How data science and drones combine to assess Ireland’s habitats

14 Oct 2020

Image: Charmaine Cruz

Charmaine Cruz of DCU and Insight is finding out if drones can be an effective data collection tool for farmers, climate scientists and others.

Charmaine Cruz undertook BSc and MSc degrees in geomatics engineering at the University of the Philippines Diliman and graduated in 2018. In 2019, she moved to Ireland to study for a PhD at the Insight Centre for Data Analytics, based at Dublin City University (DCU).

Her research interests include the use of geospatial technologies in natural resources management and studies related to the climate crisis.

What inspired you to become a researcher?

My interest in research started when I received an opportunity to be involved in a lidar mapping project in 2012. At that time, in the Philippines, lidar was quite a new technology.

During my seven-year involvement in research projects in the Philippines, I found myself passionate about two things: science and the environment. My research focuses on doing science to enhance environmental protection.

I am very interested in using geospatial technologies, such as geographical information systems (GIS) and remote sensing for natural resource management – for example, mapping habitat distribution and biomass as well as habitat condition.

I am also interested in environment and climate change-related studies that support conservation and sustainable development initiatives. I love doing research and I get excited when I learn new ideas and methods and when they yield new insights. Also, I really enjoy the extra things that come with research, such as being in the field (or ocean) to collect my data.

Can you tell us about the research you’re currently working on?

I am working on a remote sensing project called ‘Habitat Mapping, Monitoring and Assessment using High-Resolution Imagery’ or iHabiMap.

It is a multidisciplinary project that is developing analytical approaches to determine if images acquired by drones can be analysed using machine learning techniques to effectively map, monitor and assess three types of Irish habitats – uplands, grasslands and coastal.

I’m working together with a great team consisting of experts from remote sensing, machine learning, ecology and geography. This project is for four years and is funded by the Irish Environmental Protection Agency.

This summer has included an extensive field campaign. We are collecting UAV [uncrewed aerial vehicle] images and ecological field data at five study sites throughout Ireland. We will collect data at the same site several times to capture the inter-seasonal and inter-annual variability of the habitats.

One of our sites is a grassland site. It is very dynamic and varies widely within a growing season, between April and September. It is necessary that we start recording at this site early in the year and continue until close to the end of the growing season. This could help us to accurately and effectively classify the vegetation and vegetation dynamics on site.

In your opinion, why is your research important?

In my opinion, my current research is important because it has the potential to be used in effective mapping and monitoring the environment. This is particularly useful because Ireland, as part of the EU, is required to conduct assessments of ecosystems and to report the conservation status of its habitats to the European Commission every six years.

Delivering this report is a huge task for a periodic national assessment, if only field-based collected data is the source of information. So there is a need for another approach that could complement effectively with the existing habitat mapping and monitoring efforts.

What are some of the biggest challenges you face as a researcher in your field?

There are numerous challenges to environmental research. The weather is one: overcast conditions and high winds can result in poor data quality.

As the spatial, spectral and temporal resolutions of the data get higher, file size also becomes larger. Huge file storage space and high computing power computers should be considered when conducting large remote sensing projects. Calibrating the classification algorithms and validating the results require ground-truth data. Finally, data availability is a challenge.

Doing this kind of environmental research requires different datasets such as weather and topography. However, not all these data are either available to the public or applicable to use in remote sensing analysis.

Are there any common misconceptions about this area of research?

For me, it is the level of detail that remote sensing can provide. The remote sensing images that we use are only as good as their resolution.

The spatial resolution of the images can be anywhere from a few hundred meters (lower resolution) to centimetres (higher resolution) per pixel. Lower resolution images are often free, while higher resolution images are usually commercial products. Although remote sensing has been effective in a wide range of applications, it is still not a panacea that will provide solutions to every research field.

Also, collecting ground-truth data is a necessary step to assess the information derived from classifying remote sensing images, otherwise it will be just a pretty, colourful map.

What are some of the areas of research you’d like to see tackled in the years ahead?

I am keen and enthusiastic about my current research field because I am passionate about environmental protection. I also enjoy spending my free time surrounded by nature. I would also love to focus on solution-oriented work, rather than just being research-oriented.

Solution-oriented research will allow me to address the pressing environmental issues of our times by providing my community with science-based information that is needed for effective mitigation, protection and sustainable utilisation of our natural resources and environment.

Are you a researcher with an interesting project to share? Let us know by emailing with the subject line ‘Science Uncovered’.