Daily sampling missed 28.5% of six-hour rapid-intensification storm labels in Atlantic HURDAT2 in 1982-2023

We tested loss of rapid-intensification storm labels when six-hour Atlantic HURDAT2 starts were reduced to fixed daily clocks. A valid 24-hour window had five exact six-hour fixes with tropical storm or hurricane status. The wind gain threshold was 30 kt. Among 488 eligible storms in 1982-2023, 165 had rapid intensification. The mean missed fraction across four daily clocks was 28.5%, with a conditional 95% year-bootstrap interval of 24.4% to 32.8%. Two twelve-hour grids missed 9.1% and 12.1%. Daily loss was 25.1% at 25 kt and 32.6% at 35 kt. Requiring valid starts at all daily phases gave 27.8% loss. A secondary era test found lower loss in 2003-2023 than in 1982-2002, but did not test its cause. Eligibility was held fixed with the full record. Thus, this is a sampling experiment, not a daily-only processing system or an estimate of missed true events. Wind error, land exposure, source revisions and storm mix limit inference. The package provides code, source hashes, aggregate counts, a figure and an independent count audit.

rapid intensification · Atlantic hurricanes · HURDAT2 · temporal sampling · measurement bias · storm labels