Cooling remained at day 30 in a fixed sample of 491 Atlantic storm-buoy encounters in 1983-2024

We tested whether changes in event membership can explain late cooling in a prior Atlantic buoy study. We used its saved event and control curves and held 491 storm-buoy encounters fixed across five windows through day 60. The mean excess temperature on days 28-32 was -0.209 degrees C. Its 95% year-bootstrap interval was -0.357 to -0.057 degrees C. Fixed membership changed the day-30 estimate by +0.021 degrees C. Requiring all window days and using storm or buoy clusters retained negative day-30 intervals. The days 56-60 minus days 28-32 change was +0.021 degrees C, with interval -0.114 to +0.153, so further late recovery was uncertain. The shelf-minus-Gulf contrast retained a negative interval. This is a robustness check on reused data, not independent physical evidence. Saved controls, sensor gaps, selection and incomplete dependence models limit inference. The package contains exact derived inputs, hashes, code, event evidence, draws and checks.

cold wake · sea surface temperature · fixed event sample · missing data · year bootstrap · Atlantic buoys