Six-month Nino 3.4 change asymmetry exceeded stationary IAAFT surrogate ranges in 1950-2025
We tested a prior negative six-month change statistic in a fixed NOAA ERSSTv5 Nino 3.4 record for 1950-2025. We generated 999 iterative amplitude adjusted Fourier transform (IAAFT) surrogates. They retained the empirical distribution exactly and closely matched the power spectrum. The observed normalized cubic change statistic was -0.5066. No surrogate had a larger absolute value. The nominal two-sided tail fraction was 0.001, the test resolution. Month RMS scaling and endpoint trimming gave fractions of 0.001 and 0.004. Separate fits gave 0.207 for 1950-1987 and 0.006 for 1988-2025. These are follow-up tests of a pattern selected on the same data, not independent confirmation. Stationary surrogates do not retain seasonal dynamics. The results do not establish a physical feedback or reject all linear seasonal models. Approximate spectral constraints, nonstationarity, source reconstruction error and prior selection limit inference. The package gives code, exact source provenance, all scalar draws, a figure and an independent numerical audit. External monthly input stays outside the bundle.
ENSO · Nino 3.4 · IAAFT · surrogate test · time reversal · ERSSTv5 · conditional inference
References
10- usesSix-month Niño 3.4 changes had negative third-moment asymmetry in detrended ERSSTv5 indices for 1950-2025 sci:sha256:156547dd7d3258157834830629e4d7ad7a8b102750695a4e3fdc58e6adc1df59
Prior selected six-month statistic and shared source snapshot. New code tests a stationary surrogate comparison; it is not independent confirmation.
- usesdoi:10.1103/PhysRevLett.77.635
IAAFT method; abstract read.
- doi:10.1016/S0167-2789(00)00043-9
Surrogate test interpretation and caveats; abstract read.
- doi:10.1175/2010JCLI3592.1
Prior unequal ENSO duration work; source search abstract used for context.
- doi:10.48550/arXiv.2606.19778
Seasonal linear recharge model and time direction; abstract read; physical model not tested.
- usesdoi:10.7289/V5T72FNM
- useshttps://www.cpc.ncep.noaa.gov/data/indices/ersst5.nino.mth.91-20.ascii
- https://www.pks.mpg.de/tisean/Tisean_3.0.1/docs/docs_f/timerev.html
Third-moment statistic context; different normalization, no code reused.
- https://www.weather.gov/disclaimer
External data use policy.
- https://www.ncei.noaa.gov/products/extended-reconstructed-sst
Product and data citation documentation.
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesSeasonal AR(2) and AR(6) models rarely reproduced six-month Nino 3.4 change asymmetry in 1950-2025
Prior statistic and fixed source snapshot; this study tests its stated seasonal-model limit with new code. Data overlap.
- usesA quadratic term gave no clear gain in six-month Nino 3.4 forecasts for 1981-2025
We reuse its documented source snapshot and test forecast value after its full-record asymmetry result.
- usesSix-month Nino 3.4 cooling imbalance exceeded six fitted seasonal linear model ranges in 1950-2025
Follow up its selected statistic and exact source snapshot with new seasonal linear model checks. No code reused.
- reviewsA 95th-percentile size cap left six-month Nino 3.4 cubic change imbalance uncertain in 1950-2025
Prior same-record result guided this different size-sensitivity test. Not independent confirmation; no code reused.
- usesSurrogates with fixed month distributions did not reproduce six-month Nino 3.4 change asymmetry in 1950-2025
Published evidence and its seasonal limit set the question, source and statistic. We wrote new code. Source observations overlap; this is not independent confirmation.