Implicit-tuning signal quantified through a tri-experiment design in a legacy spectral GCM

Implicit-tuning signal quantified through a tri-experiment design in a legacy spectral GCM

This is a Preprint and has not been peer reviewed. This is version 1 of this Preprint.

Add a Comment

You must log in to post a comment.


Comments

There are no comments or no comments have been made public for this article.

Downloads

Download Preprint

Supplementary Files

Authors

Dmitrii Vokhmintsev 

Abstract

We quantify the climate response of the MGO atmospheric general circulation model (MGO-03 T42L25) to two alternative implementations of a structural fix in the vertical-exchange parameterization. A companion parallelization study (under review) identifies an unintended inter-latitude state leakage through SAVE-persistent working arrays in the boundary-layer scheme. Here we contrast the legacy code (REF) against two candidate repairs—zero-reinitialization at each latitude (CTRL_A) and per-latitude snapshot/restore of the prior state (CTRL_B)—in a three-member initial-condition ensemble of 30-year AMIP-style integrations (1979-2008), and evaluate the simulated climatologies against ERA5, CERES EBAF, GPCP, and ISCCP-H observational datasets. Pairwise comparisons with pooled-paired t-tests across the ensemble, an effective-sample-size correction for temporal autocorrelation, and Benjamini-Hochberg control of the false discovery rate at q = 0.05 reveal that (i) the choice between the two implementations—contrasted at identical compilation settings (CTRL_B-CTRL_A)—is itself FDR-significant in 20.34% of T42 cells in annual surface air temperature, exceeding the CTRL_B-REF leak-elimination signal in both area-weighted amplitude and FDR-significant spatial extent; (ii) both repairs enlarge the model bias in radiation and cloud cover against observations relative to REF, a pattern consistent with the hypothesis that the legacy configuration was implicitly tuned against the leaky baseline; the ensemble establishes the signal's robustness but does not isolate the mechanism among several plausible alternatives. We additionally document that the legacy implementation is not thread-reproducible across OpenMP partitions. A formal bit-reproducibility theorem, identifying three implementation conditions under which OpenMP parallelization of the model is bitwise reproducible by construction, is presented in the Supporting Information.

DOI

https://doi.org/10.31223/X54F51

Subjects

Atmospheric Sciences, Climate, Numerical Analysis and Scientific Computing

Keywords

atmospheric general circulation model, OpenMP parallelization, bitwise reproducibility, implicit tuning, AMIP, vertical-exchange parameterization, MGO GCM, T42L25, ensemble, Benjamini-Hochberg FDR, pooled-paired t-test, Trenberth N_eff, replication package

Dates

Published: 2026-06-14 16:21

Last Updated: 2026-06-14 16:21

License

CC BY Attribution 4.0 International

Additional Metadata

Conflict of interest statement:
The author was previously employed at the Voeikov Main Geophysical Observatory (Saint Petersburg, Russia) for a 2.5-year period during which the parallelization of the MGO atmospheric general circulation model was initiated but not completed; this employment concluded in July 2025. The substantive work reported in this manuscript — including the discovery of the inter-latitude state leakage in the vertical-exchange parameterization (VE_JRESET), the formal bit-reproducibility theorem, the three-member tri-experiment ensemble design (REF / CTRL_A / CTRL_B), the paired statistical framework, and the entire climate-impact analysis — was completed over the subsequent year as an independent, self-directed effort on personal computational resources, without institutional affiliation or external funding. The author has no current affiliation with, or financial interest in, the Voeikov Main Geophysical Observatory or any related institution. The author holds one issued Rospatent computer-program registration covering the structural repair, the deterministic-reduction modules, and the analysis pipeline reported here, and has two related Rospatent patent applications under examination on the parallel-reduction and thread-safety methodology; these are derivative works that do not cover the proprietary MGO model itself. There are no other competing interests.

Data Availability:
The replication package (input data manifests, analysis code, derived NetCDF, figure scripts, and verification artifacts) is publicly archived at Zenodo: https://doi.org/10.5281/zenodo.20447240

Metrics

Views: 22

Downloads: 0