Workshop on 3D mapping of habitats and biological communities with underwater photogrammetry

For the past decades, photogrammetry has been increasingly used for monitoring spatial arrangement or temporal dynamics of submerged man-made structures and natural systems. As photogrammetry remains a nascent technique for data collection in the underwater environment, acquisition workflows have evolved constrained by specific methodological practicalities (e.g. euphotic environments vs. deep-sea waters). The annual GeoHab conference gathers a world-wide range of scientists interested in mapping and is, therefore, an adequate event to set up a state-of-the-art workshop on (underwater) photogrammetry. More specifically, a preliminary survey identified the overall lack of


Introduction
For the past decade, with the development of acquisition platforms and computer power, publications involving photogrammetry by Structure from Motion (SfM) have increased by more than tenfold (Bayley andMogg 2020, Pulido Mantas et al. 2023), with ~ 25% of investigations performed in the marine environment.From coastal to abyssal areas, this technique allows mapping of objects at a high level of positional accuracy (< cm) in a 3D space.This type of data is conducive for precise spatio-temporal monitoring studies (D'Urban Jackson et al. 2020, Lange and Perry 2020, Fukunaga et al. 2022).Therefore, the value of this technique has evolved considerably as its development continues and research communities continue to innovate methods to analyse these types of data, tackling a broad range of topics, from ecology to geology.The annual international GeoHab symposium covers topics ranging from marine geosciences, mapping, ecology and targeting presentation of novel data-acquisition approaches.As a result, this event gathers a broad range of stakeholders and scientists from the marine community that may empower the use of photogrammetry to forward this field of research.
The "3D underwater mapping for habitats and biological communities" -GeoHab 2023 workshop took place on Monday 8 May 2023 at the Hotel Le Récif in Saint-Gilles-les-Bains (La Réunion).Throughout diverse activities, the workshop aimed: 1.
to provide GeoHab participants with knowledge on photogrammetry and to give them an overview of the potentials uses of this technique; 2.
to show-case studies on marine environments from shallow waters (with aerial and underwater surveys) to the deep seas (with underwater platforms); 3.
to train attendees to design and perform photogrammetry investigations that could be optionally monitored by a panel expert who ran that workshop and finally; 4.
to empower the audience and, more generally the marine community, in the domain of photogrammetry and 5.
to lead a hands-on session to practise photogrammetry sampling, processing with different software (e.g.Open Drone Map, ColMap, Meshroom, Agisoft Mestashape, CloudCompare) and methods for analyses (e.g.annotation on 3D mesh with 3D metrics, 3D dense cloud, digital elevation models and orthomosaics).
The goal of this report is to present the conception and running of the GeoHab 2023 workshop in order to help reproduce such a workshop.The present document includes details on the participants, the type of the audience, the programme and detailed information of the content of several activities and sessions.All associated data and supplementary materials were published in an open-access Zenodo repository: 10.5281/ zenodo.7934452.

Organising committee
The organisers comprised: Loïc Van Audenhaege (National Oceanography Centre, UK), Vincent Mahamadaly (CREOCEAN, FR), David Price (University of the Azores, PT), Alexandre Sneessens (CREOCEAN OI, FR), Isabel Urbina-Barreto (French National Institute for Sustainable Development IRD, La Réunion, FR; Fig. 1).The organisers work in the private and public academic sectors.Their scientific background was diverse in terms of environments of interest (i.e. from coastal to deep-sea marine ecosystems) and platforms for usual data acquisition (i.e.drone, scuba and free diving and remotely operated vehicles).We believed that the workshop benefitted from the collaboration of diverse experts because it enabled a holistic approach in the photogrammetry domain highlighting the differences and similarities across scientific disciplines, thus displaying an exhaustive state-of-the-art of photogrammetry.
The preparation of the workshop started in early February 2023 with one-hour meetings every two weeks or so.

List of participants
In total, 61 people participated to the workshop (Table 1; total number of registration = 64).List of participants to the GeoHab 2023 workshop.All participants shared consent to publish the following information.
Workshop on 3D mapping of habitats and biological communities with underwater ...Although all five continents were represented, participant origins were relatively uneven, with Europe (48.4%),North America (20.3%),Africa (15.6%),Oceania (12.5%),South America (1.6%) and Asia (1.6%).For future similar events, it could be helpful to consider a hybrid format and recording activities that could encourage attendance from zones with a large time difference (e.g.Oceania, America and Asia).

