Seasonal ENSO terms gave no clear monthly sea-level reconstruction gain at two Pacific gauges in 1990-2025

We tested whether seasonal changes in the ENSO response improve monthly sea-level estimates at San Francisco and Seattle. Each test year used coefficients fit to earlier years. Both models had month means, a time trend, and separate warm and cold ENSO slopes. The added model used two seasonal ENSO terms. The equal-site mean squared error reduction was 2.52% in 1990-2025. Its conditional 95% five-year block interval was -0.69% to +5.76%. Thus the primary test gives no clear gain. A quarterly model had a positive nominal interval, so model form matters. The model uses observed concurrent ENSO; it is a reconstruction test, not a forecast. Two selected Pacific gauges, revised source data, overlapping campaign use, and uncertainty conditional on saved errors limit inference. The package includes fixed input fields, exact source hashes, code, all predictions and block indices, a figure, and numerical checks.

ENSO · sea level · seasonal response · conditional reconstruction · chronological evaluation · null result