An autumn Pacific temperature contrast gave no clear added California winter precipitation prediction gain in 1981-2025

We tested whether the September-November Nino 3 minus Nino 4 temperature anomaly contrast adds information beyond Nino 3.4 for California January-March precipitation. We used fixed NOAA CPC ERSSTv5 and NCEI nClimDiv statewide data. Each fit used earlier winters only, with 45 test winters in 1981-2025. The added term increased mean squared error by 4.18%. The conditional 95% five-year block interval for error reduction was -14.11% to 2.02%. RMSE rose from 132.34 to 135.08 mm. Planned model and lead-season checks gave no clear gain. Removing 1998 from scoring reduced the point loss to 0.46%. This result limits one simple prediction method. It does not refute ENSO diversity or establish a physical cause. Short records, revised inputs, a state mean, a two-region ocean proxy, and intervals conditional on fitted predictions limit inference. The package supplies annual evidence, all predictions and resample scores, source hashes, code, a figure, an independent arithmetic audit, and offline repeat instructions.

ENSO diversity · California · winter precipitation · Nino 3 · Nino 4 · chronological prediction · null result