Good question, and researchers have actually looked into this. A 2024 paper in Environmental Hazards found that after correcting for historical underreporting, the true long-term number of disasters appears surprisingly stable. Before 1970, only major events were logged and minor floods, heatwaves, and storms simply weren't recorded, so a lot of that "tenfold increase" is really just better paperwork. That said, economic damages from the most extreme events do show a real upward trend. The frequency numbers are inflated by better detection, but the worst events genuinely are getting worse.

Found it thanks! "Incompleteness of natural disaster data and its implications on the interpretation of trends" 

The detection problem isn't uniform across hazard types, which I think is worth keeping in mind. Earthquakes have had decent instrumentation since the early 1900s, so the completeness issue is much less severe there. Heatwaves and small floods though? Barely tracked until recently, and still underreported in many lower-income regions today.

Those also happen to be the events most associated with climate change in public conversation. So the correction ends up mattering most exactly where the debate is most charged. Just something to be aware of when citing the paper.

John Bradley wrote:

Found it thanks! "Incompleteness of natural disaster data and its implications on the interpretation of trends" 

@John Bradley Any chance you could share the link? 

Leo Roche wrote:

John Bradley wrote:
Found it thanks! "Incompleteness of natural disaster data and its implications on the interpretation of trends" 

@John Bradley Any chance you could share the link? 

https://www.tandfonline.com/doi/full/10.1080/17477891.2024.2377561

Here you go!