From the Sydney Opera House to the ancient marshlands of southern Iraq, thousands of the world’s most treasured heritage sites sit close enough to the coast to face growing exposure to flooding and erosion as sea levels rise.
And, until now, no publicly available dataset existed to show – at a global scale and in fine spatial detail – exactly where these sites are and how far their boundaries extend.
That gap has finally been closed by the Global Coastal Heritage (Glo-CoH) dataset, published in Scientific Data, a journal of the Nature portfolio. The paper, “Mapping the global distribution of coastal World Heritage Sites: Glo-CoH, a new Global Coastal Heritage dataset”, digitised the boundaries of every UNESCO World Heritage property situated at or below 20 metres above mean sea level – the zone most exposed to coastal hazards. The result is a dataset of 1 250 individually mapped sites and component parts, covering more than 235 million hectares.
The project was jointly conceived by Dr Nicholas Simpson, chief research officer at the University of Cape Town’s (UCT) African Climate and Development Initiative (ACDI) together with David Bescoby (University of East Anglia), Joanne Clarke (University of East Anglia and Durham University), and Robert Nicholls (University of Southampton).
“There has been no shortage of studies warning that heritage sites are exposed to climate change,” says Dr Simpson. “What has been missing is the basic spatial data needed to actually quantify that exposure, site by site, at a global scale. Glo-CoH gives researchers and heritage managers a common, freely available foundation to build that analysis on.”
The dataset was produced by a team of 14 digitisers working from institutions across three continents, coordinated through a hierarchical quality-control workflow to ensure consistency across the full dataset. Among them were several UCT postgraduate researchers, whose contribution the authors were keen to recognise: Chukwuma Okolie, who went on to undertake the dataset’s independent sample validation and is a co-author on the paper; and Kevin Msungu, Kumbirai Zingore and Katie Carlson, who carried out digitisation work across batches of sites assigned to them.
“This was, frankly, onerous work – hundreds of individual site boundaries, each requiring careful judgement about what was actually visible on the ground,” says Dr Simpson. “The postgraduates who did this deserve real credit. Datasets like this don’t get built without people willing to do the detailed, repetitive work that never makes it into a headline.”
To ensure the dataset could be trusted for downstream risk modelling, the team developed a novel scoring procedure to independently validate a stratified 5% sample of sites, assessing the quality of source mapping, the visibility of site boundaries in satellite imagery, and the resulting positional accuracy of each digitised polygon. That validation exercise found that 92% of sampled sites met the threshold for optimal digitisation accuracy, with the small remainder flagged for correction.
The dataset covers 386 UNESCO World Heritage Sites that have components at or below 20 metres above mean sea level – of the 1 199 sites inscribed worldwide by the end of 2023 – split into 1 250 discrete polygons to allow precise modelling of exposure to sea-level rise and coastal erosion including for sites made up of multiple separate components such as islands or atoll elements.
UCT’s Climate Risk Lab, based in ACDI, is already using Glo-CoH to power two further global studies: a global sea-level risk assessment of coastal heritage sites; and an assessment of climate risk to biodiversity at these sites focused on thermal exposure of reef-building coral species and mangroves.
“This dataset isn’t just a one-off contribution – it’s already the backbone of two major studies coming out of our lab,” says Dr Simpson. “That’s exactly the kind of downstream value we hoped for: a foundational resource other researchers, including our own team, can build directly on.”
Clarke says: “This dataset represents years of painstaking, detailed work translating UNESCO’s own site records and satellite imagery into boundaries that are accurate enough to support real risk assessment. It’s the kind of unglamorous groundwork that almost never gets funded, and almost always turns out to be essential.”
Bescoby, who oversaw the digitisation process, adds: “Every one of these 1 250 sites had to be individually identified, checked against the source mapping, and hand-digitised by our team – often with only partial or outdated UNESCO documentation to work from. Getting the boundaries right, and validating that accuracy independently, was the whole point of the exercise.”
The research was supported by Irish Aid, via ODI Global’s Climate and Sustainability Programme; the British Academy’s ODA Challenge-Oriented Research Grants; and the UK Natural Environment Research Council’s Discipline Hopping for Environmental Solutions initiative.