The Atlantic total-to-thermal squared-change ratio fell from one season to one year in NOAA upper-ocean fields for 2005-2025

We tested local squared steric changes at three lags in NOAA upper-700-m fields for 2005-2025. All lags used 76 start seasons in 2005-2023. We removed fixed seasonal means and cell trends. The Atlantic total-to-thermal mean squared-change ratio was 0.752 at one season and 0.709 at one year. The paired log contrast was -0.0587, with conditional 95% three-year block limits of -0.0925 to -0.0244. Its sign held with two- and four-year blocks, without trend removal, and between 60 S and 60 N. The Pacific contrast was uncertain. The secondary Indian contrast had the opposite sign. A post-score identity showed that the cross term with total-minus-thermal changes accounted for most of the point contrast. This is not an independent salinity measurement or a causal test. One mapped product, 19 start years, overlapping pairs, and uncertainty conditional on fitted residuals limit inference. Code, basin aggregates, source hashes, all scores, draws, audits and a figure are included.

steric sea level · squared change · time scale · salinity compensation · NOAA · block bootstrap