4.

MARINE SAMPLING FIELD MANUAL FOR

AUVS (AUTONOMOUS UNDERWATER
VEHICLES)

Jacquomo Monk*, Neville Barrett**, Tom Bridge, Andrew Carroll, Ariell

Friedman, Nicole Hill, Daniel Ierodiaconou, Alan Jordan, Gary Kendrick,
Vanessa Lucieer

* Jacquomo.monk@utas.edu.au
** neville.barrett@utas.edu.au

Chapter citation:
Monk J, Barrett N, Bridge T, Carroll A, Friedman A, Ierodiaconou D, Jordan A, Kendrick G,
Lucieer V. 2018. Marine sampling field manual for autonomous underwater vehicles (AUVs). In
Field Manuals for Marine Sampling to Monitor Australian Waters, Przeslawski R, Foster S
(Eds). National Environmental Science Programme (NESP). pp. 65-81.
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

4.1 Platform Description

Autonomous Underwater Vehicles (AUVs) are untethered robotic platforms that operate
independently to complete pre-determined surveys. The endurance of AUVs typically range from
hours to several days (Huvenne et al. 2018). However, with the rapid development of battery
technology long-period deployments ranging from weeks to months are now possible (Furlong et al.
2012; Hobson et al. 2012). Maximum operational depths range from a few hundred metres for the
smaller vehicles (Wynn et al. 2014) to over 6000 m for larger units (Huvenne et al. 2009).

Huvenne et al. (2018) classify AUVs as either "cruising" or "hovering" vehicles (Figure 4.1). Cruising
AUVs are traditionally torpedo-shaped, driven by a single propeller at speeds up to 2 ms-1, and are
optimised to cover large distances along pre-designed survey tracks (Wynn et al. 2014). These
cruising AUVs are usually not well suited to photographically surveying high-relief seabed terrain
due their lack of vertical agility. Traditionally, cruising AUVs are the main type of AUVs used in the
commercial world, with prominent scientific examples including the Autosub series from the National
Oceanography Centre (UK), the AsterX and IdefiX from French Research Institute for Exploitation of
the Sea (IFREMER; France) and the Dorado series from Monterey Bay Aquarium Research
Institute (USA) (Furlong et al. 2012; Rigaud 2007). By contrast, hovering AUVs are equipped with
several propellers, which facilitate multi-directional manoeuvrability capabilities, similar to a remotely
operated vehicle (ROV). Hovering AUVs are designed for precision operations, slow motion surveys
(e.g. seabed photography) and work in distinctly 3-dimensional terrains, such as around high-relief
reefs (Williams et al. 2012). Among the best-known scientific examples of hovering AUVs are ABE
and Sentry from Woods Hole Oceanographic Institute (USA) (e.g. Tivey et al. 1998; Wagner et al.
2013) and Sirius from Australian Centre for Field Robotics (Australia) (e.g. Bewley et al. 2015;
Williams et al. 2016; Williams et al. 2012).

Depending on the size of an AUV they can be equipped with a range of sensors such as
conductivity, temperature, depth, acoustic doppler current profilers, chemical sensors, photo
cameras, sonars, magnetometers and gravimeters (Connelly et al. 2012; Sumner et al. 2013;
Williams et al. 2010). Importantly, on-board battery capacity is the primary limitation to the number
of sensors and survey duration for AUVs. Furthermore, AUVs are currently not yet equipped for
extensive physical sampling of seabed or fauna, although sampling of the water column can be
achieved (Pennington et al. 2016). Overall, AUVs are more suited for survey operations, acquiring
sensor data along pre-programmed transects, while ROVs are optimal for high-resolution, highly
detailed and interactive work, including high-definition video surveying and physical sampling. An
extensive review of the use and capabilities of AUVs for geological research was recently published
by Wynn et al. (2014). There is, however, no equivalent review discussing the capabilities of AUVs
for ecological research (but see section 3.3 in Wynn et al. 2014; Durden et al. 2016).

This document focuses on hover class AUVs can control their position and heading at very low
speeds, which makes them suitable for operations over rough terrain while maintaining an
appropriate altitude for imaging small scale targets. When equipped with navigational sensors such
as GPS, Ultra Short Baseline Acoustic Positioning System (USBL), acoustic doppler profiler, and
forward-looking obstacle avoidance sonar, hover class AUVs enable precise tracking along the pre-
programmed routes. These characteristics make them particularly suited to collecting highly detailed
sonar and optical images over high-relief seabed terrain, which can be geo-referenced with high
precision. These can then be stitched together into photomosaics to focus on large features or
specific details on the seafloor.

While most of the well-known AUVs used in scientific research are custom built, technological
developments over the last five years have seen a number of ready-built, commercial units
becoming available, with examples such as the cruising Iver and hovering Subsea 7 AUVs. The
release of these units into the market will likely increase the uptake of AUVs for scientific research.
Page | 66
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Figure 4.1 Examples of AUV classes. Left: an example of the cruising class AUV Nupiri muka operated by the University
of Tasmania (photo credit: Damien Guihen). Right: an example of the hovering class AUV Sirius operated by Australian
Centre for Field Robotics for Integrated Marine Observing System (Photo credit: Asher Flatt).

4.2 Scope
The primary aim of this field manual is to establish a consistent approach to marine benthic
sampling using AUVs and facilitate statistically sound comparisons between studies. This manual
will focus on hover class AUVs designed to survey the seabed due to their proven use in marine
benthic monitoring compared to other marine imagery platforms (described in next section of this
chapter). It will not consider cruising class AUVs. The scope of the manual is to cover everything
required from equipment, pre-survey preparation, field procedures and post-survey procedure for
using hover class AUVs to photographically survey seabed assemblages found on Australia’s
continental shelf regions. Deep-sea environments are currently excluded from this field manual as
we do not currently have an AUV in Australia capable of image-based surveys at these depths.
Although it should be noted that AUV-based photographic surveys of the deep-sea benthos have
been successfully undertaken internationally (e.g. Morris et al. 2014; 2016; Milligan et al. 2016).

