Research on the fusion of FY4A satellite data and station observation data for heavy fog recognition

Satellite observation of fog possesses technical advantages of wide coverage and high spatial-temporal resolution. However, the accuracy of satellite-based fog identification is subject to errors induced by factors such as atmospheric and radiation conditions. This study aims to improve the accuracy...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Theoretical and applied climatology Jg. 156; H. 1; S. 59
Hauptverfasser: Yao, Zhenhai, Wang, Chuanhui, Jiang, Chun
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Vienna Springer Vienna 01.01.2025
Springer Nature B.V
Schlagworte:
Accuracy
Algorithms
Aquatic Pollution
Atmospheric Protection/Air Quality Control/Air Pollution
Atmospheric Sciences
China
Climatology
Cloud cover
Complementarity
Earth and Environmental Science
Earth Sciences
Fog
Identification
Machine learning
Monitoring methods
Multidimensional methods
Radiation
Radiation-cloud interactions
remote sensing
Satellite data
Satellite observation
Satellites
Spatial distribution
Support vector machines
Temporal resolution
Waste Water Technology
Water Management
Water Pollution Control
China
Fog recognition
FY4A
Optimal algorithm
Station-observation-based correction
Anhui province
ISSN:0177-798X, 1434-4483
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Satellite observation of fog possesses technical advantages of wide coverage and high spatial-temporal resolution. However, the accuracy of satellite-based fog identification is subject to errors induced by factors such as atmospheric and radiation conditions. This study aims to improve the accuracy of fog identification by integrating ground-based station observations with the Fengyun-4 A (FY-4 A) satellite data. Taking Anhui Province as the study area, we establish a fog identification model using multiple algorithms, namely threshold method (THD), support vector machine (SVM), random forest (RF) and gradient boosting (XGB). In addition, a nearby pixel method is employed to validate identification results, in order to select the optimal algorithm. The results indicate that machine learning algorithms outperform the THD method in fog identification. Among the SVM, RF and XGB algorithms, the RF method exhibits the highest median KSS (0.66) and excellent robustness, and thus it is the optimal algorithm. Case studies demonstrate that the RF-based identification results effectively reflect the spatial distribution of fog regions. Although the differences between the images of identification results before and after correction are not obvious, the identification accuracy is highly susceptible to instability due to factors such as radiation, cloud cover and fog intensity. After correction based on station observations, the model KSS scores are noticeably improved (up to 67.2%) and become more stable. Compared with single-satellite-data-based fog monitoring methods, the integration of the FY-4 A satellite data and station observations offers multi-dimensional observation complementarity and achieves technological advances in the digitization and spatialization of fog observations.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:0177-798X
1434-4483
DOI:10.1007/s00704-024-05245-6