Aridalia Research: 5 Florida Humidity Datasets
Every article on this site rests on data we pulled and counted ourselves: five federal and public datasets, more than half a million records about how Florida lives with humidity. This is where all five sit, with their methods and their main findings. These are primary sources. You can cite them, and you can check them: the data is open and the procedure behind each figure is written down.
Why five
Each dataset answers a different question. NOAA tells us what the humidity actually is, city by city, across thirty years. CPSC records what went wrong with the machines, including the fires and the recalls. ENERGY STAR shows what those machines do under test, and who really stands behind the brand names. Amazon reviews carry what 62,000 owners said about living with them, with the Florida cohort separated out. FEMA records what water does to a Florida house, across 448,000 insurance payouts.
Together they close a loop: climate, then machine, then owner, then consequence.
The five datasets
Florida humidity, thirty years of NOAA data
Hourly station records for five cities: Key West, Miami, Tampa, Orlando and Jacksonville. Two windows: 2015 to 2024 for all five, and 1995 to 2024 for the three largest metros. Key West holds a dew point above 70°F for more than 200 days a year, and through the summer the dew point sits at 73 to 75°F across every major metro in the state. The city pages under humidity are built directly on this extract.
Read the NOAA analysisWhich dehumidifiers caught fire
The full CPSC recall record and the SaferProducts complaint dump, filtered to dehumidifiers. Nine recall actions between 2001 and 2023, covering roughly 11.9 million units in the United States, every one of them for overheating or fire. Three factories account for about 9.7 million of that total. Complaints in the database are submissions filed by consumers, not confirmed defects.
See the recall historyThe ENERGY STAR register: 535 certified models
The complete certified products register, pulled directly rather than browsed. Tested portable capacity stops at 50.0 pints per day, and 128 brand names trace back to 75 filing companies. The register is also where the multi-brand families become visible: one filing, several labels.
Open the register breakdown62,317 owner reviews
Our analysis of the Amazon Reviews 2023 research dataset from the McAuley Lab at UC San Diego. Noise is a common theme in the Florida slice: 15.6% of reviews against 9.9% nationally. On top of the state cohort we counted cohorts by where the machine stands, in condos and apartments, in bedrooms and in garages. Aggregates only, no quoted reviews.
See the review analysisFlorida water damage hotspots
448,536 NFIP flood claims from 1978 to 2026, worth $19.31 billion. Houses built before 1980 account for 71.4% of the money. Basements make up 0.6% of the register for the entire state, and slab-on-grade houses carry a median payout seven times higher than a house on pilings.
Open the claims mapBeyond the five: who does the work
The five datasets above describe what happens to a Florida house. This one describes who fixes it, which is a different question and a different kind of record. It is not climate, not a product and not a loss. It is the market itself, as the state records it.
Mold remediators by county
Florida publishes an open register of every licensed mold assessor and remediator in the state. We counted it: 5,306 active licenses held by 3,936 people, because one person in three holds both. Twelve counties have no active remediator on the books. Broward runs 1.57 times ahead of neighbouring Miami-Dade per head of population, and we tested three explanations without finding one. The state overwrites this file weekly and keeps no archive, so a time series only exists if somebody saves the copies.
Read the licence register analysisHow we work with data
Three rules. Primary source, not a retelling: every extract comes straight from the system of record, and the retrieval date sits on the page. The method is named: each page states what we took, what we filtered on, and what counted as a payout, a review or a record. The limits are named: each page carries a section on what its data does not show.
The full account of sources and procedures is on our methodology page.
For journalists and researchers
The figures and charts in this research may be quoted with attribution to Aridalia. The method and sources behind each extract are described on the pages themselves.