← Back

About the Tornado Impact Score

Raw tornado records rank events by physical measurements alone — wind rating, path width, track length. Those describe the storm, not what it did to the people and places in its way. The Tornado Impact Score (TIS) is an attempt at the second thing: a single number estimating how significantly a tornado affected a community, blending human cost and economic cost, with economic cost weighted by how much it actually meant to the local economy it hit.

Philosophy

The same dollar figure means very different things in different places. $50M in damage might be a rounding error for a major metro county's economy, or a genuinely catastrophic loss for a small rural one — so damage is scored relative to the affected counties' own GDP, not as a flat number. That relative impact is then weighted further: the same percentage loss is treated as more consequential in lower-income counties than wealthier ones, on the premise that a given loss is felt more heavily where there's less economic cushion to absorb it. Casualties are scored on their own tiered scale rather than folded into the economic side, since loss of life isn't reducible to a dollar figure.

Structure

• Damage bump — tiered points from inflation-adjusted (2024 dollars) damage: ≥$1B→50, ≥$100M→25, ≥$10M→10, ≥$1M→3. This is a flat floor independent of county wealth, so catastrophic dollar losses always score meaningfully even where the relative-GDP ratio looks modest.
• Fatality bump — tiered points from fatality count: ≥100→120+0.5×(fat−100), ≥50→85, ≥25→60, ≥10→40, ≥5→25, ≥1→15. The top tier is open-ended so historically exceptional death tolls keep differentiating from each other rather than capping out together.
• Injury bump — same tiered approach, smaller scale: ≥500→40, ≥200→28, ≥50→18, ≥10→10, ≥1→5.
• Economic ratio — total inflation-adjusted damage (summed across all counties on a multi-county track) as a percentage of those counties' combined GDP, capped at 100% for scoring purposes (the card display itself still shows the real, uncapped figure honestly). Every other factor here is deliberately tiered with a ceiling; this one originally had none, and in practice that let a single small-county outlier (Greensburg, KS 2007 — damage came to 280% of Kiowa County's entire annual GDP) outscore far more devastating disasters like Joplin. 100% is the cutoff because once damage matches a full year of county GDP, that already reads as complete local economic devastation — whether the true multiple is 1x or 3x isn't a meaningfully different signal for a broad-strokes score.
• Poverty weighting — the (capped) economic ratio is multiplied by a factor from 1.0× (affected counties rank at the very top nationally for GDP per capita) up to 2.0× (the very bottom), based on where the affected counties' average GDP per capita ranks against every county in the country.

TIS = damage bump + fatality bump + injury bump + (min(economic ratio, 100) × poverty weight)

Known limitations

• County GDP and population reflect 2024 only, used as a flat stand-in for every year in the dataset — official county-level GDP doesn't exist for years before 2001 at all, so there's no historically accurate alternative.
• Pre-1996 damage figures are themselves estimates derived from SPC's old severity-coded bins, not directly reported amounts — see the note on individual cards.
• Connecticut is unavailable (shows as a "Geography mismatch") — the state's tornado records use its traditional counties, while the population and GDP sources use the newer planning regions that replaced them, and the two don't share matching boundaries or codes.
• About 1% of all tornado records (concentrated more heavily among older F5/EF5s specifically) never had a county resolved for them at all, back in the original data conversion — these show as "No county data" rather than a fabricated score, and are a separate, pre-existing gap unrelated to the Connecticut issue above.
• Multi-county tornadoes sum the full GDP of every county touched, even ones only clipped briefly at the edge of the track — damage isn't actually distributed evenly across a track, so this can understate the ratio for a geographically compact but severe tornado that happened to graze a second, larger county.