Preliminary survey
A preliminary survey aimed to collect information about participants to better understand their experience level and diversity.That helped adapt and define the content of the workshop.The reader will be assisted in determining how similar that audience was to their audience in order to determine whether our workshop is applicable to their particular situation.

Methods
A Google form was set up to collect answers from each participant to the seven following questions: 1. What is your current position?*2.
What field of research do you consider yourself involved in?* 3.
What ecosystem(s) does your work relate to?* 4.
How long have you used photogrammetric models for?* 5.
What aspects of photogrammetry do you particularly intend to focus on during the workshop?*6.
What platform do you usually use to perform photogrammetry?*N.B.All questions required an answer from the survey participant.

Results
All participants consented to share their answers anonymously.The total answer rate was 84.4% (54 participants out of the 64 registered).

What field of research do you consider yourself involved in?
Geologists, biologists/ecologists and mappers predominated (90%; Fig. 3) in roughly equal proportions, reflecting the general GeoHab conference's focus.The rest was mostly made of subgroups of one to two people (e.g.marine energy, computer sciences).

What ecosystem(s) does your work relate to?
Most of the attendees' work focuses on the marine environment (from the littoral to the deep sea; Fig. 4).A large share of attendees were identified as having experience in the sublittoral ecosystem (0-200 m depth).Still, the deep-sea realm was well represented with 44.4% working in bathyal and with 18.5% working in abyssal waters.Seventeen responders (31.5%) work on both sublittoral and bathyal environments.

How long have you used photogrammetric models for?
Half of the participants had no experience at all with 3D photogrammetry (Fig. 5).Still, 35.2% had less than two years of experience with photogrammetry.This overall 'lack of 5. What aspects of photogrammetry do you particularly intend to focus on during the workshop?
Answers from participants showed a relatively balanced interest amongst data collection, model computation and result extraction (Fig. 6).

Any additional comment?
From a selection of comments: • "Suggestions on best software (affordability vs. quality) would be very helpful, also if we came away with generic workflows".• "How to make the most of available data that was gathered without using any controlling methods (e.g.laser pointers, stabilisation)"?• "How can we get an orthophoto from georeferenced images" (x2)?• "I am interested in QPS SfM or QGIS Plugins".• "I am interested to know more about using AI or programming for post-processing and analyses" (x2).• "would be interested in 'filling the gap' nearshore between where multibeam stops and terrestrial mapping starts".• "Calibration/bundle adjustment and georectification".
Comments highlighted that the workshop duration was not sufficient to approach deeply enough all aspects of photogrammetry.However, those comments provided valuable information for modulating the hands-on session and the panel discussion (e.g.first comment on standardisation of photogrammetry and selection of software).

What platforms do you usually use to perform photogrammetry?
A balance predominated amongst the use of drones (33.3%), scuba diving (29.6%) and the use of underwater platforms, such as remotely operated vehicles (20.4 to 24.1%), towed cameras (40.7%), and, surprisingly, the more recent technologies of autonomous underwater vehicles (33.3%;Fig. 7).Workshop on 3D mapping of habitats and biological communities with underwater ...

Workshop activities
The programme was intended to cover multiple aspects of photogrammetry by SfM considering a meaningful progression amongst the different workshop sessions.
The first session of the morning focused on the theorical principles, presenting the mathematical rationale behind photogrammetry, the algorithms involved and the types of files that can be extracted to be displayed in a virtual environment (Fig. 8).Second, various presentations focused on sampling/survey designs, including what optical camera parameters should be considered or, depending on the sampled environment and acquisition platform, what practical constraints and limiting factors need to be considered.Legislation for drone acquisition was also discussed and underlined the specific regulation for each country.Third, the participants attended a hands-on session aiming to teach them image acquisition to live model reconstruction while leaving room for more specific questions at the same time.The afternoon consisted of case studies presentations which were focused on describing various applications from various environments as proofs of concept for the variety of research contexts in which photogrammetry can be used (Fig. 9).For this session, additional speakers from academia were invited as guest speakers (Simon Delsol, Iason-Zois Gazis and Valentin Danet).The day ended with a feedback panel discussion.

