Making drone data FAIR
Drones: Image by Pexels
The use of drones to capture data is increasingly common across a range of disciplines, particularly the physical and environmental sciences. While this data is valuable for reuse, its provenance is particularly complex. In collaboration with UCT eResearch and the University of Notre Dame’s Center for Research Computing, Dr Jane Wyngaard of UCT’s Department of Electrical Engineering is working to build an open-source data toolkit called LANDRS (Linked-data API for Networked Drones) for capturing and publishing drone-captured data. In conjunction with other tools being developed - such as the Australian Scalable Drone Cloud – LANDRS tools will allow researchers to process and publish their drone data more easily and in a way that fully captures and exposes the data’s provenance thus helping to make it findable, accessible, interoperable, and reusable (FAIR).
Increasingly, researchers, including oceanographers, climatologists, and ecologists are collecting valuable data 120 metres above the planet, to improve both our monitoring and understanding of the climate and environmental processes. Drone-data collection is adding a valuable new scope to so-called ‘small science’, explains Wyngaard.
“Individual researchers, who have collected enough funding to buy a drone, generally lack data management support that larger institutions typically provide. They are capturing valuable data all over the planet but that data often goes no further than that single researcher and their immediate work.”
“Individual researchers, who have collected enough funding to buy a drone, generally lack data management support that larger institutions typically provide. They are capturing valuable data all over the planet but that data often goes no further than that single researcher and their immediate work.”
“While this is a common problem in research, the loss of drone data for reuse is a particular concern because of its high potential reuse value.”
“The problem,” she says, “is that the provenance pipeline for drone data is particularly complicated for scientifically-robust reuse of the data, even within the same discipline.”
Wyngaard’s goal is to provide researchers who do not have access to extensive data curation support, with the tools needed to publish their drone-captured data with sufficient metadata for its reuse, even across disciplines. An added complication to this challenge is the need to align discipline-specific ontologies. Different disciplines use different words to describe different and even the same parameters or phenomena. In order to fuse data from different domains, an understanding is needed as to how these terms relate to one another.
The use of formal ontologies, in combination with emerging semantic web technologies, can make this process automatable. The toolkit takes advantage of this possibility providing automated data annotation that archives and publishes metadata that is readable by both machine and human. This in turn can be used to make data more FAIR at the cost of minimal manual intervention, despite the complexity of the drone data pipeline.
Australia’s Scalable Drone Cloud
Wyngaard also enjoys a long-standing relationship with a group of Australian researchers, who, under the umbrella of the Australian Research Data Commons, are working to build a national ecosystem for drone-data management called Australia’s Scalable Drone Cloud. A visit to Australia in early 2020 by UCT’s Vice-Chancellor Professor, Mamokgethi Phakeng, and Deputy Vice-Chancellor for Research and Internationalisation, Professor Sue Harrison, resulted in plans for a more formal research collaboration between the two groups.
Connecting with UCT eResearch
The increase in rate and volume of data being captured by researchers in general, in combination with advances in data technologies, societal need, and a drive from funders to publish data, has sparked a global trend of eResearch facilities pairing with science experts to improve data publishing in general. Most recently, this has included eResearch facilities seeking to also support science drone users. Wyngaard has thus been collaborating with UCT eResearch Director, Dr Dale Peters, to support drone research at UCT. The goal of this collaboration is to support UCT researchers using drones by finding ways of deploying the LANDRS tool stacks developed by Wyngaard and the Australian Scalable Drone Cloud.
The end goal is to create an end-to-end support infrastructure for small-scale drone research.
The end goal, explains Wyngaard, is to create an end-to-end support infrastructure for small-scale drone research, by combining the data publication pipeline she is building, with the drone-data cloud-processing pipeline the Australian group is building with eResearch support.