4.3 AUVs in Marine Monitoring

Application of AUVs for monitoring benthic marine ecosystems has experienced a rapid increase
over the past two decades. Researchers have used hover class AUVs in monitoring the impacts of
invasive species (Ling et al. 2016; Perkins et al. 2015), for ecosystem-based fisheries management
(Smale et al. 2012), assessing population trends in demersal fishes (Clarke et al. 2009; Seiler et al.
2012), mapping of benthic habitats (Lucieer et al. 2013), examining diversity in reef communities
(Bridge et al. 2011; James et al. 2017; Monk et al. 2016), changes in structural complexity of coral
reefs (Ferrari et al. 2016a, b), and mapping the spatial and depth extent of kelp forests (Marzinelli et
al. 2015).

Compared to other marine imagery platforms (e.g. towed systems), hover class AUVs have several
strengths applicable to marine monitoring:

• They navigate precisely defined flight paths and the geolocation of individual images along
this path. The geolocation of imagery and flight paths allows relatively precise repeat
transects to be conducted, and also for the imagery to be used to ground-truth multibeam
sonar (Lucieer et al. 2013) as well as for modelling the environmental factors driving species’
distributions (Hill et al. 2014).

Page | 67
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

• The time-gain it provides over an ROV. This particularly the case if the AUV system can be
left alone (i.e. that are truly autonomous).
• An AUV will follow the set path, will not slow down or divert for something pretty, exciting or
scary in the water: something that tends to happen to the humans when piloting an ROV.
• They generate spatially accurate photomosaics and finescale digital elevation models.
Multibeam data which is often available with accurate georeferencing can provide important
information regarding habitat types and structural complexity but is often limited to cell
resolutions of 50 cm to 5 m. Finescale digital elevation models from AUV photomosaics can
be done at 1-10cm cell resolution, thus enabling extremely detailed structural information to
be extracted (Ferrari et al. 2016a,b). Additionally, and perhaps more importantly, the benefits
of using AUV to provide digital elevation models is that the AUVs also provide colour
information (via the photomosaics), which is crucial for species identification and the
evaluation condition (e.g. live vs. dead coral).

The manner that data is extracted from imagery (i.e. image annotation) is context-dependent and
ranges from the simple scoring of presence-absence of indicator organisms or habitats within
individual images (e.g. Perkins et al. 2016) to automated habitat classification that uses
sophisticated algorithms (e.g. Friedman et al. 2011). Random point count is one of the commonly
employed approaches in the quantification of the cover of benthic habitats or organisms (e.g. James
et al. 2017; Monk et al. 2016; Perkins et al. 2016). Whilst pattern recognition annotation has the
potential to substantially speed up the image scoring process, it is not a point yet where it is
accurate enough to replace manual point-counts. Accordingly, this manual will focus on point-count
annotation approaches.

4.4 Pre-Survey Preparations

Ensure all permits, safety plans and approvals have been obtained. Any research undertaken within
Australian Marine Parks (AMPs) requires a research permit issued from Parks Australia. See
Appendix B for a list of potential permits needed.

Define question/aim of project.

Confirm sampling design is statistically sound with adequate spatial coverage and replication, and
addresses the initial question/aim. This is generally achieved through the use of an explicit
randomization procedure to ensure that independent replicates are obtained (Foster et al. 2017;
Smith et al. 2017). See Chapter 2 for further details on sampling design.

Select appropriate transect design for AUV deployment. Two AUV transect designs are
recommended for marine monitoring: 1) broad grids and 2) dense grids. Foster et al. (2014)
evaluated a number of broad grid designs and determined that a grid consisting of three long
parallel transects (each generally covering a total of 2000-4000 m) was generally the most optimal
design for monitoring purposes (Figure 4.2). The dense grid transects are used to get a complete
coverage photomosaic that covers a 25 x 25 m scale (Figure 4.2). Combinations of both within a
survey can be applied if required (e.g. Morris et al. 2016). Essentially, broad grids cover more
ground but are less repeatable, whereas dense grids are more repeatable but less general
(essentially you get more information about less).

The decision to which transect design is most appropriate is driven by the question being
addressed, as well as the environment, available time and logistics of AUV deployment and
retrieval. For example, in the deeper regions (> 100m) within the AMPs that are exposed to strong
currents, dense grids are not recommended for temporal monitoring purposes because the
challenges with maintaining physical position in these conditions make it difficult to successfully
Page | 68
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

repeat the same 25 x 25 m grid. This ultimately results in limited temporal overlap between
sampling points over time (Figure 4.3). Where inference is the primary objective of the study it is
recommended that broad grids are used to increase sampling power (Chapter 2). Conversely, if the
physical structure of the seafloor or biota (e.g. corals; Ferrari et al. 2016a) are the focus then dense
grids are best suited.

Broad grids are generally used in mid-outer continental shelf Tasmanian waters as result of strong
currents. Conversely in Western Australia, the patchy nature of inshore reefs, coupled with a lack of
shelf slope to encompass a wide depth range along broad grid designs meant that dense grids
surveys undertaken within each of a replicate number of patch reef systems and depths was the
most pragmatic solution. In southern Queensland, dense grids were the primary method used due
to the initial process-based research focus, however, the missions are time intensive, as is post
processing and analysis, and could readily be modified to a broad grid design in the future to
simplify analysis. In NSW a combination of both broad and dense grids has been conducted at most
sites over several time periods, although more recent surveys in the Sydney region have just used
broad grids.

Stereo-cameras must be pre- or post-calibrated in shallow water using the techniques similar to
those outlined in Boutros et al. (2015).

Decide on appropriate navigational systems (e.g. USBL). Accurately geo-referenced imagery is

crucial to the success of any AUV deployment, and appropriate effort must be given to this during
the survey planning phase.

Ensure appropriate software is installed on onboard laptops (e.g. AUV navigation software platform,
GIS, etc), and potential users are familiar with it so that the AUV can be tracked and its mission
success monitored while underway.

Figure 4.2 Examples of AUV transect designs over multibeam mapped reef features. Left: stand-alone 25 x 25 m dense
grid transect. Middle: stand-alone broad grid. Right: combination of broad grid with a dense grid imbedded. Note with this
design broad grid transects are usually shorter due to the time required to complete both grid types.

