Geng Report: Temperature retrieval of near space with the combined use of O2(a1Δg) and O2(b1Σ+g) dayglow emissions under self-absorption effect correction

Academic fraud report · Geng Detector

Summary

Verdict: No fraud / clean paper. Optics Express paper DOI 10.1364/OE.532076 by Wang, Li, He, Wang, Li, and Wu (2024) passed the six checks (image reuse, data fabrication, image splicing, statistics, output anomalies, citation/methodology) with no positive indicators of academic misconduct. Key supporting observations: (1) cross-validation noise patterns vs SABER, ACE-FTS, and MIPAS satellite data show physically driven error profiles rather than fabricated uniform accuracy; (2) authors openly report an early linear-fit slope of only 0.19 with line-intensity retrievals before switching to Einstein-coefficient-based retrieval to reach 0.98; (3) data, instruments, and versions are internally consistent (Envisat SCIAMACHY orbits 47085/47086/47775/47776 from March–April 2011; HITRAN 2020; NRLMSIS 2.0); (4) submission-to-acceptance timeline (June 5 to July 25, 2024) is reasonable. Confidence: high for checks 2 and 4–6; medium for image-based checks because pixel-level comparison was not possible.

Verdict

No evidence of academic fraud. The paper is considered clean under the six checks applied, with appropriate caveats for image-based checks due to the absence of pixel-level data.

Key findings

Image reuse (Figures 1–20): Pixel-level comparison not possible; captions describe distinct chart purposes (reaction flow, spectrum comparison, inversion results, error analysis). Severity: 🟡 inconclusive pending original images.
Data fabrication (Abstract, Section 4.2, Section 5): Cross-validation errors against SABER, ACE-FTS, and MIPAS show physically motivated variation (e.g., best accuracy <10 K between 40–70 km, peak error ~40 K at 120 km, increased scatter near the surface due to scattering). Pattern is consistent with real measurements, not fabricated uniform errors.
Image splicing: Spectral figures are remote-sensing style; the described "numerical substitution" of bad pixels is standard remote-sensing processing and is openly disclosed.
Statistics (Section 4.2.2): RMS and mean temperature differences follow physics: minimum ~3 K at 60–90 km, larger deviation at 35 km (cumulative self-absorption) and 120 km (low SNR).
Output anomalies: Received 5 June 2024, revised 24 July 2024, accepted 25 July 2024, published 2 August 2024; ~1.5-month timeline typical for Optics Express.
Citing & methodology: Envisat SCIAMACHY observation dates (March–April 2011; orbits 47085, 47086, 47775, 47776), the satellite's actual end of mission (April 2012), HITRAN 2020 (Gordon et al., 2022), and NRLMSIS 2.0 (2021) align with a 2024 publication.

Evidence highlights

Self-reported honest negative result: with line-intensity only, Figure 17 linear fit slope = 0.19; after switching to Einstein coefficients, slope rose to 0.98.
Quantitative claims retained verbatim from report:
"40–70 km error better than 10 K"; "120 km peak error ~40 K"
Best agreement ~3 K in the 60–90 km range vs ACE-FTS
Timeline numbers: received 2024-06-05; accepted 2024-07-25; published 2024-08-02.
Instrument/coverage: SCIAMACHY MLT mode 2008-07 to 2012-04.
Database versions: HITRAN 2020 (Gordon et al., 2022); NRLMSIS 2.0 (Emmert et al., 2021).

Notes

Image-reuse and splicing verifications are textual only; final confirmation would require access to the original image files for pixel-level comparison, specifically Figures 4 and 10 as suggested by the source report.
All quoted numbers preserve the original Chinese report's precision and phrasing.
DOI retained exactly as provided: 10.1364/OE.532076.
Report generated as an AI-assisted academic-integrity screening; final determination requires institutional investigation.