometry increases the number of reflections,
making interpretation more difficult. The
technology would be well suited to large
tabular icebergs where the length and width
is much greater than the thickness or for
flying over the ice-shelves from which the
icebergs are calved.
Conclusions
UAVs have proven to be a valuable tool to
survey and monitor icebergs. They have
enabled us to deploy sensor packages that,
previous to this program, could only have
been achieved via helicopter which would
have been a much greater cost and risk.
Moreover, we have been able to deploy GTUs
onto over 100 icebergs in a single season.
56
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
Copyright Journal of Ocean Technology 2020
Figure 7: (top) Radargram of an iceberg with annotated interpretations; (bottom, left) field photograph with flight planning sketch to show
ideal flight overpass to acquire Ice Penetrating Radar (IPR) data; (bottom, right) full 3D reconstruction of the same iceberg used to compare
with the IPR measurements.
This is a step change in data collection. They
have allowed us to position optical camera
systems and radar sensors over an iceberg and
to derive dimensional measurements.
The same tools and techniques can be used
to deploy alternative sensor packages onto
or over icebergs or in other hard-to-reach
and challenging environments. For example,
we have developed an iceberg motion
accelerometer sensor package which is similar
to the form factor and weight of the GPS
tracking unit and so is transportable by the
delivery UAV. This will be deployed at the
next opportunity. Similarly, during the 2017
field campaign, we performed an opportunistic
deployment of a GTU onto a sea-ice floe
which gave us a sea-ice drift track. The UAV
IPR could be used in a glacial or river ice
environment where it would be unsafe to
deploy people.
However, UAVs do come with their own
set of challenges and difficulties which are
compounded when operating from a vessel
in an offshore environment and even further
compounded when operating around icebergs.
The most critical lesson we have learned is
that one has to plan for and expect a UAV to
be damaged or lost when operating in a harsh
environment. During the first phase of this
project, we were torn between working with
large, powerful, and expensive (~ $100K)
UAV systems and smaller cheaper systems
(~$1-10K). As we have spent more time out on
the ocean working with the icebergs, we have
learned that incidents will occur. For example,
chaotic wind fields around an iceberg can
unexpectedly cause the UAV to lose altitude
or lurch, sometimes by many metres, which
can induce collisions; we never considered
the fact that seabirds would find the UAV a
threat and behave aggressively toward them –
The Journal of Ocean Technology, Vol. 15, No. 3, 2020
57
Copyright Journal of Ocean Technology 2020
if a seagull did attack, it is likely that neither
the bird nor the UAV would survive. To cope
with these risks, one can either avoid flying in
all but the most benign circumstances which
is an ineffective operational data collection
strategy, or one can accept the risk to the
UAV and design and plan for it. We feel that
our operational approach of developing the
UAVs to be as simple and as functional as
possible and making multiple units has proven
the correct path. Ship and personnel time are
expensive; UAVs much less so. Adopting this
approach, when we have damaged or lost a
UAV, we simply pulled another UAV out of
the ship’s hold and continued on. This attitude
allowed us to target more of the challenging
icebergs and increased the window of met-
ocean conditions within which we would
attempt to could collect data. Future work
will focus on further reductions in cost and
increased autonomy.
u
Acknowledgements
This program would not have been possible
without funding provided by the Department
of Tourism, Culture, Industry, and Innovation
of the Government of Newfoundland and
Labrador, members of the local offshore
industry, C-CORE, and the United States
Department of Homeland Security Office of
Science and Technology under an Interagency
Agreement with the United States Army Combat
Capabilities Development Command C5ISR
Center NVESD. The authors also acknowledge
the support and professionalism of the captains
and crew of Knave I, Atlantic Eagle, USCGC
Juniper, and Patrick and William.
Dr. Robert Briggs is a Research
Engineer and Scientist at C-CORE
with over 20 years’ industrial and
academic experience. He holds a
PhD in glacial system dynamics,
a M.Sc. in oceanography, and a
M.Eng. in systems engineering.
Due to his diverse and multi-
disciplinary background, he
works on projects that crosscut and integrate the disciplines
of engineering and natural science. Usually this involves
developing tools and technologies to collect, analyze, and
distribute data from harsh environments to solve real-world
operational problems.
Carl Thibault is the founder of
MakeTech Aerospace, which
develops UAV technology
for industrial applications in
harsh environments. He has
a bachelor’s in mechatronics
and a master’s in electrical and
computer engineering from the
University of New Brunswick. He
was the first professional UAV operator and developer in
Atlantic Canada.
Laurent Mingo is a broadly
experienced systems engineer
with over 20 years’ experience in
data acquisition and management,
and system integration. Through
his company, Blue System
Integration, he develops portable
ice-penetrating radar systems for
cryospheric studies. His systems
are now regularly used by many research teams around the
world, from the large ice-sheets found in Antarctica and
Greenland, to ice-islands in the Canadian Arctic, and other
glaciers in North America, Europe, Asia, and South America.
