Tom Knutson's Global Warming and Hurricanes — the field's standard living review for two decades — preserved verbatim now that its author has retired, set beside the IPCC's calibrated statements and the measurements on this site: claim by claim, in a running audit that will track AR7 as it arrives.
Knutson, T. R. — Global Warming and
Hurricanes: An Overview of Current Research Results
GFDL/NOAA · maintained
2007–2025 · archived here verbatim 2026-09-18 · original:
gfdl.noaa.gov/global-warming-and-hurricanes/
For two decades this page was where the field — and honest journalists — went for the state of the science, updated as papers landed, written with the calibrated care its subject deserves. With Tom Knutson's retirement it stopped being a living document. The copy preserved here is byte-faithful to the final state (asset filenames mechanically cleaned for static serving; article text untouched), committed to version control so that the last version remains permanently citable and any future changes to the original are diffable. Nothing in that directory will ever be edited; commentary lives on this page instead.
The archived summary makes two kinds of claims — projections for a 2 °C-warmer world, and detection/attribution statements about changes already observed — and it is scrupulous about the difference. The audit below honors that structure. This is not an adversarial exercise: much of what this site measures confirms the review's careful formulations, sometimes against louder versions of the same claims; where our evidence refines or complicates a statement, that is noted as what it is — new information the living document never got the chance to absorb.
| Claim | GFDL review (Knutson, final) | IPCC AR6 WG1 Ch. 11 | climatlas evidence |
|---|---|---|---|
| Observed rise in major-TC proportion | Increase in the fraction of Cat 3+ intensity fixes 1979–2017 (citing Kossin et al.); notes the increase in proportion was assessed as mostly due to a global decrease in Cat 1–5 fixes with a slight, non-significant increase in Cat 3–5 fixes. | likely that the global proportion of Category 3–5 TC instances has increased over the past four decades. | Reproduced and extended. The corrected Kossin result (+5%/dec [0.4, 11]) replicates on ADT-HURSAT v01 and holds at +5.4%/dec through 2024 (paper §7). Our decomposition supports Knutson's denominator nuance in the best track (its hurricane-fix denominator shrinks) but finds the homogeneous record's trend carried by genuine duration-at-major (§8). |
| Long-term TC frequency trends | No detectable century-scale increase in Atlantic hurricane frequency after accounting for observing changes; global frequency trends not established. | low confidence in most reported long-term (multidecadal to centennial) trends in TC frequency- or intensity-based metrics. | Consistent. Global PDI/ACE trendless 1960–2025; hurricane landfalls trendless 1970–2025 under both definitions (landfalls); the 24-month global ACE swung from the 13th to the 91st percentile inside 15 years — variability, not drift (§12). |
| Projected intensity increase | +1 to +10% in mean intensity for 2 °C warming; larger percentage increase in "destructive potential per storm, assuming no reduction in storm size". | Proportion of intense TCs and peak winds increase with warming (high confidence for the proportion statement in projections). | Refinement offered on the caveat GFDL itself flagged. The "assuming no reduction in storm size" proviso is exactly where our measurement bears: dissipation scales as V≈2.1 across the observed population because area contracts with intensity — the destructive-potential amplification is roughly half the cubic arithmetic, conditional on that structure transferring (§9). |
| Projected TC rain rates | +14% (+6 to +22%) globally within ~100 km of center for 2 °C. | Peak TC rain rates increase at least at ~7%/°C (medium–high confidence). | Not assessed here. This site carries no precipitation analysis; the claim stands on its own literature. |
| Projected global TC frequency | Most studies project a decrease or little change. | Global TC frequency projected to decrease or remain unchanged (medium confidence). | Observationally consistent to date: no rise in global storm counts in 66 years of the merged record. |
| Poleward migration (WNP) | Detectable poleward shift of latitude of maximum intensity in the NW Pacific, unusual relative to natural variability; contribution of anthropogenic forcing "not confidently estimated". | Poleward migration assessed with the same regional framing (medium confidence on attribution). | Present in the satellite era, absent over 66 years. Global median LMI latitude: +0.24°/dec since 1982 (p=.04); flat 1960–2025. Landfall latitude flat. Our anatomy assigns the exposure change to track character, and majors themselves peak equatorward of minors (majors paper) — migration is a composition property, worth carrying into AR7 framing. |
| Rapid intensification increase | Observed increase in RI probability (1982–2017; Atlantic, NW Pacific, global), highly unusual vs one model's internal variability and consistent in sign with forced response. | Increased intensification rates assessed as part of the observed-change evidence base. | Strongly corroborated — our cleanest two-record agreement: RI-fix share +9.5 and +8.6%/dec since 1990 in best track and ADT-HURSAT respectively (§11), with a size gate (compact storms RI at >2× the rate) and a shortening runway to peak among majors. |
| Translation-speed slowdown | Treated cautiously in the review's later updates, noting the data-era critiques. | Not a headline AR6 claim. | Caution vindicated: absent at storm level — flat 1966–2025, +1.8%/dec since 1990 (§8). |
| SST as the driver of PDI | The review moved with Emanuel from SST correlation to potential-intensity framing after 2007. | PI-based framing throughout. | The move was right. Emanuel's 2005 SST–PDI correlations have largely dissolved on the extended record — as his own 2007 revision first indicated for the WNP — while measured peak-season (Jul–Oct) PI rose over 1979–2025 at about the rate he expected — concentrated before the mid-2000s, flat since 1996 outside the West Pacific — with utilization not rising (§13). |
The AR7 cycle is underway (WG1 assessment expected late this decade). As drafts and the report arrive, this page will hold each tropical-cyclone statement against the archived GFDL baseline and the measurements here — the same three-column discipline, with the calibrated-language changes between AR6 and AR7 made explicit. Items this site is positioned to inform: the exceedance/proportion statement (now reproduced independently, with a duration decomposition); the structural term in destructive-potential language; the landfall record (trendless through 2025 under the original WMP12 definition); RI (the robust one); and the storm-level uncertainty quantification of the ledger — count bands, not point counts, belong in assessment prose.