Page | 69
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Figure 4.3 Example of spatial mismatch between sample time points for a 25x25 m grid in a high current/wave action
environment. Note the limited overlap between all three sampling points.

4.5 Field Procedures

4.5.1 Onboard sample acquisition

Complete an on-site briefing. Prior to deployment, a deployment briefing should always be

completed to ensure the operation can be completed safely. Always take a precautionary approach
to risks associated with vehicle deployment. See Chapter 1 for further information about risk
assessments.

Set up and test AUV system. Allow sufficient time during survey mobilisation to undertake system
checks, calibrations and testing of equipment and account for unforeseen problems; in most cases it
will be possible to complete all system setup and tests within half a day. The conduct of pre-start
checks should be noted in the trip log and any test failures specifically recorded for later-reference.
Detailed settings for each component should be made using relevant operations manuals (e.g.
USBL operations manual etc.).

On-deck tests should include, but not limited to, the following checks:
• on-board data storage
• on-board power
• cameras
• strobe lighting
• iridium beacon, RF and emergency strobes
• propellers
• all blanking plugs are installed
• correct and new corrodible link attached emergence ascent drop weight
• crane and associated shackles are working order
• check all seals/o-rings and blanking plugs are good working order
• check all surface communications

Wet testing should include checks of the following:

• USBL and internal navigation (e.g. compass and avoidance sonar)
Page | 70
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

• cameras and strobes

• through-water communications

Acoustic tracking setup

• Set position of GPS receiver. Differential GPS is mandatory for repeat site monitoring.
• Deploy USBL transceiver (e.g. pole or vessel mounted).
• Measure offsets of USBL transceiver head to GPS receiver and put offsets into navigation
system.

Conduct AUV transects

Pre- deployment
• Transects should only be undertaken in areas where the substratum is known/mapped
(often in the form of multibeam mapping) as to avoid entrapment and potential loss of AUV.
Do not deploy blind, as this increases the risk of equipment loss and damage, as well as
unnecessary impact on potentially vulnerable ecosystems.
• Once final transect locations have been determined, provide the locations of the transects
(usually in ESRI shapefile format) and associated multibeam maps (in geotif format) to the
AUV engineers responsible for uploading missions. Cross-check the uploaded transect
corresponds to the correct area on the geotif (i.e. ensure the geographic coordinates are
defined for all spatial data).
• The flight elevation of AUV should be set and maintained at ~ 2m from the seafloor to
facilitate a consistent field of view. General sampling methodology can be found in Williams
et al. (2012). Although this needs to be informed by 'survey question', camera type and
performance, illumination type and output power, etc.
• Prepare for AUV launch and recovery on deck, and ensure only essential personnel
participate in its preparation and deployment.
• Place USBL transceiver in water and ensure functionality.
• Correctly insert the deployment release pin.

AUV deployment and retrieval

1. Disconnect any power or data cables, ensuring any blanking plugs are fitted prior to
deployment.
2. Install sacrificial ballast weights. Ensure that there is sufficient time allocated to transect
when selecting corrodible link.
3. Vessel master must ensure the vessel is positioned at the start of the transect start
location.
4. Following the signal to deploy from the vessel Master, use the crane and/or A-Frame to
lift and guide the AUV from the deck into the water.
5. Minimise the time taken from when the AUV is let out of reach, to when it is lowered in the
water, so as to reduce potential swing and impact against the vessel.
6. Using appropriate software (see Pre-Survey Preparations), monitor the AUVs progress to
the seabed and start of transect location. Note the start time of transect using a timer as
this will be used to determine when the sacrificial weight will be automatically released (if
fitted) in the case of an emergency.
7. Confirm data is being recorded where possible (e.g. recording indicators, hard drive
operating).
8. Ask the vessel's Master to follow the AUV during transects, to maintain USBL
communication and AUV tracking.

Page | 71
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

9. Monitor weather forecast conditions prior to and during deployment to maintain safe
working environment. Consider aborting operations if local weather and forecast
conditions are marginal.
10. When the transect is complete or if the transect is being aborted, advise the vessel
Master of the intention to retrieve the AUV.
11. Watch for the AUV to resurface, ensuring only required personnel are near open transom.
Avoid approaching the AUV looking into the sun as this increases the risks of collision.
12. Use grapple hook to connect the lift line to the AUV for retrieval. At least three personnel
should be present with hooks to avoid the AUV colliding with vessel [Recommended].
13. Shut down the AUV and connect relevant power or data cables.
14. Remove the sacrificial ballast weights.
15. For the last transect of the day, wash down the AUV with freshwater, unplug the USBL
and turn off emergency beacons.
16. Raise the USBL transducer (if pole mounted) before moving vessel to next location.

Procedures for seabed entanglement or loss of communications with AUV

Potential entanglement of the AUV is always a possibility. The following procedures should be
followed upon entanglement:

1. Log the last known position of the AUV.

2. Send an abort code to AUV to manually end the transect.

3. If the AUV appears entangled (i.e. not moving), a mini remotely operated vehicle (ROV)
should be used to locate and retrieved the unit. If the AUV is trapped under a ledge/cave, or
ensnared in fishing line or kelp, the automatic release of the sacrificial weights may cause
issues with recovery of the unit. Under such circumstances it is recommended that a ROV is
deployed to recover the AUV.
4. If the AUV is fitted with a sacrificial dump weight, which automatically releases after a user
defined period, it may surface on its own. Once it’s on the surface, use the fitted iridium
beacon, RF, GPS and emergency strobes to locate unit.
5. Ensure that you check AUV thoroughly for damage before redeployment.

Completion of operations

Prior to any vessel movement or engine start-up, operators should check the following:

• All equipment is clear of the water, including the USBL transducer pole.
• AUV is shut down.
• All gear is safely stowed.
• All power and data cables are connected.
• An “All Clear to Move” command is given to vessel Master when the AUV team is satisfied it
is OK for the vessel to move on.

