US Extreme Weather and Climate Change Dashboard

Loss Normalization

A disaster loss "normalization" asks what a historical storm, tornado outbreak, or flood would cost if it happened today, adjusting for the fact that there is simply more built up, more populated, and more valuable property in harm's way now than there was decades ago -- inflation, population growth, and growth in wealth per person, all combined. Economic loss data is not appropriate for the detection or attribution of changes in climate variables, weather and climate data should always be used for that purpose.

Methodology

Each series uses the "full" normalization method -- Consumer Price Index inflation plus growth in exposed wealth (population and per-capita/structure wealth) -- following each hazard's own published methodology, rebased onto a single common 2026-dollar basis. Hurricanes and tornadoes are treated differently from flooding for the "share of GDP" figure: for hurricanes and tornadoes it is a rescaling of the normalized-dollar series against a single current US GDP level, while flooding's share-of-GDP is the authoritative series (each year's loss divided by that same year's own real GDP), with its dollar figure derived from it. Full methods, base years, data sources, and references are in the Methodology & Sources PDF.

Hurricanes

Continental US hurricane losses normalized using the Weinkle, Landsea, Collins, Musulin, Crompton, Klotzbach & Pielke Jr. (2018) method, extended through 2025. Coastal development is the biggest driver of why raw damage totals have grown: Florida alone went from 445,397 homes in 1950 to over 9 million by 2019, a twentyfold increase in exposed property along the same coastline.

Tornadoes

US tornado losses normalized using the Simmons, Sutter & Pielke Jr. (2013) method, extended through 2025. Normalized losses have trended down, not up: the 1954-1963 decade averaged roughly $4.8 billion a year, versus roughly $1.9 billion a year in 2015-2025 -- a decline that tracks a real drop in violent (F3/EF3+) tornado incidence over the same period, not just changing exposure.

Floods

Flood's share-of-GDP figure -- each year's loss (Downton & Pielke Jr. method through 2019, extended through 2025 via NWS Storm Data) divided by that same year's own real GDP -- is the basis for both charts below; the dollar chart is derived from it, not the other way around. This surfaces two of the most catastrophic floods in US history as the largest normalized-loss years on record: 1913 (the Great Dayton Flood) and 1937 (the Ohio River flood), both well ahead of 2005 and 2017 -- a real, checked result, not an artifact.