All About the Data.

Every storm on this site traces back to HURDAT2, the official Atlantic hurricane database maintained by NOAA's Hurricane Research Division. HURDAT2 records the position, maximum sustained wind, and central pressure of every known tropical and subtropical cyclone in the Atlantic Basin at six-hourly intervals, going back to 1851. It's the same dataset the National Hurricane Center itself relies on for historical analysis, and it's updated and reanalyzed by NOAA on an ongoing basis as new research uncovers more accurate wind speeds, tracks, and even entirely new storms that were missed in the era before satellites.

HURDAT2 flags the specific six-hourly points in a storm's life where it made landfall, but it doesn't sort those points by state, territory, or country on its own. To build my database, every flagged landfall point is checked against precise geographic boundaries I built by hand for every U.S. state and Atlantic Basin country or territory — not simple rectangles, but boundaries shaped to follow real coastlines, barrier islands, and the sometimes-odd angles of state lines where they meet the water. A storm's landfall only counts if it was at tropical storm strength or higher; tropical depressions are excluded entirely.

This is the part that takes the longest, and the part I care most about getting right. HURDAT2's six-hourly positions are the storm's center at fixed synoptic times — they weren't designed to pinpoint the exact moment and place a coastline was crossed. In addition, HURDAT2 does not flag landfalls for the entire Atlantic Basin prior to 1946 – only for the United States (with often unmarked landfalls). The 1960s through the early 1980s also feature a lack of landfall indicators. This means I had to sort the database by hand, comparing coordinates to coastlines to see if any storms made landfall. If a storm did make landfall, I add a precise landfall coordinate: the real latitude and longitude of where the coast was actually crossed, at the time it was, rather than the nearest six-hourly checkpoint. This is a very slow process, and I’m still actively catching errors, since covering a database this large by hand almost certainly presents errors. The base data used on this website is official, and is tweaked as accurately as possible on the coastal scale to work with an interactive map setting.

Storms are labeled with their official name wherever one exists. Atlantic storms weren't formally named until 1950, so anything before that is labeled by its numerical order within that season — "Hurricane Three (1926)," for example — unless it's one of a couple dozen storms significant enough in the historical record to be widely known by an unofficial nickname (the 1926 Great Miami Hurricane, the 1938 New England Hurricane, and similar), in which case that name is used instead. Where the National Hurricane Center's official Tropical Cyclone Reports document a storm's fatalities or damage total, that figure is included; where it isn't documented, the site says so plainly rather than guessing.

HURDAT2 isn’t meant to be exact. It’s meant to display a storm’s track on a larger scale for archive purposes. Not being exact isn’t good enough for this database, which is meant to display landfalls on a smaller, zoomed-in scale. Oftentimes HURDAT marked landfalls that were either just off the coast, or just inland. This created a visually messy map with landfalls that didn't perfectly line up with the coastline. The manual work is slow because getting it right takes actually looking at each storm individually, not just accepting whatever the nearest data point to a coastline says. Make no mistake however, the HURDAT2 database is often cited as the cleanest, most accurate and extensive hurricane database of its kind.