Comparison between current and future environmental satellite imagers on cloud classification using MODIS

• Published in: Remote sensing of environment Vol. 108; no. 3; pp. 311 - 326
• Main Authors: Li, Zhenglong, Li, Jun, Menzel, W. Paul, Schmit, Timothy J., Ackerman, Steven A.
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Inc 15.06.2007; Elsevier Science
• Subjects: ABI; Animal, plant and microbial ecology; Applied geophysics; Biological and medical sciences; Cloud classification; Cloud detection; Earth sciences; Earth, ocean, space; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Internal geophysics; MODIS; Teledetection and vegetation maps; VIIRS; Cloud detection; ABI; VIIRS; Cloud classification; MODIS; satellites; Classification scheme; Automatic classification; detection; Spectral band; color; Imager; measurement sensor; Method; clouds; Image; Information; currents; Type; snow; signal-to-noise ratio; Environment; spatial resolution; radiance
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/j.rse.2006.11.023

Future Satellite Imagers are expected to improve current ones on environmental and meteorological applications. In this study, an automatic classification scheme using radiance measurements with a clustering method is applied in an attempt to compare the capability on cloud classification by different sensors: AVHRR/3, the current GOES-12 Imager, SEVIRI, VIIRS, and ABI. The MODIS cloud mask is used as the initial classification. The results are analyzed with the help of true color and RGB composite images as well as other information about surface and cloud types. Results indicate that the future sensors (ABI and VIIRS) provide much better overall cloud classification capabilities than their corresponding current sensors (the current GOES-12 Imager and AVHRR/3) from the two chosen demonstration cases. However, for a specific class, it is not always true that more spectral bands result in better classification. In order to optimally use the spectral information, it is necessary to determine which bands are more sensitive for a specific class. Spatial resolution and the signal-to-noise ratio (SNR) of satellite sensors can significantly affect the classification. The 2.13 μm band could be useful for thin low cloud detection and the 3.7 μm band is useful for fresh snow detection.