Unusual but amazing experience as an early career visiting scientist during global pandemic

By Milica Stankovic

The year 2019 ended wonderfully when I found out that I will be joining the IORA Blue Carbon Hub at the Indian Ocean Marine Research Centre in Perth, as an early career visiting scientist for an 8 week program. The project I had proposed involved developing a new workflow to map seagrass meadows using commercial drones.

My fellow participants (Jacqueline Raw from South Africa and Lahiru Sandaruwan from Sri Lanka) and I were very warmly welcomed by Mat Vanderklift and Lauren Hardiman from CSIRO. During the first week they introduced us to the facilities of the Indian Ocean Marine Research Centre and to everyone working there. Everyone we met was very welcoming, nice, and always ready to help. Although, signs of the global pandemic of COVID-19 were very much visible in Perth (such as shortages of toilet paper in the stores). Mat and Lauren went above and beyond to make sure that we were well taken care of and that we had a wonderful experience. Unfortunately, day to day changes of travel requirements and border restrictions meant it was a very stressful time, but Mat and Lauren made sure that we were comfortable every step of the way. Due to the sudden border closure I had to return to Thailand very soon after my arrival, and they both went beyond the program requirements to ensure I was not trapped outside Thailand.

Research project
My proposed project was to develop a new workflow on drone image analysis and processing to distinguish seagrass species in the meadow and to compare to the current methodology using software such as Pix4D, AgiSoft and ArcGIS. However, due to the travel restrictions my project had to change slightly, as field data collections were not possible, but it turned out for the best as we decided to use recorded drone images from the various places across the Indo-Pacific. This allowed me to explore this workflow on a much larger scale and under different conditions.
The first part of my project was to check if the seagrass species can be spectrally differentiated. Although I did not have a chance to go to the field to collect the data, I was able to obtain a lot of metadata on tropical species spectra. The results showed that there is a slight difference between certain species of seagrass, but for most the species spectra is very similar, meaning that mapping of the species using spectra would be very challenging.

I was not discouraged by these results, and I continued to work on the second part of the project which is the actual workflow. The goal is to design a workflow which can be used for image processing and analysis without use of the commercial software for any analysis step. The workflow consists of image pre-processing, where the drone images are renamed, correctly aligned, rotated, colour corrected, and the spectral data is obtained from the images every few meters, after which image processing is done. The big difference between this methodology compared to the image stitching software is that this workflow works on each drone image (the images are not stitched), from which the spectra are extracted on a small scale. The current pre-processing workflow works on the intertidal seagrasses which are exposed during the lowest tide, but it is planned to incorporate water column corrections and to include subtidal seagrass meadows. The project is still ongoing and there are many more things to do, but the next steps include classification and accuracy assessment for the intertidal meadows.

I hope that the results of this project will help anyone working on seagrass mapping using drones. Although there is a large increase of studies which incorporate drones in the mapping of seagrass, there is still a huge gap of knowledge between the experienced practitioners and beginners. I hope that this workflow will fill in the gaps for all practitioners. Since this workflow will be available to anyone, I hope that the elimination of commercial software will increase the possibility of the seagrass mapping, especially in the Indo-Pacific region where mapping of the seagrass distribution is still limited.

Figure 1. Showing the results of the discriminant analysis on the spectra of tropical seagrass species

Figure 2. Drone image polygons of a site in Thailand. Each rectangular polygon represents one image, and the large polygon under them is the area of interest where the seagrass meadows are present.

Figure 3. Showing tiles from different images (each tile is one image) show the same area of the seagrass suggesting proper image alignment and rotation. The area on the tile includes front and side overlap of the various images.

Program overview

The opportunity to be part of the IORA Blue Carbon Hub early career visiting scientist program has provided me with so many valuable experiences, that I will keep on using through the rest of my career. Through the program I have been able to develop new programming skills in Python and learnt so much more about drones on the technical side, which as a biologist/ecologist we do not think about. The program encouraged us to develop collaborations in all aspects related to our project. The support and encouragement that I have received from Mat with this regard was the key instrument for my project. As my project had to change due to travel restrictions, Mat was very helpful in reaching out and I have received many valuable suggestions, help and inputs from Janet Anstee and Arnold Dekker from CSIRO, Karen Joyce and Lucas Langlois from James Cook University, Chris Roelfsema from University of Queensland and Katia Ballorain from the marine observatory of La Réunionin in Kelonia. Since I am currently working remotely on the project from Thailand, I have enormous gratitude, and appreciation for Nick Mortimer from CSIRO, who is devoting his valuable time and knowledge working with and teaching me, and without whom this project would not be possible.

Participating in the program remotely was not very challenging, as Mat and Lauren made sure that the transition was very smooth by having continuous support throughout the next few weeks and months. We have weekly Zoom meetings where we catch up on our work progress or sometimes lack of progress, but through this lockdown and working from home, the meetings became one of the best parts of the week. We were also welcomed to join a couple of “blue carbon” and mapping webinars hosted by CSIRO. This allowed me to really connect with the current research in the other regions and helped me to broaden my research horizons.

Being part of IORA Blue Carbon Hub early career visiting scientist program has improved many of my research skills and helped me to establish so many wonderful contacts with experts from various fields. Also, I am very glad that I had the chance to meet and become friends with Jackie and Lahiru, from whom I had the opportunity to learn, share experiences and stress out during the first weeks of the pandemic when we did not know what to do. I am sincerely grateful to the Hub and Mat for granting me this opportunity and I am sure that many future participants will have amazing experiences and chances to improve their research.