Integrity Check Report: Optimization and accuracy validation of near-space 1.27 μm O2(a1Δg) dayglow forward model based on limb observation mode

Academic fraud report · Geng Detector

Summary

Verdict: No integrity concerns identified; the paper appears legitimate. The Geng-style audit examined six dimensions: image reuse, data fabrication, image splicing, statistical anomalies, output productivity, and citation/methodology. No substantive issues were detected across any dimension. Key supportive evidence includes natural numerical variation in error metrics (e.g., mean relative errors of 14.19%, 22.46%, 15.40%, and 20.56% across four orbits rather than uniform values), physically consistent perturbation responses in sensitivity tables, honest disclosure of model limitations (e.g., the 50–85 km ozone overestimation bias in MIPAS_2001 affecting orbit 43,350), and an unusually transparent data availability statement depositing all data and plotting code on Zenodo (10.5281/zenodo.15354984). The first author's timeline—undergraduate graduation in 2022, doctoral candidate, single first-author methodology paper—is consistent with normal PhD progression. Limitations of the audit: original high-resolution image files were not available, so visual similarity and pixel-level splicing checks could not be performed.

Verdict

No integrity concerns identified. The paper appears legitimate across all six audit dimensions (image reuse, data fabrication, image splicing, statistical anomalies, productivity, and methodology/citations). Confidence is moderate-to-high; the only residual limitation is the absence of raw image pixel files for full visual comparison.

Key findings

Data numerics are consistent with genuine computation: Table 1 perturbation responses retain realistic floating-point tails (e.g., −13.00, −14.20, −12.85, −17.14) rather than fabricated constant offsets.
Error metrics show natural variability: Four-orbit mean relative errors of 14.19%, 22.46%, 15.40%, and 20.56% (orbits 42,033; 43,350; 44,667; 45,988) are non-uniform and physically plausible.
Honest limitation reporting: Authors disclose that the MIPAS_2001 ozone climatology overestimates O3 between 50–85 km, producing visible bias in the 43,350 orbit simulation—consistent with authentic research practice.
Ablation experiment arithmetic holds: Residual reductions to 0.8% and 0.99% for the full model versus baseline are mathematically consistent.
Open data and code deposited on Zenodo: DOI 10.5281/zenodo.15354984 provides all experimental/model data and figure-generation code—an unusually strong transparency signal.
Methodology tools verified: MODTRAN5, HITRAN, and SASKTRAN are established atmospheric-optics resources; 2010-vintage SCIAMACHY and SABER satellite data are appropriate for algorithm validation.
Author timeline normal: First author Guoyin Zeng (B.Sc. 2022, current PhD student) publishing a single first-author methodology paper is consistent with typical doctoral trajectories.

Evidence highlights

DOI: 10.1080/10095020.2026.2633870
Authors: Guoyin Zeng, Wei Xiong, Zhiwei Li, Haiyan Luo, Yuan An
Journal: Geo-spatial Information Science (published online 09 Apr 2026)
Submitted July 2025; accepted February 2026
Zenodo data/code archive: https://doi.org/10.5281/zenodo.15354984
Table 1 (Section 4.2.1, pp. 15–16): sensitivity perturbation grid with non-rounded floating-point responses
Section 4 (Accuracy validation) and Section 5.3 (Ablation experiment): residual/error analysis with disclosed model bias

Notes

Image-based checks (visual similarity, splicing artifacts, background noise) could not be executed because only text and partial captions were available for analysis; figures were not assessed at pixel level.
The paper is an atmospheric remote-sensing modeling study, so biological/statistical red flags (Western blot duplication, p-value anomalies) are not applicable.
Verification of Zenodo archive contents (file completeness, reproducibility of figures) was not performed in this audit and is recommended as an optional follow-up if higher assurance is required.
All findings are based on textual and metadata evidence; final determination of academic integrity rests with institutional investigation.