Publications

Kim, DC; Shin, DB (2024). Gaussian Mixture Model-Based Cloud- Phase Estimation From GEO- KOMPSAT-2A Observations. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 62, 4103515.

Abstract
Cloud-phase algorithms based on satellite infra-red (IR) observations typically use empirical thresholds andadditional cloud properties to predict cloud phases, but these set-tings may not be consistently accurate under various conditions.This study introduces an unsupervised machine learning (ML)approach based on a Gaussian mixture model (GMM) to avoidthe empirical determination of thresholds and auxiliary cloudproducts for cloud-phase estimation. The GMM-based cloud-phase algorithm proposed in the present study consists of threeGMMs that distinguish cluster types representing water, ice, andundetermined phases using the brightness temperature (TB) at11.2 mu m and the difference in TB between 8.6 and 11.2 mu mfrom Geostationary-Korean Multi-Purpose Satellite-2A (GEO-KOMPSAT-2A, GK2A) satellite. The first GMM initially classifiesfour clusters, while the second and third GMMs regroup theinitial clusters to distinguish supercooled water, mixed-phase,and optically thin cirrus clouds. The GMM-derived estimatesare compared with operational cloud-phase products from theModerate Resolution Imaging Spectroradiometer (MODIS) andCloud-Aerosol Lidar with Infrared Pathfinder Satellite Obser-vations (CALIOP). Results show that the water and ice phasesestimated using the GMM-based algorithm are in good agreementwith both the MODIS and CALIOP products. The GMM-basedalgorithm also significantly reduces the misidentified area forundetermined phases observed in the GK2A operational product.Water and ice phases are also effectively estimated in warmregions, resulting in distributions similar to those derived fromMODIS and CALIOP products. Unlike most IR cloud-phasealgorithms that utilize thresholds and other cloud parameters, theGMM-based cloud-phase algorithm has the advantage of usingonly TB, thus avoiding auxiliary cloud properties

DOI:
10.1109/TGRS.2024.3383888

ISSN:
1558-0644