Tony King has 20 years’
experience in the ice engineering
field and has been director of
C-CORE’s Ice Engineering Group
for the past eight years. He has
a background in risk analysis,
probabilistic methodologies,
and numerical modelling, and
has performed ice risk analyses
for subsea assets in most ice-prone regions of the world.
He obtained both an undergraduate and master’s from
Memorial University’s engineering program.
... The major improvements are: implementation of gradient-boosted trees instead of NN, and a hybrid approach instead of a "black box" model. In addition, the model is trained on new GPS iceberg tracks collected during an iceberg tagging campaign off the coast of Labrador in 2019 (Briggs et al., 2020). ...
... The iceberg GPS tracks were collected during a tagging campaign offshore Newfoundland and Labrador in 2019. Unmanned Aerial Vehicles (UAVs) were used to deliver the GPS trackers from the vessel to the icebergs (Briggs et al., 2020). Each of the trackers consisted of a GPS unit attached to a rubber mat with toothed metal plates. ...
Conference Paper
Full-text available
A machine learning approach has been developed to improve iceberg drift forecasting on a tactical timescale to mitigate potential impacts between icebergs and offshore facilities in the North Atlantic. The problem of iceberg forecasting involves many uncertain terms. Although iceberg drift trajectories can be measured with sufficient accuracy, the iceberg shapes are generally unknown. Moreover, the corresponding metocean forecasts, in particular ocean current forecasts, carry their own uncertainties. The forecasting error for this highly uncertain drift process can be minimized using a hybrid approach that calculates Coriolis acceleration explicitly and estimates combined hydrodynamic and wind drag accelerations using machine learning. The latter is an implementation of gradient boosted trees algorithm that predicts and integrates the unknown accelerations. The training dataset consists of iceberg GPS tracks obtained during 2019 iceberg data collection campaign offshore Labrador. The trained model is cross-validated using the k-folds approach. The evolution of the average error between the observed and forecast tracks is used as the measure of performance. The resulting error curve is promising and can be improved even further if more, and more accurate in-situ data are collected. The trained model can be used operationally, together with conventional dynamic approach, as an additional source of information in order to determine an adequate response to iceberg threat.
... Mapping icebergs, ice islands, and ice floes and estimating their deterioration and melt rates are active areas of research. Some of the notable works that have made strides in these areas include [74,26,25,19,135,11,111]. Our work builds on this rich body of work to create detailed 3D models of icebergs multiple times and calculate accurate volume loss rate estimates along with the measurement uncertainties. ...
Thesis
Full-text available
Most approaches to visual navigation make multiple assumptions about the scenes being imaged. There are implicit assumptions about the scene being predominantly static and the availability of well illuminated, texture rich, objects in the scene. In some cases these assumptions severely limit or eliminate the full applicability of Visual Simultaneous Localization and Mapping (VSLAM) and Structure from Motion (SfM) methodologies. This dissertation attempts to address problems where the assumptions of static scenes and texture rich objects are not valid. Motivated by the application of mapping rotating and translating icebergs, we propose a system level solution for addressing the problem of mapping large, low contrast, moving targets with slow but complicated dynamics. Our approach leverages the complementary nature of multiple sensing modalities and utilizes a rigidly coupled combination of a subsurface multibeam sonar (a line scan sensor) and an optical camera (an area scan sensor). This allows the system to exploit the optical camera information to perform iceberg relative navigation, which can be directly used by the multibeam sonar to map the iceberg underwater. To compensate for the effect of low contrast we conduct an in-depth analysis of features detectors and descriptors on end-to-end SfM algorithms to demonstrate and understand how methodologies such as Contrast Limited Adaptive Histogram Equalization (CLAHE) and Zernike Moment descriptors help improve the overall accuracy in these challenging applications. We merge these approaches into an algorithmic framework that allows us to compute the scale of the navigation solution and iceberg centric navigation corrections. These corrections can then be used for accurate iceberg reconstructions. This enables a quantitative analysis of our iceberg mapping efforts including volume estimation and change detection. We successfully demonstrate our approach on real field data from three of the icebergs surveyed multiple times during the 2018 and 2019 campaigns to the Sermilik fjord in Eastern Greenland. Availability of iceberg mounted Global Navigation Satellite System (GNSS) observations during these research expeditions also allowed for a comparison of this approach against ground truth, providing additional confidence in the systems level mapping efforts. The accuracy of the reconstructions is demonstrated by estimating iceberg volumes, calculating their ablation rates, and performing change detection at a granular scale.
ResearchGate has not been able to resolve any references for this publication.