Presentations
Presentations during the workshop allowed participants to discover a wide range of acquisition platforms (e.g.aerial to underwater; Fig. 10).Objects imaged varied from manmade structures, geomorphological facies and different biological scales of interest (e.g. from sessile individuals to communities).Applications ranged from the use of photogrammetry to investigate spatial organisation and to monitor temporal dynamics of man-made structures or biological communities and to raise awareness of local population regarding their marine ecosystems.All presentations emphasised the importance of photogrammetry that provides centimetric-resolution of investigations needed for various ecological applications, such as growth monitoring in natural conditions or after disturbance, characterisation of topographically complex habitat and species co-existence to investigate niche partitioning.

Hands-on
Ranked preference between three hands-on was requested to each participant during the pre-event online survey.Based on their preferences, we assigned them to a hands-on stand that was run by one person or a group (Fig. 8): • The activities of the hands-on stand were defined by the stand leader(s).However, we encouraged the stand leaders to coordinate themselves to offer a similar content structure despite featuring different environment and acquisition platforms.Content similarities included display of image acquisition, mesh acquisition and manipulation of the mesh (e.g.Figs 11,12).Some stands proposed extra activities, such as mini ROV piloting in the hotel swimming pool and visualisation of a shipwreck using virtual headsets.
As an example, for reproducibility purposes, the hands-on activities of Loïc Van Audenhaege and David Price is presented below: 1.
Prior to the hands-on, a scene was created.Sand and fragments of different shapes were collected.After the workshop, fragments were put back in their original locations to comply with the environmental regulations; Workshop on 3D mapping of habitats and biological communities with underwater ...

2.
We briefly displayed other tools of Cloudcompare, such as the ICP and point-to-point registration algorithm to overlay models together.Although time did not allow it, we intended to load the 3D model on Meshroom and Codemap to display alternative and opensource photogrammetry software.Originally, we also intended to load data in Meshlab.This software includes algorithms that Cloudcompare does not hold, such as for remeshing algorithms to lower mesh resolution and fasten the display of results.

Panel discussion
Due to time constraint, the panel discussion only lasted 40 minutes: • What (developing) technologies could complement underwater photogrammetry to acquire high-resolution mapping?Sonar, Lidar and the hyperspectral camera in development were mentioned; • Practical aspects of scale integration of photogrammetry were discussed, notably with sonar dataset acquired at much lower resolution; • The differences between using SfM and colour band ratios to calculate bathymetry.

Feedback survey
A post-event feedback survey was conducted to allow participants to provide feedback on the workshop and identify aspects that were satisfactory from those that could have been improved.

Methods
A Google form was set up to collect answers from each participant on the seven following questions: 1. Now that you know better photogrammetry, do you think it could be a valuable tool for your work (if you were using photogrammetry before, please answer N/A)?* 2.
Do you now feel confident enough to be able to run a 3D reconstruction on your own (if you were using photogrammetry before, please answer N/A)?*

3.
Do you think that some aspects photogrammetry remained unclear or should have deserved more time?* 4.
Feel free to comment your answer.* 5.
Do you think that some aspects of photogrammetry were too extensively detailed?*
Do you think of anything that could improve the presentations?* 9.
On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how satisfied were you with the hands on?* 10.
How would you improve the hands on?* 11.
On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how satisfied are you with that workshop?*12.
Do you think of anything missing or that could be improved for that workshop?*N.B.All questions required an answer from the survey participants.

