Past regional error quantiles raised high tide flood interval scores against pooled quantiles at 83 US gauges in 2022-2025

We tested whether past errors from the same coastal region improve later annual flood day ranges. We used a verified NOAA evidence table. Each test year used only earlier rows from 2019 onward. The test has 307 station-years at 83 US gauges in 2022-2025. We estimated signed error quantiles by region or across all gauges. The regional model raised the equal-year mean 68%-target interval score by 1.59 days. Lower scores are better. Scores rose in three of four years. Conditional station-and-year bootstrap limits for the score change were -0.17 to +3.53 days. Pooled mean score was 16.05 days for regional limits and 14.47 days for global limits. Coverage was 55.4% and 58.0%. Regional error structure in the full record did not give this simple correction a later-year gain. Only four test years, sparse regional samples, data revisions and shared coastal conditions limit the result. The test does not rule out other regional models or identify a physical cause. Code, all limits, source hashes, a separate arithmetic check and a figure are included.

high tide flooding · regional errors · interval score · chronological validation · NOAA · null result