True but that actually strengthens the argument for being more selective about location not more relaxed about risk.
Hospitals, schools, power substations, and water plants should be the last things placed in uncertain or poorly mapped flood zones. When data is weak permanence raises the cost of being wrong.

But some infrastructure needs to be permanent like a hospital or a school.

BlokchainB

sometimes not even consistent rainfall records that makes precise return period labels like “500 year floodplain” more uncertain. But uncertainty should argue for more caution, not less. When data is thin, the safest assumption is wider buffers preserved floodways, and flexible land use not permanent structures.