Why storm by storm
A count is a sum of individual decisions
Every trend in the climatology paper is an aggregate — a basin total, a global proportion, a decade mean. But a "global Category 5 count" is not a measurement; it is a sum of classification decisions, each made for one storm, by one agency, in one era, from one observing system. The aggregate inherits every one of those decisions. This page takes the record apart into its individual storms and asks two questions: what kind of object is each storm, and how well do we actually know its peak intensity?
The storm families
k-means over five per-storm variables (LMI, lifetime, latitude of peak, translation speed, peak 24-h intensification), 1979–2025. Families ordered by mean intensity; every storm belongs to exactly one.
Correlations
What travels together, storm by storm
The four voices
Four independent estimates of every storm's peak
For every storm the ledger carries up to four lifetime-maximum intensities: the merged best track (JTWC + NHC — the record every trend paper uses); the ADT-HURSAT satellite record — the Advanced Dvorak algorithm reading the raw HURSAT imagery, windowed to the storm's own track and quantile-mapped to the aircraft-reconnaissance Atlantic (1995–2024, n = ) so its known low bias is removed before any comparison — the raw satellite bias against aircraft truth at major intensity is just kt, with a per-storm residual of ± kt; the Xu et al. (2024) ERA5 random forest, matched storm-by-storm by time and position; and the strongest other IBTrACS agency (JMA, CMA, HKO, IMD, Réunion, BOM, Nadi, Wellington), converted toward 1-minute winds. Where the voices agree the storm is well measured; where they diverge, somebody's book entry is wrong.
Category 4–5 storms per decade: the count and its uncertainty
Global storms with LMI ≥ 113 kt. Bars: best track and bias-adjusted satellite. Whiskers: the agreement band — storms both records call Cat 4–5 (floor) to storms either calls Cat 4–5 (ceiling).
How well each decade's storms are measured
Share of storms by agreement grade: great (voices within 12 kt, 3+ sources) · good (≤22 kt) · bad (≤35 kt) · worst (>35 kt or ≤1 source).
The missed
Candidate majors the best track never booked
The overcalled
And the book entries that run high
The super-optimized track
Leaving the USA-only record: consensus, provenance, and the trend under uncertainty
The merged best track is deliberately USA-centric — JTWC and NHC throughout, Neumann filling the early south. But IBTrACS carries every other agency's book (JMA, CMA, HKO, KMA, IMD, Réunion, and the SPEArTC constituents BOM/Nadi/Wellington), and the audit adds the adjusted satellite record and the ERA5 reconstruction. A super-optimized track takes, for every storm, the median of every available ~1-minute-equivalent voice — and, more importantly, keeps the spread, so the uncertainty can be carried into every downstream number rather than discarded at the bookkeeping step.
Does the trend depend on whose book you read?
Landfalls, storm by storm
How firm is a landfall category?
Method
Choices, conversions, and what a candidate is not
References
- Hoarau, K., J. Lander, R. De Guzman & R. Guard (2012). Intense tropical cyclone activities in the northern Indian Ocean. Int. J. Climatol. 32 — the Dvorak reanalysis showing the 1980s NIO undercount this audit generalizes.
- Landsea, C. W., Harper, B. A., Hoarau, K. & Knaff, J. A. (2006). Can we detect trends in extreme tropical cyclones? Science 313, 452–454.
- Knapp, K. R., Olander, T. L., Velden, C. S., Gahtan, J. & Schreck, C. J. (2025). ADT-HURSAT v01 (NCEI Accession 0307249). doi:10.25921/n6va-0b18
- Kossin, J. P., Knapp, K. R., Olander, T. L. & Velden, C. S. (2020). Global increase in major tropical cyclone exceedance probability. PNAS 117, 11975–11980.
- Xu, Z., et al. (2024). Global tropical cyclone size and intensity reconstruction dataset for 1959–2022. ESSD 16, 5753.
- Knapp, K. R. & Kruk, M. C. (2010). Quantifying interagency differences in tropical cyclone best-track wind speed estimates. Mon. Wea. Rev. 138, 1459–1473. (the wind-averaging conversions)
- Weinkle, J., Maue, R. & Pielke, R. Jr. (2012). Historical global tropical cyclone landfalls. J. Climate 25, 4729–4735 — the landfall definition audited here, on its own page.
- Maue, R. N. (2026). Global Climatology of Tropical Cyclone Power Dissipation — the aggregate record this ledger decomposes.