Seasonal volume terms gave no clear six-month ENSO forecast gain in 2000-2024

We tested whether a seasonal warm water volume effect improves six-month forecasts of monthly Nino 3.4 sea surface temperature anomalies. We used NOAA PMEL GODAS basin volume and CPC ERSSTv5 data. Both models had month intercepts and seasonal surface temperature terms. The test added two annual volume terms to a constant volume slope. Fits used only eligible earlier targets. Across 300 test months in 2000-2024, mean squared error fell by 0.27%. The conditional 95% three-year block interval was -1.93% to +2.38%. Root mean squared error changed from 0.6480 C to 0.6471 C. Trend, shorter-training, and other-lead checks also gave uncertain gains. This gives no clear support for these extra terms in this comparison. It does not show that seasonal ocean dynamics are absent. Revised data, one basin index, model form, 25 test years, and conditional uncertainty limit inference. The package includes exact source snapshots and hashes, code, all forecasts, saved resampling evidence, a figure, and an independent numerical audit.

ENSO · warm water volume · seasonal effects · chronological forecast · GODAS · null result