How this storm history is built

Each city is placed at the population-weighted centre of its ZIP codes, counting the ZIP codes with at least 2,000 residents. Every distance on a city page is measured from that point, so the outskirts of a large city can be several miles nearer to a storm, or farther from it, than the figure shown.

Hurricanes and tropical storms: 75 miles. A storm is listed when the centre of its track passed within 75 miles of the city while it was at least a tropical storm, with sustained winds of 39 mph (34 knots) or more. That is the reach of damaging wind and rain for a typical hurricane. HURDAT2 gives a position every six hours, and a storm moving at 20 mph covers 120 miles between two of them, so the distance is measured to the line joining consecutive positions, not to the positions themselves. Being on the list does not mean the city took a direct hit; the "closest" column says how near the centre came.

Tornadoes: 25 miles. A tornado is counted when any part of its path, taken as a straight line from where it touched down to where it lifted, came within 25 miles of the city. Tornado paths are narrow, so counting only paths through the city would leave most cities with nothing on record; the wider radius describes how often tornadoes occur in the area around the city. A tornado that crossed a state line appears in the SPC file once for its whole track and again for each state; only the whole-track record is used, so it is counted once. Deaths and injuries are those of the whole tornado, not only those in the city.

Hail: 10 miles. A hail report is counted when it was made within 10 miles of the city's centre. Hail falls in swaths a few miles wide and is reported where people live, so a tighter radius than the tornado one keeps a city from collecting the next town's storms. Each report is a point where someone saw hail, and one storm can leave several reports near the same city; that is why the page also gives the number of separate days with hail.

"Strength when closest" is the sustained wind on the track at the point nearest the city, taken between the two surrounding positions in proportion to where that point falls between them. It is put on the Saffir-Simpson scale from the wind in knots: Category 1 from 64 knots (74 mph), Category 2 from 83 knots (96 mph), Category 3 from 96 knots (111 mph), Category 4 from 113 knots (130 mph) and Category 5 from 137 knots (157 mph).

A city's hurricane count includes only storms that were at hurricane strength at their closest point, so a storm that peaked far out at sea can count as a tropical storm for a city it reached later; the storm's own page gives its peak. The wind on the track is not the wind felt in the city: a hurricane's worst winds sit within a few dozen miles of its centre, and inland cities are usually past that.

Tornado ratings are as SPC records them: the original Fujita scale (F0 to F5) before February 2007 and the Enhanced Fujita scale (EF0 to EF5) from then on.

Some city pages carry a map of every HURDAT2 track that passed within 75 miles of the city. Each stretch between two six-hourly positions is split in half, and each half takes the colour of its own position: Category 1 to 5 by the Saffir-Simpson thresholds above, tropical storm from 34 knots, and grey dashes for a depression or a system HURDAT2 marks as no longer tropical (extratropical, remnant low, wave or disturbance). The three strongest passes by wind at the closest point are drawn thicker and labelled, then storms with a notable-storm page that lists the city, up to five labels. The map covers about 526 by 362 miles around the city.

The text under the map is computed from the same tracks. The direction is the way the track was moving at its closest point to the city, to the nearest of eight compass points. A landfall is a HURDAT2 position with the record identifier L within 75 miles of the city. "The last 30 seasons" are 1996 to 2025; the earlier rate runs from 1851 (1949 for cities reached only by eastern Pacific storms, whose HURDAT2 file starts then). Early decades are less complete: before aircraft reconnaissance and satellites, weaker storms that stayed offshore could go unrecorded, so a higher recent rate is partly better observation.

Coastlines and state lines are the U.S. Census Bureau's cartographic boundary files (2017 edition, as simplified in the us-atlas package); land outside the United States is Natural Earth's 1:10 million countries. Census boundary files are "free to use in a product or publication" with the Census Bureau acknowledged as the source and are meant for display at 1:5,000,000 or smaller, which the map keeps at its full width; Natural Earth is in the public domain. Boundaries are for orientation only. For now the maps are on the city pages readers reach most often from search.

Where a city page compares the city with its state, it counts the state's cities that have a page on this site, not every town in the state. When at least as many of them have a lower count as a higher one, the page gives the share with a strictly lower count ("more than 87% of the 171 Kansas cities"); otherwise it gives the share with a strictly higher count. Cities with the same count fall in neither group. Hurricanes are compared by the number of hurricanes, or by all tropical systems for a city that has had none at hurricane strength; tornadoes by the number of tornadoes; hail by the number of reports. States with fewer than five such cities get no comparison.

Each storm page opens with a summary written from the storm's HURDAT2 track and its matches to cities: how long it was tracked, when it first reached tropical-storm, hurricane and major-hurricane strength, its peak wind and lowest central pressure, its landfalls near US cities, and the cities and states it came within 75 miles of. A landfall is a point HURDAT2 itself marks with the record identifier L, where the centre crossed a coastline; one counts as near a US city when a city on this site is within 50 miles of it. Storms without such a mark are described by their closest approaches, never by an assumed landfall.

The full summary goes to storms that were Category 4 or 5 at their closest pass to a US city, storms since 1990 with a landfall at Category 3 or stronger near a US city, and the storms on the National Hurricane Center's Hurricanes in History page; every other storm gets a shorter one from the same figures. Unnamed storms on that page carry the name the Hurricane Center uses — the Galveston Hurricane of 1900, for example — next to their HURDAT2 number. The summaries say nothing about damage, deaths or tornadoes: HURDAT2 does not record them, and tornadoes near a track cannot be attributed to the storm from the SPC file alone.

