Every hurricane, typhoon and cyclone in NOAA's global best track archive, 13,186 storms in all, drawn by strength. Where they form, where they hit, the strongest on record, and the hard question: are they getting stronger? Since 1980, more of them reach Category 4 or 5, and they peak further from the equator. Before satellites, the record misses too many storms to say much more.
Since 1851, the first season in the American record, ships, coastal stations and, later, aircraft and satellites have tracked the great storms of the tropics. NOAA's International Best Track Archive for Climate Stewardship, IBTrACS, gathers the tracks of every agency into one global record: the position, wind and pressure of every tropical cyclone, every six hours, from the Atlantic hurricanes of the 1850s to this year's typhoons. That is 352,746 positions for 13,289 storms.
Here is all of it on one map, coloured on the Saffir-Simpson scale, from tropical depressions in pale blue to Category 5 in deep plum. Below the map I count where they form, where they come ashore and which were the strongest, and then look honestly at whether they are getting stronger. The honest part matters: before weather satellites, many storms that stayed out at sea were never seen, so the early record is thin, and its winds were often guessed or not recorded at all.
Each line is a storm, coloured by its strength at each point. Play the years, pick a period, filter by basin or strength, or search for a storm by name. Click a track for the storm's name, year, peak wind and the countries it hit. Landfalls, where storms form, and the base map and labels are under Map layers.
Tracks: NOAA NCEI, IBTrACS v04r01, all basins, 6-hourly positions, main and provisional tracks (the 2025 and 2026 storms are partly provisional). Colour: Saffir-Simpson category from the 1-minute sustained wind of the US agencies (NHC, CPHC, JTWC) where given, else the responsible agency's wind (often a 10-minute average, which reads lower); grey where no wind was recorded. Tracks are simplified for the web. Landfalls: where a track crosses from sea onto land in Natural Earth's 1:10m countries, at tropical storm strength or more. Base map tiles © Esri.
Since 1980, when geostationary satellites watched the whole tropics, an average of 89.1 storms a year have reached tropical storm strength (winds of 34 knots, 39 mph, or more), and 47.3 have become hurricanes, typhoons or severe cyclones (64 knots, 74 mph). The quietest season was 1991 with 73, the busiest 2005 with 106. The western North Pacific, the typhoon basin, has 29% of them; the North Atlantic, the most studied, only 15%. Over those 46 seasons the number has not clearly changed: the fitted trend is +1.1 storms a decade, with a 95% interval of -0.8 to +2.9.
Storms per season, by basin where they formed
Hover a bar. The shaded years before 1970 are before satellite coverage: the record there misses storms, and outside the Atlantic it has no winds at all before the 1940s.
Look at the long record and the storms seem to multiply. In the North Atlantic, the one basin with winds back to 1851, there were 7.5 tropical storms a season from 1851 to 1900, and 13.3 from 1976 to 2025. But this is mostly a story about seeing, not about storms. In the nineteenth century a storm was recorded only if it hit land or a ship ran into it. Storms that stayed far out at sea are exactly the ones that were missed: only 12% of the early Atlantic storms never came within 500 km of land, against 23% in the last fifty years. For hurricanes, which are harder to miss, the rise is smaller: 5.2 a season then, 6.8 now. Elsewhere it is starker: the archive has tracks for the Bay of Bengal from the 1850s and for the South Indian Ocean from 1851, but no winds until the 1957s and 1972s. Only 15% of the systems of the 1900s have any wind at all, against 99% in the 1990s.
Polar-orbiting weather satellites were declared operational in 1965, and it took several years for warning centres to learn to read storms from them; the Dvorak technique, which still plays a large role in estimating a storm's strength from satellite images, was developed in the 1970s. So I treat the years before 1970 as a different, patchier record, and make every comparison of intensity within 1980 to 2025. Even then, most storms are measured from satellite images rather than by aircraft, and the methods have kept improving, which can itself create trends.
