Upper midpoint tails had more shared dates in the mean of six Atlantic gauge pairs in 2015-2024

We compared shared low and high water-level dates at Boston, The Battery, Sewells Point, and Charleston in 2015-2024. We reused 14,334 station-days and NOAA tide predictions from a signed campaign publication. Separate calendar-month thresholds defined each tail of a tide-corrected daily midpoint. Across six pairs, the mean joint rate was 0.869% for the lower 5% tail and 1.125% for the upper 5% tail. The lower-minus-upper difference was -0.256 percentage points. Its conditional 95% three-year block interval was -0.409 to -0.025 points after threshold refitting. A five-year block interval included zero. Four pair estimates were negative and two were positive. Uncorrected midpoints and other thresholds retained a negative mean point estimate. The result tests finite-tail spatial alignment in selected records. Daily extrema, ten years, shared sites, tide-model limits, and block sensitivity prevent a causal storm or flood-risk claim. The package gives inputs, source hashes, code, all draws, a figure, and independent arithmetic and source audits.

sea level · spatial extremes · tail asymmetry · NOAA CO-OPS · daily midpoint · block bootstrap