Yuzhu Tang

Sea fog strongly affects marine transportation and offshore operations, yet its vertical extent remains difficult to determine over broad ocean areas. Active remote sensing can provide accurate vertical information, but its narrow sampling and low temporal resolution make it unsuitable for continuous, large‐scale monitoring of sea fog top height (SFTH). Existing widely used passive satellite cloud top height (CTH) retrieval methods are not well suited to shallow sea fog. Focusing on the Yellow and Bohai Seas, we present a daytime MODIS‐based method for retrieving SFTH by deriving a linear relationship between sea fog reflectance and SFTH, constrained by island peak elevation. This relationship is approximated by establishing a linear classification model that separates islands embedded in sea fog scenes into two categories, visible and obscured . Across 49 sea fog events satisfying the applicability criteria, the proposed retrieval agrees more closely with CALIOP‐derived SFTH than the operational MODIS CTH product. For the logistic regression implementation, the mean absolute error and root mean square error decrease from about 304 to 376 for MODIS to about 65 and 98 , respectively, while the hit rate, defined as the fraction of samples with absolute errors no greater than 30 , increases from about 4% to 39%. Note that the method requires at least two visible and two obscured islands, without obstruction by overlying clouds.

Sea fog has significant impacts on both human activities and the natural environment. Sea fog top height (SFTH) reflects the impact of sea fog on vertical space and is a crucial parameter for both Numerical Weather Prediction models and the estimation of sea fog dissipation. The existing SFTH retrieval methods based on spaceborne passive radiometers are prone to significant errors. We focus on the Yellow and Bohai Seas region and introduce a high-precision SFTH retrieval algorithm based on the peak elevation (PE) of islands in the area. In remote sensing images, islands of different PE within the sea fog regions exhibit two distinct appearances either visible or obscured by the fog. Thus, islands are automatically classified into two categories by support vector machine. The relationship between SFTH and the corresponding sea fog reflectance (SFR) in remote sensing images is established by defining a linear decision boundary in the SFR-PE space through logistic regression or support vector machine to effectively separate the two island categories. Validation experiments on twelve sea fog cases show that the proposed method exhibits higher accuracy compared to the MODIS cloud top height product and demonstrates good agreement with CALIPSO data.

The distribution and frequency of occurrence of different cloud types affect the energy balance of the Earth. Automatic cloud type classification of images continuously observed by the ground-based imagers could help climate researchers find the relationship between cloud type variations and climate change. However, by far it is still a huge challenge to design a powerful discriminative classifier for cloud categorization. To tackle this difficulty, in this paper, we present an improved method with region covariance descriptors (RCovDs) and the Riemannian bag-of-feature (BoF) method. RCovDs model the correlations among different dimensional features, which allows for a more discriminative representation. BoF is extended from Euclidean space to Riemannian manifold by k-means clustering, in which Stein divergence is adopted as a similarity metric. The histogram feature is extracted by encoding RCovDs of the cloud image blocks with a BoF-based codebook. The multiclass support vector machine (SVM) is utilized for the recognition of cloud types. The experiments on the ground-based cloud image datasets show that a very high prediction accuracy (more than 98 % on two datasets) can be obtained with a small number of training samples, which validate the proposed method and exhibit the competitive performance against state-of-the-art methods.