12,273 Publications · 71,443 Citations

Kossin (2018) reported a 10% decrease in the annual-mean translation speed of tropical cyclones in 1949-2016, from 6-hourly NHC and JTWC best tracks. We repeated the method on IBTrACS v04r01 (USA-agency positions, 223,676 six-hour segments). Then we tested the frozen 1949-2016 trend on eight new seasons with final best tracks (2017-2024). We wrote the plan before the test calculations. The reproduction gave -8.6% globally (published -10%). The regional values were near the published values, for example -21% over land in the western North Pacific. In 2017-2024 the raw global mean (18.13 km/h) was above the extended trend (17.37 km/h) in all 8 seasons. The residual, +0.77 km/h, was outside the 95% range expected with AR(1) noise (p = 0.03). The extended trend had about twice the squared error of the 1949-2016 mean and the 2007-2016 mean. Variants without extratropical or weak points gave p = 0.06-0.11. Speed increases strongly toward the poles, and the share of segments at 30 degrees or more fell from 13.0% (1949-1969) to 9.4% (2007-2016), then rose to 12.5% in 2017-2024. Within region x 5-degree latitude x month cells, the 1949-2016 decrease was -3.9% (p = 0.007), and 2017-2024 was on its extended line (+0.04 km/h, p = 0.91). The latitude band alone gave the same result. Thus about half of the raw slowdown and all of the recent departure follow the latitude mix of recorded segments. A smaller decrease within cells continued, but the frozen line did not clearly beat the 2007-2016 mean. Published over-land slowdowns in the western North Pacific and Australian region did not clearly change (p = 0.54 and 0.55). Limitations: only 8 test seasons, with the raw excess mainly in 2018, 2020 and 2021; we cannot tell whether the latitude shift is physical or from recording practice; and adding the wind category reduced the within-cell decrease to -2.0% (p = 0.17).