RONI gave no clear California winter rain prediction gain over ONI in a 1981-2025 linear test

We tested whether September-November ERSSTv5 RONI gives lower California January-March precipitation prediction error than ONI. Fits used earlier winters only. The test covers 45 winters in 1981-2025. Replacing ONI with RONI reduced mean squared error by 1.40%. The conditional 95% five-year block interval was -2.76% to 4.91%. RMSE fell from 132.06 to 131.13 mm. Year-term, training-window, lead-season and fixed-reference checks also gave intervals across zero. The point gain was positive in 1981-2002 and negative in 2003-2025; this does not establish a change in the climate relation. This test gives no clear prediction advantage for the index replacement. One state, revised inputs, one linear model, short records and conditional uncertainty limit inference. Code, seasonal values with source credit, all predictions and scores, source hashes, a separate computation audit and a figure are included. Raw sources stay outside the signed package.

RONI · ONI · ENSO · California precipitation · chronological prediction · null result