Page | 72
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

4.5.2 Onboard data processing and storage

6. Once the AUV transect is complete, it is good practice to download associated raw imagery
and associated positional data. Imagery and associated positional data should be checked
to ensure no failures have occurred, including but not limited to the following:

• Miss-timing between image capture and strobes (i.e. dark/black imagery)

• Failure of one of the stereo cameras
• Failure of positional logging

2. Name data files according to established conventions. File naming conventions are
important for ensuring both efficient and effective management of field data and its
integration into appropriate data management repositories. It is important to note that these
conventions will differ among agencies and academic institutions.

3. Ensure accurate recording of metadata. Metadata is a descriptive data source comprised of

information that may be used to process the images or information therein Durden et al.
(2016). While it is important to follow agency specific protocols for capturing metadata, it is
also essential that metadata is sufficient enough in detail to satisfy conformance checks for
subsequent data release via AODN. Minimum data for each transect should contain as
follows:

• Campaign (i.e. Survey identifier)

• Station/event number
• Platform
• Latitude and longitude (WGS 1984 in decimal degrees with a minimum of 6 decimal
places [Recommend])
• Altitude
• Depth
• Time and date stamp
• AUV orientation (roll, pitch, heading)
• Precision details (e.g. type of navigation system used and its associated errors)
• Data provenance

4. Backup data. This is necessary to ensure all data collected in the field is safely returned and
securely backed-up at host facilities, prior to quality control and public release. Onboard
copies of data should be made as soon as practical following acquisition. When operating
external to a network, it is recommended that all data be backed up on a RAID or a NAS that
contain built-in storage redundancy in case of hard-drive failure. A duplicate copy of all data
onto external hard drives for transportation back to host facilities is [Recommended].

4.6 Post-Survey Procedures

4.6.1 Data processing

A general workflow for data processing methodology can be found in Williams et al. (2012). Key
requirements for raw image processing and positional data are as follows:
• It is recommended that at least one of the stereo images is in colour and enhanced following
similar procedures as outlined by Bryson et al. (2016).
• All stereo images should be georectified following Williams et al. (2012). If not stereo then
processing routines can be found in Morris et al. (2014).
Page | 73
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

• Positional data should be post-processed using Simultaneous Localisation and Mapping

(SLAM) as demonstrated in Barkby et al. (2009) and Palomer et al. (2013)

4.6.2 Data annotation

Scoring of individual images can be done using a number of annotation software tools. Examples
include, Transect measure, Coral Point Count, CoralNet and Squidle+. For national consistency
Squidle+ (http://squidle.greybits.com.au) is recommended as it is free and allows for different
approaches in image subsampling, which appears to influence inferences from data (Monk et al.
unpublished data), as well as stratified and random point count distribution on images. It also
automatically imports the collected AUV data once it is uploaded to the AODN making it ready for
analysis, and has tools for exploring survey data as well as analysis. In addition, it supports multiple
annotation schemes, and will provide consistency through translation between schemes, which is
an important point that differentiates Squidle+.

There are three approaches recommended for annotating georeferenced imagery from AUVs:
• Annotation of individual images
• Annotation of photomosaics
• Extracting structural complexity from orthomosaics

Annotation of individual images or photomosaics can be undertaken using three methods:

• Full assemblage scoring of imagery across space and time. It is important to note that this is
a time-consuming process, requiring a lot of replicate images to be scored to enable
sufficient power to detect biologically meaningful change as most morphospecies are < 10 %
cover within images. This approach appears to be good for delineating bioregional and
cross-shelf patterns at a morphospecies (Monk, et al. unpublished data) and CATAMI
(Althaus et al. 2015) level (James et al. 2017; Monk et al. 2016). This approach will no doubt
be effective in choosing initial suite of indicators for national level monitoring and reporting.

As a general guideline, and dependant on the survey question, we recommend that 25

random points per image from at least 50 images per transect leg are a good starting point
for recording most morphospecies present within images (based on Perkins et al. 2016). It is
important to note that the properties of the organism themselves will also influence the
number of points/images to score. Obviously morphospecies that are less abundant require
more effort, but also the 'clumpiness' of species will affect the scoring effort needed (Perkins
et al. 2016). Van Rein et al. (2011) and Perkins et al. (2016) suggest that, while a higher
number of points per image can increase the detection rate of more organisms within an
image, increasing the number of scored images using fewer points is likely have a similar (or
greater) effect. Ideally, increasing both the number of images scored and the number of
points scored within an image would result in greater power (Roelfsema et al. 2006), but
preference is usually for increasing the number of images (Perkins et al. 2016).
Unfortunately, the adoption of this approach is likely to result in substantial increases in
processing time and thus cost.

• Targeted scoring of indicators or proxies (such as grouping fine level morphospecies into
broader level CATAMI classes; Monk et al. unpublished data). This approach has been
shown to work very well at an indicator morphospecies level for detecting change at a
regional level (Perkins et al. 2017) as well as for detecting invasive species trends (Ling et
al. 2016; Perkins et al. 2015). More recently this approach has been extended to mobile
species, such as fish (Seiler et al. 2012) and lobster (Bessell et al. unpublished data). Care
needs to be taken if length data (using photogrammetry or structure from motion) is
extracted from stereo pairs from Sirius data as both Seiler et al. (2012) and Bessell et al.
Page | 74
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

(unpublished data) found precision can be poor for mobile species if camera separation is
inadequate (see Boutros et al. 2015)

Since this approach requires substantially less effort to score each image, more images (i.e.
often all images) can be scored and, thus, increased statistical power. The drawback is that
narrower understanding of the environment is produced.

• Automated analysis of imagery potentially provides a cost-effective alternative to annotating

imagery from AUVs. It is important to note that automated imagery analysis is a relatively
new, and largely developmental, way of annotating images. Despite this some studies
suggest that coral and macroalgae can be reliably identified using automated image analysis
(Table 7).

The last approach to annotating AUV imagery involves the extraction of 3D structural information
from stereo images using structure from motion techniques outlined in Ferrari et al. (2016) and
Pizarro et al. (2017). This approach works particularly well too for sessile species to track changes
in growth form through time at a 25 x 25 m scale (Ferrari et al. 2016).

Table 4.1 A brief summary of methods for automated benthic image classification. The number of classes and the main
taxa included in the respective studies are also shown.

