Warm and cold monthly SST tails gave no clear global persistence difference after trend removal in ERSSTv5 for 1982-2025

We tested whether extreme warm and cold ocean months had different chances of staying extreme next month. We used NOAA ERSSTv5 for 1982-2025 at 7,877 grid cells from 60 S to 60 N, with minimum sea surface temperature above -1 C. We removed a separate linear year trend for each cell and calendar month. The upper and lower tails each selected five of 44 years per month. The area-weighted warm-tail probability was 56.33% and the cold-tail probability was 56.20%. Their difference was 0.13 percentage points. Its 95% circular three-year block interval was [-3.99, 4.18] points with fixed fits and [-1.85, 3.42] points when trends and thresholds were refitted. Both intervals include zero. Thus, this record gives no clear global warm-versus-cold difference for this monthly metric. It does not establish equal physical recovery rates, daily heatwave durations, regional symmetry, or absence of a trend. One coarse reconstruction, a short climate record, trend choice, and conditional resampling limit inference. Original code, source hashes, annual aggregate evidence, all resampling weights and draws, two figures, and an independent arithmetic audit support reproduction. Raw grids remain external.

sea surface temperature · ocean extremes · warm and cold tails · monthly persistence · ERSSTv5 · block bootstrap · null result