Florida’s damaging-weather season peaks in June, not September

Ten years of NOAA records, counted as distinct storm days rather than reports. Tropical systems account for 62 of them. Thunderstorm wind accounts for 817.

Tropical systems are 62 days in ten years. Thunderstorm wind is 817.

This is a compilation of public data, not a study, and it sits with the rest of what we have and have not measured on the research and data page. We downloaded the NOAA NCEI Storm Events Database for 2015 through 2024, filtered to Florida, and counted. Every figure below comes from those files. Nothing is modelled, and where the data will not support a claim we say so rather than estimating.

The finding that reorganised the page: over ten years Florida logged 62 distinct tropical episode-days statewide, counting hurricanes, tropical storms and tropical depressions together. That is 6.2 days a year across the whole state. Over the same decade, thunderstorm wind logged 817 distinct episode-days and hail logged 300.

Hurricane season dominates the conversation about Florida roofs because a hurricane is a single memorable event that damages a great deal at once. The record says the recurring damaging weather is summer convection, thirteen times more often, and it peaks in June and July.

One record is one report, and we had the inflation factor too low

One methodological point governs everything below, and it is the reason most published storm counts are wrong.

The database logs one record per report, not one per storm. A single storm crossing three counties with several spotter reports produces several rows. We had been publishing that this inflates counts by roughly 1.7 to 2 times. Measuring it directly on this pull, Florida runs 7.07 reports per distinct episode-day across all event types. For tropical systems it is 16.32, because one hurricane generates a row for every forecast zone it touches. Our published estimate was too low and this page corrects it.

So the unit throughout is the distinct episode-day: a calendar date on which at least one event of that type was recorded. Where a raw row count appears, it is labelled a report.

Ten annual files were downloaded, none missing. Note that NCEI revises history: the 2024 file carried a compile stamp of 28 July 2026, the same day we pulled it, and 2017 and 2022 were also recent recompiles. These figures are as of that date and will drift.

Distinct storm days by month, 2015 to 2024

Distinct episode-days by month, all ten years combined. Read down the tropical column and then the thunderstorm wind column.

Hail runs three to five months ahead of the tropical peak

Hail runs on a different calendar from wind, and the gap matters if you are timing anything around it. Restricting to sizes generally associated with roof damage sharpens the picture considerably.

Florida is a wind market before it is a hail market

A question worth settling before anyone builds a Florida roofing programme around hail: is Florida a hail market at all? We ran the identical computation on Texas from the same files.

What the county rankings do and do not measure

The county rankings come with a caveat we would rather state than bury, because it changes how they should be read.

For hail, sixteen of the top twenty counties are metro or metro-adjacent, led by Duval at 35 episode-days, Palm Beach at 33, Broward at 31, Orange at 26 and Miami-Dade at 24. Pinellas, Sarasota, Manatee, Marion, Pasco and Alachua do not appear in the top twenty at all.

For thunderstorm wind the map inverts. The leaders are Duval at 124 episode-days, Leon at 116, Walton at 100, Jackson at 90 and St. Johns at 88. Every South Florida and Tampa Bay metro county is absent from that top twenty. That does not mean South Florida is safe from thunderstorm wind. Leon County recorded 473 reports across 116 episode-days, 4.1 reports per day, against Broward’s 1.5. The rankings measure spotter-network density and forecast-office practice alongside weather, and North Florida is watched more closely.

Tropical exposure does not discriminate usefully between counties at this sample size. Rolled up from forecast zones to counties, the top twenty spans 16 episode-days down to 11. Ten years is not enough tropical events to rank Florida counties against each other, and any page that does so is reading noise.

The damage field will not carry one

There is no dollar figure anywhere on this page. That is deliberate, and the reason is worth publishing because the same field appears in a great deal of storm marketing.

The NCEI property-damage field is blank on 10.1% of Florida records and reads zero on a further 65.3%. Only 24.6% carry any value at all. For hail specifically it is populated on 1.0% of records: nine reports out of 928 across ten years. There is no basis in this database for a statement about the dollar cost of Florida hail damage.

Completeness also varies by county in a way that has nothing to do with damage. Seminole populates a value on 9.8% of its records and Pasco on 75.4%, an eightfold spread inside one state and one database. That is a county-office practice difference.

The tropical figures fail differently and more dangerously. Where a value exists, it is frequently one episode-level estimate stamped onto every zone row. Hurricane Helene’s $1.00 billion appears on four separate rows sharing one episode identifier, covering four rural counties. Michael’s $100 million appears on five rows and $500 million on two more. Fred’s $50,000 is replicated across eighteen consecutive Panhandle zones. Fourteen of the thirty-two storm windows in this decade show the pattern, and summing them produces figures wrong by multiples. Hurricane Ian, the costliest storm in the window, carries a value on only 19 of its 70 rows.

