Error Structure Diagnosis and Correctability of FY-4B/GIIRS Temperature Profiles over the Central-Eastern Tibetan Plateau
Atmosphere · 25.07.2026 · 17 (8) · 725
DOI: 10.3390/atmos17080725 ↗Kurz gefasst
Fachartikel in der Zeitschrift Atmosphere, online erschienen am 25.07.2026. Verfasst von Wan Feng und 2 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
FY-4B/GIIRS temperature-profile products provide an important source of atmospheric sounding information for the radiosonde-sparse Tibetan Plateau, but their error structure and correctability over complex terrain remain insufficiently understood. In this study, the FY-4B/GIIRS temperature profiles over the central-eastern Tibetan Plateau were systematically evaluated using collocated satellite and radiosonde observations from 2023 to 2025. In addition to conventional validation metrics, an error-structure diagnosis was conducted to distinguish the relative contributions of systematic and random components of the satellite–radiosonde differences across 12 cloud–season–layer scenarios. The results show that product quality varies markedly with cloud condition and season. Clear-sky samples are more frequent during the cold season, whereas cloudy samples increase during the warm season, and the proportion of high-quality retrievals decreases substantially under cloudy conditions. Retrieval errors also exhibit clear vertical dependence, with the highest accuracy in the middle layer and larger errors in the lower and upper layers. The error-structure diagnosis further reveals pronounced scenario dependence: random errors dominate most clear-sky conditions, whereas systematic biases are particularly evident in the clear-sky lower layer in winter, the clear-sky upper layer in summer, and most cloudy scenarios. A scenario-based LightGBM framework effectively reduces these systematic differences across the 12 cloud–season–layer scenarios. Test-set R2 ranges from 0.43 to 0.82, while mean Bias decreases from −1.66 to 0.01 K and mean RMSE from 3.35 to 1.78 K, corresponding to an average reduction of 45.56%. In contrast, the residual-error magnitude is less predictable, with STD-model R2 values of 0.09–0.37. These findings indicate that error-structure diagnosis can help identify correctable retrieval-error components and that scenario-based correction provides an effective approach for improving the reliability of FY-4B/GIIRS temperature profiles over complex plateau terrain.
Zitieren
APA
Feng, W., Wang, W., & Zhou, X. (2026). Error Structure Diagnosis and Correctability of FY-4B/GIIRS Temperature Profiles over the Central-Eastern Tibetan Plateau. Atmosphere, 17(8), 725. https://doi.org/10.3390/atmos17080725
Vancouver
Feng W, Wang W, Zhou X. Error Structure Diagnosis and Correctability of FY-4B/GIIRS Temperature Profiles over the Central-Eastern Tibetan Plateau. Atmosphere. 2026;17(8):725. doi:10.3390/atmos17080725
BibTeX
@article{Feng2026,
author = {Feng, Wan and Wang, Wei and Zhou, Xueying},
title = {Error Structure Diagnosis and Correctability of FY-4B/GIIRS Temperature Profiles over the Central-Eastern Tibetan Plateau},
journal = {Atmosphere},
year = {2026},
volume = {17},
number = {8},
pages = {725},
doi = {10.3390/atmos17080725}
}
EUVIMED