A frequency benchmark matched most of the mean warm-month continuation rise in six ocean boxes in 1982-2023

We tested whether warm ocean months continued more often after their frequency rose. We used monthly regional sea surface temperature in six fixed boxes in ERSSTv5 and HadISST1. Warm months exceeded the calendar-month 90th percentile for 1982-2011. We compared 1982-2002 with 2003-2023. Mean continuation rose by 23.25 percentage points in ERSST and 15.14 in HadISST. A calendar-month independence benchmark rose by 22.35 and 18.75 points. The excess over that benchmark changed by +0.90 and -3.61 points. Thus the point changes gave no common increase beyond the frequency benchmark. Uncertainty was large. In 2511 of 10000 ERSST three-year block draws, at least one region had no early warm first month. The mean ratio was then undefined. We report these failures and bound their possible effects. This is a descriptive check of monthly regional extremes. It does not test daily marine heatwaves, prove constant ocean memory, or identify a physical cause. Threshold choice, short records, source reconstruction and shared observations limit the result. Code, monthly values, source hashes, counts, uncertainty draws and a figure are included.

sea surface temperature · warm months · continuation · frequency benchmark · ocean extremes · sparse denominators