A city's weather table comes from the nearest station that has normals for the high, the low and precipitation in all twelve months, each from at least ten years of readings, and only when that station is within 25 miles; the page names the station and its distance. Normals describe the station: a city on the coast with an inland airport, or a valley town with a station on a ridge, can differ from them. 13,604 of 13,642 cities have a station in range.

The hurricane season sentence (June to November against December to May) is written only for cities with at least 10 tropical systems on record. Water temperature is shown when a NOAA tide station with a water sensor is within 25 miles and has at least three years of readings in every month, each month at least half covered. The sea level trend is NOAA's linear trend for the nearest tide gauge within 50 miles, shown only for cities with a tide station within 25 miles; the trend describes that gauge, where the land itself may be sinking or rising.

A state has a weather-by-month page when at least 10 normals stations in it serve its city pages and at least one of its city pages is in search; 38 states qualify. The page takes the median of those stations month by month, names the station with the highest average high, the lowest average low and the most and least precipitation in each month, and counts which month is wettest and driest station by station. States with at least 10 tropical systems within 75 miles of their cities add the hurricane season by month: each system counted once, in the month of its first pass near one of the state's cities. States with NOAA tide stations add their monthly water temperature.

The notable storms were chosen by a rule fixed before the pages were built: the 10 tropical systems with the most NFIP flood insurance claims under their name; then the 6 with the most deaths in National Weather Service reports that name them (reports from 1996 on, dated within the storm's track, whose account says “Hurricane”, “Tropical Storm” or “remnants of” with its name); then the 4 named systems with the strongest wind at their closest point to a city on this site; and the 6 deadliest tornadoes since 1950 plus the 2 deadliest since 2012, from the SPC database: 28 pages. A tornado's NWS account is the one logged for the county where the SPC track begins; disaster declarations are those for its state with an incident start within three days.

Each year in numbers (2020–2025) adds up the same files for one calendar year: tropical systems that passed within 75 miles of a city here, SPC tornadoes against the 1991–2020 average, SPC hail reports, all NWS Storm Events reports, NFIP claims by year of loss and FEMA declarations by declaration date. Claims for recent years keep growing as they are settled.

A city's count covers flash floods, floods, heavy rain, thunderstorm wind and lightning whose starting point in the Storm Events Database is within 10 miles of the city, from 1996; the section is shown when there are at least 5. Coastal flooding, storm surge and heat are logged by forecast zone or county without a point, so they are not counted. Deaths and injuries add the direct and indirect figures. Damage is the NWS estimate of property and crop damage in the dollars of the year, not insured losses. The three events listed are those with the most deaths, then injuries, then damage; the NWS account of an event is quoted on the page of the city nearest to where it began.

Claims are added up over every ZIP code that carries the city's name, from 1978. A city is shown with at least 10 claims, and a named event or a year only with at least 10, so no figure describes a single household. Payments are the building and contents payments in the dollars of each year, not adjusted for inflation. A named flood event links to its storm page when the name and year match a storm on this site. Disaster declarations are counted by county, once per disaster number, so every city in a county shows the same line.

The National Risk Index sentences describe the county too: its composite rating, the rating for each hazard above that has one (hazards FEMA marks as not applicable or without data are left out), its modelled expected annual loss from all 18 hazards and that loss's rank among the state's counties. Connecticut's counties are not rated: the index uses its planning regions instead.

A city has a page when at least one tropical system or tornado is on record near it, and every page stays up and linked from its state and from nearby cities. Only the cities with the fullest records are offered to search engines. Each city is scored on three layers, each weighted for severity: tropical systems (a hurricane counts twice, a tropical storm once), tornadoes (EF3 and stronger count twice) and hail (each day with a report, plus two for every report of 2 inches or more). In each layer the city gets its percentile, the share of all 13,642 cities with a lower figure, and the three percentiles are added, so a city with a long record in several layers ranks above one with a long record in only one.

The 1,500 cities with the highest score are indexed and listed in the sitemap, and so are 223 more that readers were already landing on in the 30 days before this rule was set (October 2026). The other 11,919 ask search engines not to index them and carry no advertising. A page whose hurricanes, tropical storms, tornadoes and days with a hail report add up to 3 or fewer (136 cities) is never in the indexed group. Any city's record can be set against another's with the comparison tool.

Early years. Atlantic records before aircraft reconnaissance in the 1940s and weather satellites in the 1960s are less complete: storms that stayed away from land and shipping could be missed, and positions and winds for early storms are estimates. Tornado and hail records before the 1990s are thinner, because weak tornadoes and small hail in open country often went unreported before Doppler radar and mobile phones. A rise in the decade counts is mostly better reporting, not a change in the weather.

Unnamed storms. Atlantic storms were not named before 1950, and HURDAT2 lists them by number only. They appear here as "Unnamed 1893 storm" and so on, with the NHC number on the storm's page. Names are reused, so every storm is named together with its year.

Hail sizes and reporting. Sizes are hailstone diameters in inches as reported, with the names from the Storm Prediction Center's size table (1 inch is quarter size, 1.75 inches golf ball, 2.75 inches baseball, 4 inches softball). The Weather Service raised its severe-hail threshold from 3/4 inch to 1 inch in 2010, and smaller hail is seldom in the file, so counts of small hail do not compare well across decades. Hail is on record only where someone saw and reported it. Rows that repeat a report exactly (same time, place and size) are counted once, and sizes above 8 inches, the largest hailstone measured in the US, are treated as unknown.

Distances at the edge of a radius. Distances are worked out on a flat map centred on the city, which is accurate to a small fraction of a mile at these ranges. A track or report almost exactly on the edge of a radius can still fall on the other side of it than a great-circle calculation would put it: three hail reports from one spot 10.0 miles from Pensacola are counted here and would not be by a great-circle measure.