Why numbers on heritage risk need careful reading
Headlines about monuments “threatened by climate change” usually compress a long chain of reasoning into one figure or one ranking. Behind the figure there is typically a climate projection, a map of where a site sits, an estimate of how sensitive its materials are, and a set of assumptions about what people will do in response. Each link in that chain carries uncertainty, and the uncertainty tends to disappear by the time the result reaches a headline.
This article explains how risks to monuments, historic towns and cultural landscapes are usually assessed, what the resulting models can reasonably say, and where they stop. It does not offer figures of its own, because the useful skill here is judging the figures other people publish.
Hazard, exposure and vulnerability
Most assessments borrow a framework from disaster-risk work that separates three ideas.
- Hazard is the physical change that could cause harm: sea-level rise, heavier rainfall, flooding, drought, wildfire, heat, or more frequent freeze-thaw cycles.
- Exposure is whether a site is actually in the path of that hazard. A coastal fortress and an inland chapel face different sea-level risks even if the regional climate trend is the same.
- Vulnerability is how badly a particular site would be affected. Porous stone, earthen walls, timber frames, painted plaster and waterlogged archaeological deposits all respond differently to the same conditions.
Risk is generally treated as the combination of the three. A site can sit in a high-hazard area and still be fairly resilient, or be in a moderate-hazard area and be fragile. This is why a regional map alone is a weak guide to any single monument.
How the evidence is typically gathered
Studies of heritage and climate often combine several kinds of input, and each has its own weakness.
Inventories and registers. Lists such as national heritage registers or the World Heritage List provide a defined set of places to analyse. They are not a neutral sample of everything valuable: what gets listed reflects history, funding, expertise and politics. A study that overlays a flood map on a register describes the listed sites, not heritage in general.
Spatial overlays. Researchers commonly place site coordinates over projected flood zones, erosion zones or temperature maps and count how many sites fall inside. This is quick and repeatable, but it depends on the resolution of the climate data. A coastal projection built for regional planning may be too coarse to say anything reliable about a single headland.
Condition surveys. Site visits record cracking, salt damage, biological growth, subsidence and previous repairs. These records are more specific than a map, but they are expensive, uneven in quality, and often one-off rather than repeated.
Material and laboratory studies. Tests on stone, mortar, timber or pigments can show how a material responds to moisture, salt or temperature swings. Lab conditions simplify what happens on a real site.
Local knowledge and historical records. Accounts of past floods, repairs and maintenance routines fill gaps that instruments miss, though they are harder to compare across places.
What risk models can and cannot predict
A well-built model can usually do a few things reasonably well. It can rank places by relative exposure under stated scenarios, flag where more monitoring or detailed survey is warranted, and show which hazards matter most for which kinds of material. These are comparisons and priorities, which is what planners often need.
What models generally cannot do is say when a particular site will be damaged, or by how much. Several reasons recur.
- Scenario dependence. Climate projections are conditional on assumptions about future emissions. A result is better read as “under this scenario” than as a prediction.
- Downscaling limits. Global models are translated into local conditions through further modelling, which adds error.
- Damage is not just climate. Tourism pressure, pollution, neglect, conflict, development and ordinary ageing are often entangled with climate effects. Separating them is a problem of confounding variables, and attributing a specific loss to climate is a form of the causal claims problem that careful studies treat with caution.
- Human response changes outcomes. Repairs, drainage works, relocation, documentation and changes in management can all alter what actually happens. A model that ignores adaptation may overstate loss; one that assumes generous adaptation may understate it.
- Slow processes are hard to observe. Gradual weathering shows up over decades, while most monitoring programmes and funding cycles last a few years.
Reading a headline number
When you meet a claim such as “a large share of sites are at risk,” a few questions usually help.
- What is the population? All sites, listed sites, or sites in one region?
- What does “at risk” mean? Located in a hazard zone, assessed as vulnerable, or observed to be damaged?
- Which scenario and time horizon? A result for the end of the century is a different statement from one about the next decade.
- Was anything observed? Overlay studies describe exposure, not damage that has occurred.
- Is there another explanation? If damage is rising, increased reporting, better surveying or heavier visitor use may partly explain it.
The last point matters because more attention can create more data. If more sites are being surveyed than before, a rising count of recorded problems does not by itself show that conditions are worsening.
Visitors, counts and the wider picture
Tourism pressure is a good example of a related factor that is easy to mix up with climate. Visitor numbers affect wear and funding at many sites, and debates such as the overtourism debate show how contested even the definitions are. Likewise, the way tourism measurement is carried out varies between countries, so comparing visitor figures with condition data takes care. Climate may also change when and where people travel, which affects both income for conservation and pressure on a site.
Values, not only physical fabric
Heritage is more than stone and timber. Landscapes, settlements and living traditions matter because people value them, and some effects of climate change fall on use and meaning rather than structure. Those aspects are harder to count, which is one reason risk models focused on physical fabric can give an incomplete picture. Related questions about how communities experience change are taken up in separate discussions of identity and practice.
A practical way to hold the evidence
Treat site-level climate risk numbers as structured estimates of where to look, not as verdicts. Prefer studies that state their scenarios, their data sources and their limits. Be wary of single scores that blend hazard, exposure and vulnerability without showing the parts. And remember that the most reliable information about any one place usually comes from repeated, local observation over time, which is exactly what is hardest to fund.