Authors Classes Main Species

Marcos et al. (2005) 3 Corals
Stokes & Deane (2009) 18 Corals, Macroalgae
Pizarro et al. (2008) 8 Corals, Macroalgae
Beijbom et al. (2012) 9 Corals, Macroalgae
Denuelle & Dunbabin (2010) 2 Kelp
Bewley et al. (2012) 19 Corals, Algae and Kelp
Bewley et al. (2014) 19 Corals, Algae and Kelp
Beijbom et al. (2016) 10 Corals, Macroalgae
Mahmood et al.(2016a) 9 Corals, Macroalgae
Mahmood et al. (2016b) 2 Corals, Macroalgae

4.6.3 Data curation and quality control

A national AUV steering group has been set up to oversee a nationally coordinated AUV benthic
monitoring program which is supported by the Integrated Marine Observing System (IMOS) (Table
4.2). Any new AUV deployments should be discussed with this steering group to ensure that,
wherever possible, they can be integrated within the national program [Recommended].

Table 4.2 Key contacts in national AUV steering group as of Jan 2018

Name State Organisation

Neville Barrett* Tasmania IMAS
Craig Johnson Tasmania IMAS
Peter Steinberg New South Wales SIMS
Alan Jordan New South Wales NSW DPI
Page | 75
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Stefan Williams New South Wales USyd

Gary Kendrick Western Australia UWA
Russ Babcock Western Australia CSIRO
Paul Van Ruth South Australia SARDI
Hugh Sweatman Queensland AIMS
Tom Bridge Queensland JCU/QLD Museum
Daniel Ierodiaconou Victoria Deakin
* Chair

Data quality control at both the collection and annotation stage is critical. Most importantly, the
annotation schema needs to be consistent between studies. Morphospecies and associated
CATAMI parent classes be used [Recommended]. An initial morphospecies catalogue for
southeastern shelf waters is currently held and maintained at the Institute for Marine and Antarctic
Studies (IMAS) (contact Dr Neville Barrett or Dr Jacquomo Monk).

Other annotation schema are available, and can be applied. In such situations where an alternative
schema are used to annotate AUV imagery, it must be able to be mapped to CATAMI so that
comparisons can be made with previous studies or between regions. Translations between schema
can be readily applied within Squidle+. The quality control of all annotations undertaken by novice
scores should be assessed against an experienced analyst (e.g. using confusion matrices; Figure
4.4). Logically, it is important to correct any discrepancies between annotators. This can be done by
re-examining the images to ensure an agreement can be reached between annotators.
Alternatively, if an agreement cannot be reached, then the miss-classified morphospecies could be
potentially grouped into a higher level CATAMI class.

Page | 76
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Figure 4.4 Confusion matrix showing the CATAMI classes scored by novice 1 (AW) and experienced (JH) for 30 co-scored
images. Black outlined boxes indicate consistent classification between scorers, the percent of all points scored as any
particular class are is shown in each box and colour coded. Blue outlined boxes indicate sponge, bryozoan/hydroid and
substratum respectively moving from left to right across the image.

4.6.4 Data release

Squidle+ is a centralised online platform for standardised analysis and annotation of georeferenced
imagery and video. Many national marine observing programs (for example IMOS through the
Australian Ocean Data Network (AODN) or the Marine Geoscience Data System (MGDS) in the
USA) routinely store imagery data online in an openly accessible location. Squidle+ operates based
on flexible distributed data storage facilities (i.e. imagery can be stored anywhere in an openly
accessible online location) to reduce data duplication and inconsistencies, and provides a flexible
annotation system with the capability to translate between different annotation schemes.

Page | 77
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Following the steps listed below will ensure the timely release of imagery and associated annotation
data in a standardised, highly discoverable format.

1. Create a metadata record describing the data collection. Provide as much detail as possible on
the deployment (either directly in the metadata record itself, or in the form of attached field
sheets as .csv, .txt or similar). Details of minimum metadata requirements are provided in
Onboard Data Processing and Storage section above.

2. Publish metadata record(s) to the Australian Ocean Data Network (AODN) catalogue as soon as
possible after metadata has been QC-d. This can be done in one of two ways:

• If metadata from your agency is regularly harvested by the AODN, follow agency-specific
protocols for metadata and data release.
• Otherwise, metadata records can be created and submitted via the AODN Data Submission
Tool. Note that user registration is required, but this is free and immediate.

Lodging metadata with AODN in advance of annotation data being available is an important step
in documenting the methods and location of acquired imagery and enhancing future
discoverability of the data.
3. Upload raw imagery from the survey to a secure, publicly accessible online repository (contact
AODN if you require assistance in locating a suitable repository).

4. Create a Squidle + campaign as soon as possible after imagery is uploaded, choose the most
appropriate annotation schema, and commence annotation of imagery.

5. Add links to the location of the Squidle+ campaign to the previously published metadata record.
You may also wish to attach or link a copy of the annotation data directly to the record.

6. Produce a technical or post-survey report documenting the purpose of the survey, sampling
design, sampling locations, sampling equipment specifications, annotation schema (e.g.
morphospecies, CATAMI, etc.), whether the survey was assemblage-based or targeted towards
key (morpho)species, number of points, interval between images (e.g. every 50th image), and
any challenges or limitations encountered. Provide links to this report in all associated metadata.
See Appendix C for a suitable template [Recommended].

4.6.5 Data analysis

The breadth of research questions precludes any detailed advice on the analysis of data from AUV
transects. However, one common attribute of the image-based data that will have to be contented
with for all analyses is spatial proximity. The closeness of images, within and sometimes between
transects, means that image data are unlikely to be independent (due to spatial autocorrelation).
Yet, this is an assumption that many statistical methods rely upon. The failure to meet this
assumption means that the inferences from the statistical analysis may be: (i) over-confident, e.g.
having a p-value that is too small; (ii) biased, i.e. the estimates do not reflect the truth; (iii) both, or;
(iv) no effect. Obviously, the fourth category is what a researcher hopes for, but it is improbable and
must be validated. However, if it is known that the study organism exhibit particularly low
autocorrelation then the analysis need not consider it explicitly.

