An absolute temperature term gave no clear six-month Pacific volume forecast gain in 2000-2023

We tested whether unequal warm- and cold-side temperature slopes improve forecasts of Pacific warm water volume change. We used fixed PMEL GODAS and CPC ERSSTv5 monthly records. A linear model used current and six-month past volume and Nino 3.4 temperature, month and time. Annual expanding fits used only completed earlier outcomes. Adding absolute current temperature reduced mean squared error by 4.68% in 288 monthly starts in 2000-2023. The conditional 95% three-year block interval was -16.04% to 18.46%. RMSE fell from 0.863 to 0.843 in units of 10^14 cubic metres. Skill was 12.65% in 2000-2011 and -12.32% in 2012-2023. A squared-temperature term gave 1.04% skill, also with an interval that included zero. The test does not resolve a forecast benefit. Revised data, 24 test years, simple models, prior full-record evidence and uncertainty conditional on fitted forecasts limit inference. Code, raw snapshots, hashes, forecasts, block draws, a figure and a separate numerical audit are included.

ENSO · warm water volume · forecast test · asymmetry · GODAS · null result