Given the current flooding on the Mississippi, this from the Public Policy Institute of California’s discussion of flood risk in that state seems relevant:
Perception of risk directly changes pressure for improving flood management. Longer periods of time since a natural disaster reduce the perception of risk—a phenomenon referred to as the “flood memory half-life.” The problem is well in evidence in Californian’s flood insurance coverage behavior, which peaked soon after the 1997 floods—the last large floods within the state—and has declined ever
update: Discussion on twitter (see here and here) suggests interesting questions about where else this might apply, and what data one might use to examine the extent to which it generalizes to things like earthquakes and drought. Any suggestions for a) other areas this might apply, and b) data one might use?