The density layer on the map shows where the 4,097 storms that reached tropical storm strength from 1980 to 2025 began. They form over warm water in two belts, mostly between 5 and 20 degrees from the equator: 84% of them start there, and only 3.8% within 5 degrees of the equator, where the Earth's spin is too weak to wind them up. 70% form north of the equator. The busiest nursery on Earth is the sea off southern Mexico, where the 2.5 degree square around 13.8°N, 101.2°W has spawned 7.6 storms a decade; the western Pacific has the largest area of busy water. The South Atlantic almost never makes one: 3 in 46 seasons.
When each basin's storms form
Storms reaching tropical storm strength, 1980 to 2025, by basin and the month they formed. Hover a cell.
The seasons run like clockwork. Atlantic storms peak in September, eastern Pacific ones in August, typhoons in August, South Indian and South Pacific cyclones in January to March. The North Indian Ocean has two seasons, before and after the monsoon, with peaks in May and November, which is why the Bay of Bengal is so often hit in late autumn.
I counted a landfall wherever a track crosses from sea onto land at tropical storm strength or more. From 1980 to 2025, 40% of all tropical storms made landfall somewhere, and 35% of hurricanes and typhoons came ashore at hurricane strength, in 94 countries and territories. China is hit by the most storms: 263 in 46 seasons, about 5.7 a year. But the Philippines takes the hardest blows: 129 storms came ashore there at typhoon strength and 67 at Category 3 or more, both the most of any country, because typhoons reach its eastern islands at their peak, straight off the warm open Pacific. The United States had 158 landfalls, 70 at hurricane strength.
Countries hit most often, 1980 to 2025
Storms making landfall at least once in the country, by strength at landfall. A storm counts once per country. Click a row to see its landfalls on the map.
Down at the level of coasts, the most-hit place on Earth is Guangdong, China's southern coast around Hong Kong and Macau, with 118 landfalls since 1980, followed by Western Australia (96), Fujian (65), Queensland (61) and Florida (58). Counting only landfalls at Category 3 or more, the leaders are Western Australia (19), Okinawa Prefecture (16) and Cagayan, at the northern tip of Luzon (15).
The most-hit coasts: states, provinces and prefectures
Storms making landfall at tropical storm strength or more, 1980 to 2025. Click a row to zoom the map there.
A caution on strength at landfall: I interpolate the wind between six-hourly positions, so a storm that peaks right at the coast can read a category lower than its official landfall. Andrew in 1992, officially a Category 5 landfall in Florida, reads here as Category 4. Over the whole record, with its gaps, the United States has the most counted landfalls (565), simply because its coast has been watched longest.
The highest 1-minute wind in the archive is 185 knots (215 mph, 345 km/h), shared by Hurricane Patricia off Mexico in 2015 and Typhoon Nancy in 1961. Patricia's is usually given as the highest sustained wind ever recorded. Nancy's comes from the early years of aircraft reconnaissance in the Pacific, and 11 of the top 20 are typhoons of the 1950s and 1960s, whose winds were estimated in ways that are hard to compare with today's. The lowest pressure belongs to Typhoon Tip in 1979, 870 hectopascals. 324 storms reached Category 5, 209 of them typhoons in the western Pacific.
The strongest storms on record
By peak 1-minute sustained wind (US agencies), then lowest pressure. Click a row to see the track.
The strongest landfalls, by my interpolated winds, are Haiyan on Leyte in the Philippines in 2013 (166 knots), Meranti on Batanes in 2016 and Dorian on the Abaco Islands of the Bahamas in 2019.
If storms are getting stronger, a larger share of them should reach the top of the scale. Here I use only the 1-minute winds of the US agencies, so every storm is measured the same way, and only storms that reached tropical storm strength. In the 1980s, 12.5% of them reached Category 4 or 5. In the 1990s it was 20.6%, in the 2000s 20.2%, in the 2010s 22.8%, and so far in the 2020s 18.5%. Split the satellite era in half, and 16.8% of storms reached Category 4 or more in 1980 to 2002, against 21.2% in 2003 to 2025: a difference that chance would rarely produce (p under 0.001). The number of Category 4 and 5 storms a year has risen by 2.0 a decade (95% interval 0.8 to 3.1).
Share of storms reaching Category 3+ and Category 4+, by decade
Storms reaching 34 knots on the US agencies' 1-minute wind. Pale bars: fewer than 80% of the decade's systems have such a wind, or before 1970. The 2020s are 2020 to 2025. Hover a bar.
