Two ENSO slopes gave no clear winter sea-level reconstruction gain at two Pacific gauges in 1991-2025

We tested whether separate warm and cold ENSO slopes improve winter mean sea-level estimates at San Francisco and Seattle. Each target winter used coefficients fit to earlier winters only. The target winter ENSO value was supplied, so this is conditional reconstruction, not a seasonal forecast. In 34 common complete test winters in 1991-2025, the equal-site mean squared error reduction was 5.27%. Its conditional 95% five-year block interval was -13.10% to +19.60%. Error rose by 5.45% at San Francisco and fell by 16.00% at Seattle. All planned sensitivity intervals included zero. Thus, the two-slope model gave no clear gain in this benchmark. This extends a prior fitted-slope study on the same sources; it is not an untouched validation set. Two selected gauges, few strong events, revised records, a fixed SST reference, unmeasured weather, and uncertainty conditional on saved losses limit inference. The package includes exact NOAA inputs and hashes, code, predictions, all bootstrap draws, a figure, and independent numerical checks.

ENSO · sea level · nonlinear response · conditional reconstruction · chronological test · null result · tide gauge