The record shows noise, and rising reporting density

Distinct episode-days per year ran 178, 164, 172, 163, 158, 172, 156, 180, 190 and 149 from 2015 to 2024. There is no direction in that series. The lowest year of the decade is 2024, the year of Debby, Helene and Milton.

Raw report counts tell a different and false story. They rose from 939 in 2019 to 1,669 in 2023, which reads as a sharp increase in Florida storms. Over the same period reports per episode-day rose from 5.94 to 9.67. The apparent increase is reporting density. Any claim that Florida storms are becoming more frequent, built on row counts from this database, is measuring how many people filed a report.

One series has an arguable tilt: flash flood ran 8 to 20 episode-days a year from 2015 to 2019 and 20 to 33 from 2020 to 2024. With ten annual points and no significance test run, we are not calling that a trend.

For completeness on severity: of 560 Florida tornado records, 311 were EF0, 172 EF1, 38 EF2 and 8 EF3. No EF4 or EF5 tornado was recorded in Florida in this decade.

The 25 percent rule, the SB 4-D exception, and who may re-roof

Two regulatory facts sit behind every Florida re-roof, and both are widely misreported. Market-level conditions for the state sit on the Florida market page.

The 25 percent rule still exists. Florida Building Code, Existing Building, section 706.1.1 provides that no more than 25 percent of the total area or roof section may be repaired, replaced or recovered in any twelve-month period unless the entire roofing system is brought to current code. What changed is that Senate Bill 4-D, approved 26 May 2022 and effective the same day, created a statutory exception at section 553.844(5), Florida Statutes. Where the existing roof was built to the 2007 Florida Building Code or later, only the portion being worked on must meet current code. Roofs predating the 2007 code get no exception and the full-replacement trigger still applies to them. The statute also bars local governments from amending that exception by ordinance, so it is uniform statewide. No session law since has amended section 553.844.

The code in force is the 8th Edition (2023), effective 31 December 2023. The 9th Edition (2026) has a planned effective date of 31 December 2026 in the Commission’s own workplan. It is a planned date rather than a promulgated one, and it is not in force as we write.

Licensing. Florida licenses roofing contractors through the Construction Industry Licensing Board under Chapter 489, Part I. A general, building or residential contractor certified after 1973 may not act as or advertise as a roofing contractor without being certified or registered in that category. One detail worth knowing when checking a contractor: the board has been barred from issuing new registrations in these categories since 1 July 1993, so a roofer licensed in Florida today is in practice a state-certified roofing contractor. Verify certification through the department rather than accepting a licence number at face value.

What this data cannot support

Because a compilation like this invites claims it cannot carry, here is what these data do not license, stated plainly.

• Any dollar figure for Florida storm damage, at any geography. The field is unpopulated or zero on three quarters of records and replicates single estimates across zone rows.
• Any claim that Florida storms are increasing. distinct episode-days show no trend across the decade.
• Any per-household or per-roof rate. No population or housing denominator was applied, so nothing here supports a claim about what share of Florida homes were affected.
• Any claim about roof damage as such. This database records meteorological events. It contains no roof-damage variable. Every roofing inference on this page is an inference and is labelled as one.
• Anything about a specific address, ZIP code or neighbourhood. The finest geography here is the county or the forecast zone. A county episode-day means an event was recorded somewhere in that county.
• Precise hail sizes. The size distribution spikes at 0.75, 0.88, 1.00 and 1.75 inches because observers estimate against reference objects. Treat one inch as a bin rather than a measurement.
• Anything about insurance, claims, deductibles or premiums. No insurance source was consulted for this page.

NOAA NCEI Storm Events Database, bulk detail CSVs for 2015 through 2024, downloaded 28 July 2026 from ncei.noaa.gov. Ten annual files, 11,890 Florida records and 47,668 Texas records. Distinct episode-days computed as unique calendar dates parsed from the event begin timestamp within each event-type filter; report counts are raw row counts and are labelled as such. Hail magnitude is diameter in inches as recorded. NCEI revises historical years: the 2024 file carried a compile stamp of 28 July 2026 and the 2017 and 2022 files were recent recompiles, so these figures are as of that date. Florida Building Code, Existing Building section 706.1.1; section 553.844(5), Florida Statutes, as created by Senate Bill 4-D, 2022 Special Session D, chapter 2022-269, Laws of Florida, effective 26 May 2022; Florida Building Commission on the 8th Edition (2023) effective date of 31 December 2023 and the 2026 workplan. Chapter 489 Part I, Florida Statutes, sections 489.105(3)(e), 489.107, 489.111, 489.113(3)(g) and 489.117. Checked 2026-07-28.

Fair questions about this compilation

A measurement of your own beats a benchmark

A free Visibility Check runs twelve real buying questions across Google and AI answers, for your business and your market. You keep the findings.