Katrina reached Category 5 over the warm Gulf of Mexico, peaking at 150-knot winds and a central pressure of 902 mb, before weakening to Category 3 for its Louisiana landfall near Buras on 29 August 2005. The wind at the coast was survivable. The water was not. The storm had already pushed a vast surge across the shallow Gulf shelf, and in New Orleans the federal levee and floodwall system failed in more than fifty places. Around 80% of the city flooded, some districts under three metres of water, and roughly 1,392 people died across the Gulf Coast.
For risk carriers Katrina is the defining lesson in correlated loss. The peril was wind, but the catastrophe was water finding a single engineered point of failure, so losses that models had treated as independent all arrived at once. It became the benchmark US insurance event at roughly $65bn insured in 2005 dollars, and it exposed the fault line between wind cover and flood cover: private insurers paid wind, the federal flood programme paid inundation, and homeowners were caught in years of disputes over which had destroyed their homes. The protection gap that remained, between a $125bn economic loss and what was insured, is why Katrina still frames catastrophe underwriting today.