Methods to analyse data, accounting for autocorrelation are available. These include geostatistical
models (see Foster et al. 2014 for AUV-based examples). However, in certain situations
subsampling images will help (see Mitchell et al. 2017 for a marine based example), but not
necessarily alleviate completely. Further, if the study is for a broad area, where transects are small
and are well-separated, then amalgamating data to transect level may also be appropriate.
Page | 78
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

4.6.6 Field Manual Maintenance

In accordance with the universal field manual maintenance protocol described in Chapter 1 of the
Field Manual package, this manual will be updated in 2018 as Version 2. Updates will reflect user
feedback and new developments (e.g. data discoverability and accessibility). Version 2 will also
detail subsequent version control and maintenance.

The version control for Chapter 4 (field manual for AUVs) is below:

Version Description Date

Number
0 Submitted for review (NESP Marine Hub, GA, 22 Dec 2017
external reviewers as listed Appendix A.
1 Publicly released on www.nespmarine.edu 28 Feb 2018
1.1 Link to Squidle+ corrected March 2018
2 Relevant updates, including Data Release Early 2019
sections based on NESP, AODN, IMOS, GA,
and CSIRO projects

4.7 References
Althaus, F., Hill, N., Ferrari, R., Edwards, L., Przeslawski, R., Schönberg, C.H., Stuart-Smith, R., Barrett, N., Edgar, G.,
and Colquhoun, J. 2015. A standardised vocabulary for identifying benthic biota and substrata from underwater
imagery: the CATAMI classification scheme. PLoS ONE 10:e0141039.
Barkby, S., Williams, S.B., Pizarro, O., Jakuba, M., 2009. An Efficient Approach to Bathymetric SLAM. IEEE/RSJ
International Conference on Intelligent Robots and Systems, 219-224.
Beijbom, O., Edmunds, P.J., Kline, D., Mitchell, B.G., Kriegman, D. 2012. Automated Annotation of Coral Reef Survey
Images. IEEE Conference on Computer Vision and Pattern Recognition, 1170–77.
Beijbom, O., Treibitz, T., Kline, D.I., Eyal, G., Khen, A., Neal, B., Loya, Y., Mitchell, B.G., Kriegman D. 2016. Improving
Automated Annotation of Benthic Survey Images Using Wide-Band Fluorescence. Scientific Reports 6: 23166.
Bewley, M.S., Douillard, B., Nourani-Vatani, N., Friedman, A., Pizarro, O., Williams, S.B. 2012. Automated Species
Detection: An Experimental Approach to Kelp Detection from Sea-floor AUV Images.
http://www.araa.asn.au/acra/acra2012/papers/pap140.pdf.
Bewley, M.S. Nourani-Vatani, N., Rao, D., Douillard, B., Pizarro, O., Williams, S.B. 2015. Hierarchical Classification in
AUV Imagery. L. Mejias, P. Corke, and J. Roberts (eds.), Field and Service Robotics, Springer Tracts in
Advanced Robotics 105
Bewley, M., Friedman, A., Ferrari, R., Hill, N., Hovey, R., Barrett, N., Marzinelli, E.M., Pizarro, O., Figueira, W., Meyer, L.,
Babcock, R., Bellchambers, L., Byrne, M., Williams, S.B., 2015. Australian sea-floor survey data, with images
and expert annotations. Scientific Data 2, 150057.
Boutros, N., Shortis, M.R., Harvey, E.S., 2015. A comparison of calibration methods and system configurations of
underwater stereo-video systems for applications in marine ecology. Limnology and Oceanography Methods 13,
224-236.
Bridge, T.C.L., Done, T.J., Friedman, A., Beaman, R.J., Williams, S.B., Pizarro, O., Webster, J.M., 2011. Variability in
mesophotic coral reef communities along the Great Barrier Reef, Australia. Marine Ecology Progress Series 428,
63-75.
Bryson, M., Johnson-Roberson, M., Pizarro, O., Williams, S.B., 2016. True Color Correction of Autonomous Underwater
Vehicle Imagery. Journal of Field Robotics 33, 853-874.
Clarke, M.E., Tolimieri, N., Singh, H., 2009. Using the Seabed AUV to Assess Populations of Groundfish in Untrawlable
Areas, in: Beamish, R.J., Rothschild, B.J. (Eds.), The Future of Fisheries Science in North America. Springer
Netherlands, Dordrecht, pp. 357-372.
Connelly, D.P., Copley, J.T., Murton, B.J., Stansfield, K., Tyler, P.A., German, C.R., Van Dover, C.L., Amon, D., Furlong,
M., Grindlay, N., Hayman, N., Huhnerbach, V., Judge, M., Le Bas, T., McPhail, S., Meier, A., Nakamura, K., Nye,
V., Pebody, M., Pedersen, R.B., Plouviez, S., Sands, C., Searle, R.C., Stevenson, P., Taws, S., Wilcox, S., 2012.
Hydrothermal vent fields and chemosynthetic biota on the world's deepest seafloor spreading centre. Nature
Communications 3.
Denuelle, A., Dunbabin, M. 2010. Kelp detection in highly dynamic environments using texture recognition. The
Australasian Conference on Robotics & Automation.

