ENSO gave no clear six-month sea-level forecast gain beyond local history at four West Coast gauges in 1990-2024

We tested whether Nino 3.4 adds monthly sea-level information six months later after three local levels are known. We used fixed PSMSL and NOAA snapshots for San Diego, San Francisco, Astoria and Seattle. The test has 1,611 site-month targets in 1990-2024. Fits used training targets before each origin year. The equal-site mean reduction in squared error was 0.12%, with a conditional 95% five-year block interval of -2.22% to 2.17%. This gives no clear added six-month gain. The same target months gave 1.90% at a one-month label lead. The paired six-minus-one contrast was -1.79 percentage points, with an interval of -3.19 to -0.54. This secondary contrast compares error ratios with different baseline errors. It does not identify a physical loss of ENSO influence. Three- and seven-year blocks, quadratic time, recent training and selected-year exclusions left the six-month mean gain uncertain. Four selected gauges, revised data, gaps, few climate episodes and uncertainty conditional on fitted histories limit inference. The package includes code, exact used source files, source hashes, every forecast, a separate numerical audit, an offline repeat and a figure.

sea level · ENSO · six-month lead · local memory · chronological test · null result