In the 1990s, what was considered a 500-year flood was expected to occur, on average, once every five centuries. By the early 2000s, many of those same events were being reclassified as 100-year floods. By the 2020s, they increasingly resembled 10-year floods. Today, in some regions, comparable flood events are occurring every few years.
The flood itself did not change.
What changed was the climate system.
The Big Picture
Short-term stream gauge records can be heavily skewed by sampling variability, and it is absolutely true that local flood “return period” estimates can change significantly as more data are collected. That is basic hydrology.
But that is not the same as what is happening in the climate system.
We are not dealing with a single river gauge with a short record. We are dealing with thousands of independent observational datasets across independent physical domains—temperature, ocean heat content, atmospheric moisture, cryosphere mass balance, sea level rise, and precipitation extremes—all showing consistent directional change.
That is exactly why climate attribution studies and statistical detection work have been able to identify signals with high confidence across multiple decades and datasets.
On the policy side, agencies such as FEMA and others involved in risk modeling (including mortgage and insurance regulators) have updated flood maps, pricing assumptions, and risk frameworks over time. But that reflects evolving hazard conditions and improved modeling—not a refutation of climate-driven trends.
So the key distinction is this:
- A local statistical revision in a limited dataset
- Climate science is based on global, physically consistent, multi-variable observational convergence
They are not equivalent statistical problems.
Flood Insurance Policy
In the early 2000s, I met with representatives from FEMA and Fannie Mae. I went into that meeting assuming there was relatively little institutional movement toward aggressive long-term risk mitigation or structural adaptation.
I was pleasantly surprised.
At that time, FEMA had already begun shifting toward a more proactive risk-management framework, including elements of what is now often described as managed retreat in high-risk areas. There was a clear recognition that some locations face recurring or escalating flood exposure, and that long-term resilience cannot rely solely on repeated reconstruction in place.
A significant part of their effort was also focused on education—helping homeowners, lenders, and local governments understand how the U.S. flood insurance system actually works, including its limitations. Many people are still unaware that standard coverage is constrained by caps, exclusions, and policy limits that may not fully reflect total rebuilding costs, especially over the lifetime of a property.
The broader goal, as it was explained to me, was not just insurance pricing, but reducing systemic exposure over time through a combination of improved mapping, risk communication, and gradual migration away from the highest-risk zones where feasible.