Page | 79
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Durden, J.M., Schoening, T., Althaus, F., Friedman, A., Garcia, R., Glover, A.G., Greinert, J., Jacobsen Stout, N., Jones,
D.O.B., Jordt, A., Kaeli, J.W., Köser, K., Kuhnz, L.A., Lindsay, D., Morris, K.J., Nattkemper, T.W., Osterloff, J.,
Ruhl, H.A., Singh, H., Tran, M., Bett, B.J., 2016b. Perspectives in visual imaging for marine biology and ecology:
from acquisition to understanding, in: Hughes, R.N., Hughes, D.J., Smith, I.P., Dale, A.C. (Eds.), Oceanography
and Marine Biology: An Annual Review. CRC Press, Boca Raton, FL, pp. 1-72.
Ferrari, R., Bryson, M., Bridge, T., Hustache, J., Williams, S.B., Byrne, M., Figueira, W., 2016a. Quantifying the response
of structural complexity and community composition to environmental change in marine communities. Global
Change Biology 22, 1965-1975.
Ferrari, R., McKinnon, D., He, H., Smith, R.N., Corke, P., González-Rivero, M., Mumby, P.J., Upcroft, B., 2016b.
Quantifying Multiscale Habitat Structural Complexity: A Cost-Effective Framework for Underwater 3D Modelling.
Remote Sensing 8, 113.
Foster, S.D., Hosack, G.R., Hill, N.A., Barrett, N.S., Lucieer, V.L., 2014. Choosing between strategies for designing
surveys: autonomous underwater vehicles. Methods Ecology and Evolution 5, 287-297.
Foster, S.D., Hosack, G.R., Lawrence, E., Przeslawski, R., Hedge, P., Caley, M.J., Barrett, N.S., Williams, A., Li, J.,
Lynch, T., Dambacher, J.M., Sweatman, H.P.A., Hayes, K.R., 2017. Spatially balanced designs that incorporate
legacy sites. Methods in Ecology and Evolution, 8, 1433-1442.
Friedman, A., Steinberg, D., Pizarro, O., Williams, S.B., 2011. Active Learning Using a Variational Dirichlet Process Model
for Pre-Clustering and Classification of Underwater Stereo Imagery. IEEE/RSJ International Conference on
Intelligent Robots and Systems, 1533–1539.
Furlong, M.E., Paxton, D., Stevenson, P., Pebody, M., McPhail, S.D., Perrett, J., 2012. Autosub Long Range: A Long
Range Deep Diving AUV for Ocean Monitoring. IEEE/OES Autonomous Underwater Vehicles (AUV), 1-7.
Hill, N.A., Lucieer, V., Barrett, N.S., Anderson, T.J., Williams, S.B., 2014. Filling the gaps: Predicting the distribution of
temperate reef biota using high resolution biological and acoustic data. Estuarine and Coastal Shelf Science 147,
137-147.
Hobson, B.W., Bellingham, J.G., Kieft, B., McEwen, R., Godin, M., Zhang, Y., 2012. Tethys-class long range AUVs -
extending the endurance of propeller-driven cruising AUVs from days to weeks. IEEE/OES Autonomous
Underwater Vehicles (AUV) 1-8.
Huvenne, V.A.I., McPhail, S.D., Wynn, R.B., Furlong, M., Stevenson, P., 2009. Mapping Giant Scours in the Deep Ocean.
Eos, Transactions American Geophysical Union 90, 274-275.
Huvenne, V.A.I., Robert, K., Marsh, L., Lo Iacono, C., Le Bas, T., Wynn, R.B., 2018. “ROVs and AUVs.” In Submarine
Geomorphology, edited by A. Micallef, S. Krastel, and A. Savini, 572. Springer Geology. Cham, Switzerland:
Springer.
James, L.C., Marzloff, M.P., Barrett, N., Friedman, A., Johnson, C.R., 2017. Changes in deep reef benthic community
composition across a latitudinal and environmental gradient in temperate Eastern Australia. Marine Ecology
Progress Series 565, 35-52.
Ling, S.D., Mahon, I., Marzloff, M.P., Pizarro, O., Johnson, C.R., Williams, S.B., 2016. Stereo-imaging AUV detects trends
in sea urchin abundance on deep overgrazed reefs. Limnology and Oceanography Methods 14, 293-304.
Lucieer, V., Hill, N.A., Barrett, N.S., Nichol, S., 2013. Do marine substrates ‘look’ and ‘sound’ the same? Supervised
classification of multibeam acoustic data using autonomous underwater vehicle images. Estuarine and Coastal
Shelf Science 117, 94-106.
Mahmood, A., Bennamoun, M., An, S., Sohel, F., Boussaid, F., Hovey, R., Kendrick, G., Fisher, R 2016. Automatic
annotation of coral reefs using deep learning, IEEE OCEANS Monterey, 1–5.
Mahmood, A., Bennamoun, M., An, S., Sohel, F. 2016. Resfeats: Residual network based features for image
classification, arXiv preprint arXiv:1611.06656.
Marcos, M.S.A., Soriano, M., Saloma, C. 2005. Classification of Coral Reef Images from Underwater Video Using Neural
Networks. Optics Express 13: 8766–71.
Marzinelli, E.M., Williams, S.B., Babcock, R.C., Barrett, N.S., Johnson, C.R., Jordan, A., Kendrick, G.A., Pizarro, O.R.,
Smale, D.A., Steinberg, P.D., 2015. Large-scale geographic variation in distribution and abundance of Australian
deep-water kelp forests. PLoS One 10, e0118390.
Milligan, R., K. Morris, B. Bett, J. Durden, D. Jones, K. Robert, H. Ruhl & D. Bailey, 2016. High resolution study of the
spatial distributions of abyssal fishes by autonomous underwater vehicle. Scientific Reports 6:26095
doi:10.1038/srep26095.
Mitchell, PJ., Monk, J., Laurenson, L. 2017. Sensitivity of Fine-Scale Species Distribution Models to Locational Uncertainty
in Occurrence Data across Multiple Sample Sizes. Methods in Ecology and Evolution. 8:12-21.
Monk, J., Barrett, N.S., Hill, N.A., Lucieer, V.L., Nichol, S.L., Siwabessy, P.J.W., Williams, S.B., 2016. Outcropping reef
ledges drive patterns of epibenthic assemblage diversity on cross-shelf habitats. Biodiversity and Conservation
25, 485-502.
Morris, K., B. Bett, J. Durden, V. Huvenne, R. Milligan, D. Jones, S. McPhail, K. Robert, D. Bailey & H. Ruhl, 2014. A new
method for ecological surveying of the abyss using autonomous underwater vehicle photography. Limnology and
Oceanography: Methods 12:795-809.
Morris, K., B. Bett, J. Durden, N. Benoist, V. Huvenne, D. Jones, K. Robert, M. Ichino, G. Wolff & H. Ruhl, 2016.
Landscape-scale spatial heterogeneity in phytodetrital cover and megafauna biomass in the abyss links to
modest topographic variation. Scientific Reports 6:34080 doi:doi:10.1038/srep34080.
Palomer, A., Ridao, P., Ribas, D., Mallios, A., Vallicrosa, G., 2013. A Comparison of G2o Graph SLAM and EKF Pose
Based SLAM with Bathymetry Grids. IFAC Proceedings Volumes 46, 286-291.
Page | 80
Marine Sampling Field Manuals for Monitoring Australia’s Commonwealth Waters Version 1

