Integrity review: A Convective Initiation Nowcasting Algorithm Based on FY-4B Satellite AGRI and GHI Data (Atmosphere, DOI: 10.3390/atmos17040380)

Academic fraud report · Geng Detector

Summary

Verdict: questionable but not fraudulent. A systematic textual and mathematical review of Yang et al. (Atmosphere 2026, DOI 10.3390/atmos17040380) finds no evidence of data fabrication. The core evaluation numbers in Tables 5 and 7 are internally consistent: 22+19+10+13+29+20+24+29 = 166; TP(138)+FN(28) = 166; POD = 138/166 = 83.13%, matching the reported 83.1%; CSI = 138/234 = 58.97%, matching 58.9%. The timeline (FY-4B launched 2021, operational 2023; data from July–August 2023; Typhoon Doksuri landfall 28 July 2023) is plausible. Concerns are limited to presentation: a 96% CI (vs. the 95% CI used elsewhere) in Table 7 for the FAR of the GHI algorithm (33%, 26.3–39.6%), a nonsensical phrase "bright semichemic change," the tautology "cumulonimbus clouds will be cloudy," and "Chengdu University" likely being a machine-translation artifact for the Chengdu 2023 FISU World University Games. No image-level analysis was performed. Confidence in the verdict is moderate; final determination of misconduct requires institutional investigation.

Verdict

🟡 Questionable (no fraud detected) — Core experimental data are mathematically self-consistent and no fabricated results were identified. Concerns are confined to sloppy drafting/proofreading: anomalous "96% CI," apparent machine-translation artifacts, and a misrendered event name. None of these invalidate the science, but they indicate a deficient editorial process at the journal.

Key findings

Statistical self-consistency (Tables 5 & 7): Daily counts sum to 166; TP+FN = 138+28 = 166. POD = 138/166 = 83.13% (reported 83.1%). CSI = 138/234 = 58.97% (reported 58.9%). All match.
Anomalous confidence level (Table 7): For the GHI algorithm's FAR, the table reports "33% (96% CI: 26.3–39.6%)" while every other CI in the paper is 95%. This is almost certainly a typo that survived peer review.
Nonsensical terminology (Section 2.2.1): The phrase *"15-min bright semichemic change"* has no meteorological meaning and is inconsistent with the surrounding context of brightness-temperature differences.
Tautological language: *"Typically, cumulonimbus clouds will be cloudy"* is a meaningless filler sentence.
Event-name mistranslation (Sections 2.1 & 3.1): "Chengdu University" is almost certainly a machine-translation error for the Chengdu 2023 FISU World University Games (31st Summer Universiade), which is the only major Chengdu event that would align a high-frequency FY-4B/GHI scan campaign in July–August 2023.
Timeline plausibility: FY-4B launched 2021, operational in 2023; data cover 2023-07 to 2023-08; Typhoon Doksuri made landfall on 28 July 2023, matching the authors' explanation of missing Sichuan observations. Submitted 2026 — no anachronism.

Evidence highlights

22 + 19 + 10 + 13 + 29 + 20 + 24 + 29 = 166; TP 138 + FN 28 = 166 ✓
POD = 138 / (138 + 28) = 83.13% vs. reported 83.1% ✓
CSI = 138 / (138 + 28 + 68) = 58.97% vs. reported 58.9% ✓
Table 7, GHI algorithm FAR row: "33% (96% CI: 26.3–39.6%)" — outlier vs. 95% CI elsewhere.
Section 2.2.1: *"15-min bright semichemic change"* — undefined term, likely a mistranslation of brightness-temperature change.
Section 3.1: reference to *"Chengdu University"* as the temporal anchor for 2023 GHI scans — best explained as a mistranslation of the 2023 FISU World University Games held in Chengdu.

Notes

DOI: 10.3390/atmos17040380 (Atmosphere, MDPI; PDF source: atmosphere-17-00380-v2.pdf).
No image-level or pixel-level forensic analysis was performed; review is text-, arithmetic-, and background-knowledge-based.
Suggested follow-ups: (1) PubPeer comment requesting clarification of the 96% CI and "bright semichemic change"; (2) author contact recommending revision of English; (3) formal erratum request to the journal.
AI-assisted review; final misconduct determinations require institutional investigation.