For Category 3 and above the picture is weaker: 26.7% against 28.4% (p 0.21), and the year-by-year trend, +0.7 percentage points a decade, could be zero (95% interval -0.6 to +2.1). Most of the rise happened between the 1980s and the 1990s, the years when satellite methods were maturing, and the share has moved little since. By basin, the clearest rises in Category 4+ are in the South Pacific (10% to 21%), the western North Pacific (23% to 30%) and the South Indian Ocean (15% to 21%); the North Atlantic (12% to 14%) and the eastern Pacific show no clear change. The total energy of each season, measured as accumulated cyclone energy, has no clear trend at all (-7.7 a decade, 95% interval -45.1 to +25.0).
Somewhat, in the one way the record can measure fairly. Since 1980 the number of storms has not changed, but a larger share of them reaches Category 4 or 5: about one in 6 in the first half of the satellite era, one in 5 in the second. Wikipedia, summarising a 2020 study by Kossin and colleagues, reports a global increase from 1979 to 2017 in the proportion of storms reaching Category 3 or more. In this raw best-track record the rise is clear for Category 4 and above but not for Category 3 and above, and I cannot rule out that some of it comes from better measurement. Nothing before 1970 can be compared fairly. The record does not show more storms, or that every season is getting worse.
A second, subtler change is where storms reach their peak. In the Northern Hemisphere, the average latitude of each storm's lifetime peak wind has moved poleward by 45 km a decade since 1980 (95% interval 17 to 72 km), in the Southern Hemisphere by 21 km a decade (3 to 40 km). Across both, that is about 182 km over 46 seasons. It is driven by the western Pacific, where typhoons now peak 65 km a decade further north, closer to Japan, Korea and China. In the North Atlantic the average has moved the other way, though not clearly (-53 km a decade, interval -103 to 4).
Average latitude of each storm's peak, by season
Storms reaching 34 knots, distance from the equator of the point of highest 1-minute wind, mean per season, with straight-line trends. Hover a point.
The poleward shift by basin
Trend in the latitude of peak intensity, km a decade, 1980 to 2025, with 95% bootstrap intervals. Positive is towards the pole.
The number of storms has not changed since 1980. Where they peak, and how many reach the top of the scale, has.
Data. IBTrACS version 4, revision v04r01, from NOAA's National Centers for Environmental Information, reduced to the synoptic six-hourly positions (00, 06, 12 and 18 UTC): 352,746 positions of 13,289 storms, keeping the main tracks and the provisional tracks of recent storms and leaving out the short "spur" tracks that IBTrACS uses for alternative positions. Seasons follow IBTrACS: a Southern Hemisphere season is named after the year it ends. Analyses use the complete seasons 1851 to 2025; 2026 is on the map only.
Strength. The Saffir-Simpson scale is defined on the 1-minute sustained wind, so intensity is the wind of the US agencies (the National Hurricane Center, the Central Pacific Hurricane Center and the Joint Typhoon Warning Center), which IBTrACS gives as USA_WIND. Where it is missing, the map uses the wind of the agency responsible for the basin; many of those are 10-minute averages, which read lower. Shares of Category 3+ and 4+, the latitude of peak intensity and accumulated cyclone energy use the US 1-minute wind only. Storms are counted in the basin where they formed.
Landfall. Each six-hourly segment is a straight line, intersected with every land polygon of Natural Earth's 1:10m countries; a landfall is where a segment enters land from the sea while the storm was not extratropical, with the wind interpolated along the segment. Crossings of land borders do not count. The state or province is the Natural Earth unit that contains the landfall. Straight segments can cut corners or miss small islands, and interpolated winds are approximate.
Trends. Trends are least-squares lines through the seasonal values, with 95% intervals from 4,000 bootstrap resamples of seasons; the comparison of halves uses a two-proportion test. Nothing here separates natural variability, such as El Niño or the Atlantic's multidecadal swings, from long-term change, and nothing here attributes anything to climate change. Before about 1970 the record misses many storms, especially weak and short-lived ones out at sea, and outside the Atlantic it has no winds before the 1940s or later, so long-run trends in the record say as much about observation as about storms.