Pennington, J.T., Blum, M., Chavez, F.P., 2016. Seawater sampling by an autonomous underwater vehicle: “Gulper”
sample validation for nitrate, chlorophyll, phytoplankton, and primary production. Limnology and Oceanography:
Methods 14, 14-23.
Perkins, N.R., Foster, S.D., Hill, N.A., Barrett, N.S., 2016. Image subsampling and point scoring approaches for large-
scale marine benthic monitoring programs. Estuarine and Coastal Shelf Science 76, 36-46.
Perkins, N.R., Foster, S.D., Hill, N.A., Marzloff, M.P., Barrett, N.S., 2017. Temporal and spatial variability in the cover of
deep reef species: Implications for monitoring. Ecological Indicators 77, 337-347.
Perkins, N.R., Hill, N.A., Foster, S.D., Barrett, N.S., 2015. Altered niche of an ecologically significant urchin species,
Centrostephanus rodgersii, in its extended range revealed using an Autonomous Underwater Vehicle. Estuarine
and Coastal Shelf Science 155, 56-65.
Pizarro, O., Rigby, P., Johnson-Roberson, M., Williams, S.B., Colquhoun, J. 2008. Towards image-based marine habitat
classification. IEEE explore OCEANS 1–7.
Pizarro, O., Friedman, A., Bryson, M., Williams, S.B., Madin, J., 2017. A simple, fast, and repeatable survey method for
underwater visual 3D benthic mapping and monitoring. Ecology and Evolution 7, 1770-1782.
Rigaud, V., 2007. Innovation and operation with robotized underwater systems. Journal of Field Robotics 24, 449-459.
Roelfsema, C., Phinn, S., Joyce, K., 2006. Evaluating benthic survey techniques for validating maps of coral reefs derived
from remotely sensed images 10th International Coral Reef Symposium, pp. 177-1780.
Seiler, J., Williams, A., Barrett, N., 2012. Assessing size, abundance and habitat preferences of the Ocean Perch
Helicolenus percoides using a AUV-borne stereo camera system. Fisheries Research 129-130, 64-72.
Smale, D.A., Kendrick, G.A., Harvey, E.S., Langlois, T.J., Hovey, R.K., Van Niel, K.P., Waddington, K.I., Bellchambers,
L.M., Pember, M.B., Babcock, R.C., Vanderklift, M.A., Thomson, D.P., Jakuba, M.V., Pizarro, O., Williams, S.B.,
2012. Regional-scale benthic monitoring for ecosystem-based fisheries management (EBFM) using an
autonomous underwater vehicle (AUV). ICES Journal of Marine Science 69, 1108-1118.
Smith, A.N.H., Anderson, M.J., Pawley, M.D.M., 2017. Could ecologists be more random? Straightforward alternatives to
haphazard spatial sampling. Ecography 40, 1251–1255.
Stokes, M.D., Deane G.B. 2009. Automated Processing of Coral Reef Benthic Images. Limnology and Oceanography,
Methods / ASLO 7: 157–68.
Sumner, E.J., Peakall, J., Parsons, D.R., Wynn, R.B., Darby, S.E., Dorrell, R.M., McPhail, S.D., Perrett, J., Webb, A.,
White, D., 2013. First direct measurements of hydraulic jumps in an active submarine density current.
Geophysical Research Letters 40, 5904-5908.
Tivey, M.A., Johnson, H.P., Bradley, A., Yoerger, D., 1998. Thickness of a submarine lava flow determined from near-
bottom magnetic field mapping by autonomous underwater vehicle. Geophysical Research Letters 25, 805-808.
Van Rein, H., Schoeman, D.S., Brown, C.J., Quinn, R., Breen, J., 2011. Development of benthic monitoring methods
using photoquadrats and scuba on heterogeneous hard-substrata: a boulder-slope community case study.
Aquatic Conservation 21, 676-689.
Wagner, J.K.S., McEntee, M.H., Brothers, L.L., German, C.R., Kaiser, C.L., Yoerger, D.R., Van Dover, C.L., 2013. Cold-
seep habitat mapping: High-resolution spatial characterization of the Blake Ridge Diapir seep field. Deep Sea
Research Part 2 Topical Studies in Oceanography 92, 183-188.
Williams, S., Pizarro, O., Jakuba, M., Johnson, C., Barrett, N., Babcock, R., Kendrick, G., Steinberg, P., Heyward, A.,
Doherty, P., Mahon, I., Johnson-Roberson, M., Steinberg, D., Friedman, A., 2012. Monitoring of Benthic
Reference Sites: Using an Autonomous Underwater Vehicle. IEEE Robotic Automation Magazine 19, 73-84.
Williams, S.B., Pizarro, O., Steinberg, D.M., Friedman, A., Bryson, M., 2016. Reflections on a decade of autonomous
underwater vehicles operations for marine survey at the Australian Centre for Field Robotics. Annual Reviews in
Control 42, 158-165.
Williams, S.B., Pizarro, O., Webster, J.M., Beaman, R.J., Mahon, I., Johnson-Roberson, M., Bridge, T.C.L., 2010.
Autonomous underwater vehicle–assisted surveying of drowned reefs on the shelf edge of the Great Barrier
Reef, Australia. Journal of Field Robotics 27, 675-697.
Wynn, R.B., Huvenne, V.A.I., Le Bas, T.P., Murton, B.J., Connelly, D.P., Bett, B.J., Ruhl, H.A., Morris, K.J., Peakall, J.,
Parsons, D.R., Sumner, E.J., Darby, S.E., Dorrell, R.M., Hunt, J.E., 2014. Autonomous Underwater Vehicles
(AUVs): Their past, present and future contributions to the advancement of marine geoscience. Marine Geology
352, 451-468.

Page | 81