Monthly mean removal left uncertain changes in next-day high-water persistence at six US gauges in 1960-1989 and 1996-2025

We tested whether a high-water day became more likely to be followed by another high-water day at six selected US tide gauges. We compared complete accepted years within 1960-1989 and 1996-2025. High water was above the early 95th percentile for each calendar month. The main test used daily hourly maxima minus concurrent monthly mean sea level. The change in next-day high-water probability ranged from -7.2 to 4.5 percentage points. All six pointwise and six-site simultaneous 95% five-year block intervals included zero. For raw water levels, the point changes ranged from 7.1 to 26.4 percentage points. The bootstrap re-estimated thresholds in each draw. Missing records, selected years, period choice, tidal effects, and source processing limit inference. Monthly mean removal can remove part of a storm signal. The results do not establish unchanged event dynamics or identify a physical cause. The package supplies data extracts, raw file hashes, code, saved primary draws, and count checks.

coastal water levels · event persistence · NOAA CO-OPS · mean sea level · block bootstrap · null result