How the County Risk Index Is Built

The County Disaster Risk Index is a number we calculate. No government agency publishes it. It exists to let you compare one county's disaster history against another's on a single 0–100 scale, instead of reading four separate lists.

The four layers

• Federal disaster declarations — distinct FEMA declaration numbers that have covered the county, from the OpenFEMA archive (since 1953).
• Severe storms — notable NOAA/NCEI Storm Events in the county (events with a death, major damage or a strong tornado; tornadoes from 1950, all event types from 1996), plus storm deaths per 100,000 residents.
• Earthquakes — USGS-catalogued earthquakes of magnitude 4.5 or greater with an epicentre in the county (since 1900).
• Large wildfires — NIFC-mapped fire perimeters of 1,000 acres or more with a centroid in the county (since 1980).

How each layer becomes a 0–100

Raw counts are unfair to compare directly: a county with 150 wildfires and one with 150 earthquakes are not equivalent, and a few counties have enormous totals that would flatten everyone else. So for each layer we:

1. Take ln(1 + count) — a log scale that keeps the extremes from dominating. (Storm deaths are measured per 100,000 people first, then logged.)
2. Rank every county on that value and convert the rank to a percentile from 0 to 100. A county at the 90th percentile has more history in that layer than 90% of counties.

The weights

The composite score is a weighted average of the four percentiles:

• Federal declarations — 40% (the broadest and longest record, and the one that reflects real, official impact).
• Severe storms — 25%.
• Earthquakes — 20%.
• Wildfires — 15%.

The final score is that weighted average, then itself converted to a percentile so that "100" always means the single highest-scoring county and the scale uses its full range.

What it is not

• It is not a forecast. It summarises what has happened, and the past is an imperfect guide.
• It is not FEMA's National Risk Index, which models expected annual loss using exposure and vulnerability. Ours is simpler and backward-looking; theirs is the one to use for planning.
• It is not a safety rating and never a reason to avoid living somewhere. Two of the highest-scoring counties are also among the most populous and desirable in the country.
• Layers with a threshold (M4.5, 1,000 acres, notable storms) miss smaller events by design, so a low score can partly mean "nothing large enough to be catalogued".
• Earthquake and wildfire epicentres are matched to counties spatially; offshore and remote events (common in Alaska) fall outside county boundaries and are not counted at county level.

Updating

The index recomputes whenever any of the four source layers refreshes (roughly monthly). Every county page shows the counts behind its score, so you can always check our arithmetic.

Back to the Risk Index · About the Data