Results
All feedback participants consented to share their answers anonymously.The answer rate was of 34.4% (22 participants out of the 64 registered).
1. Now that you know photogrammetry better, do you think it could be a valuable tool for your work (if you were using photogrammetry before, please answer N/A)?
From the 22 participants that answered the survey, 15 did not use photogrammetry prior to this survey (Fig. 13).From those 15 participants, 13 considered photogrammetry to have potential for their own work after the workshop.
2. Do you now feel confident enough to be able to run a 3D reconstruction on your own (if you were using photogrammetry before, please answer N/A)?Answers to Question 1 of the feedback survey: 'Now that you know photogrammetry better, do you think it could be a valuable tool for your work (if you were using photogrammetry before, please answer N/A)'?
Despite the workshop, 10 beginners out of 16 still required supervision for photogrammetry (Fig. 14).This reflects the difficulty to adapt the workshop for beginners when working with time constraints.While we intended to provide an overview of the overall workflow, there remains space for them to play with their own data to empower them even more.Still, one participant attempted a reconstruction by his/her own after the workshop.
3. Do you think that some aspects of photogrammetry remained unclear or should have deserved more time?
A majority of 10 responders pointed out the need to investigate more the software from pre-processing to post-processing (Fig. 15).Five attendees found that all aspects were clear enough.
4. Feel free to comment on your answer.
• "A handout for the software usage (a "cheat sheet") for a simple case"; • "Data quantity, best practice for handling large datasets"; • "Practical acquisition and processing of ROV data"; • "Short time frame and high number of people"; • "The hands-on session covered aspects that remained difficult to understand only with presentations".Answers to Question 2 of the feedback survey: 'Do you now feel confident enough to be able to run a 3D reconstruction on your own (if you were using photogrammetry before, please answer N/A)'? 5. Do you think that some aspects of photogrammetry were too extensively detailed?
Overall, 17 participants considered that no aspect was too extensively detailed (Fig. 16).
6. Feel free to comment your answer.
• "The level of detail was good".• "This workshop provided a great overview of photogrammetry applications".
7. On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how clear were the presentations?
With a weighted average score of 4.41/5, the audience was satisfied about the presentation clarity (Fig. 17).
8. Do you think of anything that could improve the presentations?
• "More time for interaction between speaker and the attendees was needed"; • "Difficulty to see and hear"; • "Presentations on software".9. On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how satisfied were you with the hands-on?With a weighted average score of 4.23/5, the audience was satisfied with the hands-on activities (Fig. 18).
10. How would you improve the hands-on activities?
• Give the option for attendees to discuss with the stand mentor what they want to practise; • Smaller groups and not having groups in the same room were recommended several times; • Providing more details on software functionalities during the hands-on session was recommended twice.More time for the hands-on and the possibility to attend different stands was also raised.
11. On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how satisfied are you with that workshop?
With a weighted average score of 4.55/5, the audience was generally very satisfied of the workshop in overall (Fig. 19).12. Do you think of anything missing or that could be improved for that workshop?
• A cookbook was recommended; • Possibility to participate to all group activities was also recommended; • Giving the possibility for participants to bring their own data was suggested in order to discuss photogrammetry under various case studies and practicalities and help the audience to deal with their data at the same time; • For a next edition, longer discussion panel (e.g. 2 hours) is suggested to better identify the needs and the challenges that the growing photogrammetry community will likely face in the future (e.g.data publication standardisation, novel acquisition technologies, generalisation of open-source photogrammetry software).

Conclusion
This report outlined the architecture and organisation (contents and manpower) of a oneday photogrammetry workshop delivered to a scientific audience with no prior knowledge about this technique.In addition, it uses semi-quantitative approach, based on surveys to assess the relevance and the impact of the workshop.This report could help plan future photogrammetry workshops targeting a scientific community.We stress the need to survey participants prior to the event to better capture participants' experience and needs.
Activities included theoretical lessons on data acquisition followed by practical hands-on and case studies presentations.Despite the difficulty to empower attendees with high confidence on the use of photogrammetry, the workshop helped attendees to identify the potential for photogrammetry use for their own case studies.Despite logistical constraints (e.g.time and room layout), the audience was satisfied, demonstrating the success of this workshop.Future offering of such a workshop could be held in a hybrid in-person/online format and could include additional days to allow participants to attend all hands-on sessions, to lead a real field photogrammetry sampling, to practise more with the different photogrammetric software or even to give them the opportunity to play with their data in groups from beginners to more advanced users.Finally, this report not only sets the stage for a photogrammetry training, it also provides details on a wide diversity of workflows and applications that could help to identify current limits, future needs and ways for standardising photogrammetry in the future.
pool for the mini ROV demonstration and their compliance with logistical request.The organising committee would like to thank Romain Pinel who kindly offered his help for facilitating the workshop.Finally, we are grateful to the participants of this workshop who contributed in great number to the surveys of this report.

