Major Tropical Cyclone Intensity Trends Revisited: A re-analysis of Kossin et al. (2020) with the ADT-HURSAT record extended to 2024

Kossin et al. (2020, PNAS) reported a statistically significant global increase, 1979–2017, in the proportion of tropical cyclone (TC) intensity estimates at major-hurricane strength (Saffir–Simpson Category 3–5), a result cited by the IPCC Sixth Assessment Report. We identify six weaknesses in that analysis and repeat it with the updated ADT-HURSAT record (ADT 9.0 applied to HURSAT v07b, 4,852 storms, 1978–2024) using inference suited to clustered data. Four findings stand out. (1) Intensity estimates are clustered by storm and by year: the variance of the early-versus-late comparison is 5.6 times the binomial value, not the 3 assumed, so correct confidence intervals are about 37% wider than published; with the correct variance the 1979–2017 early-versus-late global change is not significant (p = 0.10). (2) The 1979–2017 annual trend is significant only when the record starts in 1979–1981; starting in 1982 or 1985 gives p ≈ 0.06. (3) With the record extended to 2024, the annual trend is robust: +4.9% per decade globally (p < 0.001) and +4.4% per decade excluding the North Atlantic (p < 0.001), significant for start years from 1979 to at least 1987; about one tenth of it is attributable to a drift in satellite viewing angle. (4) The increase is in the time hurricanes spend at major intensity (+2.1 h per decade per major hurricane, p = 0.015), not in the share of hurricanes that become major (+2.3% per decade, p = 0.16). The central conclusion of Kossin et al. is therefore supported by the longer record, but the original 1979–2017 evidence was weaker than reported, and the result is commonly misread as an increase in the chance that a hurricane becomes a major hurricane, which the data do not show.