Geng Integrity Report: "First Level 1 Product Results of the Greenhouse Gas Monitoring Instrument on the GaoFen-5 Satellite" (IEEE TGRS, 2020, DOI: 10.1109/TGRS.2020.2998729)

Academic fraud report · Geng Detector

Summary

This Geng academic-integrity report rates the paper as 'highly suspicious' (orange). The central concerns are two clear numerical contradictions between the Abstract/Conclusion and the main body. First, the claimed spectral stability values (Band 1 <0.005 cm⁻¹, Band 2 0.003 cm⁻¹, Band 3 0.002 cm⁻¹, Band 4 0.004 cm⁻¹) differ from Table III by roughly an order of magnitude (0.03, 0.011, 0.013, 0.02 cm⁻¹), consistent with a misplaced decimal point in the abstract. Second, the abstract reports an XCO₂ mean absolute error of 0.51 ppm, while Section VI reports a bias and σ of 0.53 ppm for XCO₂. A third issue notes that 109 XCO₂ but only 60 XCH₄ match-ups were produced over the same five TCCON sites, without explanation. The report cannot assess image manipulation because the figures are remote-sensing engineering plots, not micrographs. Overall confidence in the internal-inconsistency findings is high; conclusions about intent (typo vs. fabrication) cannot be drawn from the report alone.

Verdict

🟠 Highly suspicious. Multiple numerical contradictions exist between the Abstract/Conclusion and the main body. Intent (editorial error vs. fabrication) cannot be determined from this report alone.

Key findings

Internal inconsistency in spectral-stability values: Abstract/Conclusion list Band 1 <0.005 cm⁻¹, Band 2 0.003 cm⁻¹, Band 3 0.002 cm⁻¹, Band 4 0.004 cm⁻¹; Table III in Section V-C lists 0.03 cm⁻¹, 0.011 cm⁻¹, 0.013 cm⁻¹, 0.02 cm⁻¹. The discrepancy is approximately 10× (≈6–10× depending on band) and is consistent with a misplaced decimal in the abstract.
Inconsistency in XCO₂ validation accuracy: Abstract states a mean absolute error of 0.51 ppm; Section VI retrieval results report R² = 0.869, bias = 0.53 ppm, and σ = 0.53 ppm for XCO₂.
Unexplained sample-size mismatch: The retrieval experiment yields 109 XCO₂ but only 60 XCH₄ co-locations with TCCON over the same five sites using the same matching criteria (Δt < 2 h, Δlat/lon < 5°), despite GMI's four bands observing simultaneously. No explanation for the XCH₄ attrition is provided—potential selective reporting.
Image-level analysis: Not applicable. All 25 figures are remote-sensing engineering plots (orbital tracks, interferograms, flowcharts, scatter fits), not micrographs suitable for pixel-level forensics.

Evidence highlights

Spectral stability (Abstract/Conclusion): Band 1 <0.005 cm⁻¹; Band 2 0.003 cm⁻¹; Band 3 0.002 cm⁻¹; Band 4 0.004 cm⁻¹.
Spectral stability (Section V-C, Table III): 0.03 cm⁻¹, 0.011 cm⁻¹, 0.013 cm⁻¹, 0.02 cm⁻¹.
XCO₂ validation (Abstract): MAE = 0.51 ppm.
XCO₂ validation (Section VI): R² = 0.869; bias = 0.53 ppm; σ = 0.53 ppm.
Match-up counts (Section VI): 109 XCO₂ and 60 XCH₄; matching criteria identical across gases.
DOI: 10.1109/TGRS.2020.2998729.

Notes

The spectral-stability values in the Abstract are all shifted by roughly one decimal place relative to Table III, suggesting a likely typographical/transcription error rather than independent fabrication, but this is not certain.
The XCO₂ Abstract-vs-body discrepancy (0.51 vs 0.53 ppm) is small but concerns a headline validation metric; the report marks it as confirmed but moderate severity.
The XCO₂/XCH₄ count gap could reflect legitimate band-specific quality filtering (e.g., SWIR-2 channel noise, glint, or aerosol rejection); without the raw match-up list this remains an open question.
Recommended follow-ups in the original report: compare the as-accepted version with earlier online versions; request the 109/60 match-up lists from the authors; raise the stability discrepancy on PubPeer; request an Erratum for the conflicting values.
Image manipulation could not be assessed due to the nature of the figures.
This is an AI-assisted integrity report; final determination of misconduct requires investigation by an appropriate institutional or editorial body.