Figure 2 .
Figure 2. Pie chart of the answers to Question 1 of the preliminary survey: 'What is your current position'?

Figure 4 .Figure 5 .
Figure 4. Poll of the answers to Question 3 of the preliminary survey: 'What ecosystem(s) does your work relate to'?

Figure 6 .
Figure 6.Poll of the answers to Question 5 of the preliminary survey: 'What aspects of photogrammetry do you particularly intend to focus on during the workshop'?

Figure 8 .
Figure 8. Detailed programme of the "3D underwater mapping for habitats and biological communities" GeoHab 2023 workshop (Morning session).

Figure 9 .
Figure 9.Detailed programme of the "3D underwater mapping for habitats and biological communities" GeoHab 2023 workshop (Afternoon session).

Figure 10 .
Figure 10.Diversity of photogrammetry models, acquisition methods and applications presented during the workshop.(A) 3D model of Xestospongia testudinaria from scuba diving -Mozambique (2020) © Creocean; (B) Acropora sp coral nubbin 3D model for temporal survey.© Urbina-Barreto I. Future Maore Reefs project -French National Institute for Sustainable Development (IRD); (C) Coral reef model in Mayotte underwater photogrammetry, artistic and awareness actions.© Urbina-Barreto I. Future Maore Reefs project & Simon R. CORAUX project.French National Institute for Sustainable Development (IRD) & OFB Natural Marine Parc of Mayotte; (D) 3D model of the seabed geomorphology at the Dellec shoreline in the French Brittany using the Poseidon floating platform © TELEMAC project; (E) coral reef photograph acquired with an unmanned aerial vehicle -Belizes © National Oceanography Centre; (F) 3D model of Antonio Lorenzo wreck from scuba diving -La Reunion (2017) © Geolab -Comité de plongée Réunion; (G) Inspection of submerged structures for coastal road construction -Reunion Island.© Urbina-Barreto I. PhD thesis (2020); (H) 3D model of a cold-water coral reef imaged with the remotely operated vehicle Isis -Whittard Canyon, -850 m (2015) © JC125, National Oceanography Centre (I) 3D model of the > 10 m-high Capelinhos vent edifice with the remotely operated vehcle Victor6000 -Lucky Strike vent field, -1665 m (2020) © MoMARSAT 2020, IFREMER.All pictures have been reused under CC by 4.0.

Figure 11 .
Figure 11.3D model of a scene rebuilt by the participant during the hands-on session of L. Van Audenhaege (n = 42 images) and displayed on Meshlab.Note the small vignettes on the upper part to automatically scale the model in Agisoft.Note also a 3D map of La Réunion topography on the lower right part of the picture.
Figure 12. 3D model of a scene rebuilt by the participant during the hands-on session of I. Urbina-Barreto, S. Delsol and C. Delamare (n = 54 images).Hands-on was divided in image acquisition (coordinator: Delamare C.), processing with Agisoft Metashape and Open Drone Map and computation of ecological analyses (Urbina-Barreto I.) and examples of point cloud comparison (Delsol S.).

Figure 15 .
Figure 15.Answers to Question 3 of the feedback survey: 'Do you think that some aspects of photogrammetry remained unclear or should have deserved more time'?

Figure 16 .
Figure 16.Answers to Question 5 of the feedback survey: 'Do you think that some aspects of photogrammetry were too extensively detailed'?

Figure 17 .
Figure 17.Answers to Question 7 of the feedback survey: 'On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how clear were the presentations'?

Figure 18 .
Figure 18.Answers to Question 9 of the feedback survey: 'On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how satisfied were you with the hands-on?'

Figure 19 .
Figure 19.Answers to Question 11 of the feedback survey: 'On a scale from 1 (dissatisfied) to 5 (highly satisfied) by 3 (neutral), how satisfied are you with that workshop'?