Winter and summer lead signs remained opposite after predicted tide midpoint removal at two US gauges in 2015-2024

We tested whether mean filters and tide terms explain a known seasonal change in one-day water-level timing between Sewells Point and Charleston. Smooth mean removal and fixed lunar-cycle projection left the opposite winter and summer signs in 1960-2024. An added check used full NOAA hourly tide predictions for 2015-2024. After predicted daily midpoint subtraction, the north-lead minus south-lead correlation was -0.105 in winter and +0.300 in summer. The paired summer-minus-winter contrast was 0.404, with a conditional 95% two-year block interval of 0.288 to 0.540. The change from the uncorrected contrast was -0.0149, with interval -0.0329 to +0.0037. One- and three-year blocks also retained opposite signs. Thus predicted tide midpoint subtraction did not account for the reversal in this retained sample. This is a filter sensitivity extension on overlapping data, not independent confirmation. Daily extrema subtraction is not matched-hour tide removal. Short records, selected sites, missing days and conditional intervals limit inference. The result does not identify waves, moving storms, causal paths or forecast skill. Code, accepted observations, prediction extrema, source hashes, uncertainty draws and figures are supplied.

coastal sea level · seasonal lead reversal · NOAA CO-OPS · tide midpoint · lag correlation · filter sensitivity