Abstract: The Arctic region is one of the most sensitive areas in the world to climate change; therefore, accurate detection of Arctic clouds is of great significance for improving the accuracy of radiation energy balance estimation in the Arctic region and worldwide. Most of the Arctic region is covered by ice and snow throughout the year, and there are few ground observation stations. Accordingly, the application of multispectral satellite remote sensing technology has become a necessary means of Arctic cloud detection, and satellite remote sensing cloud detection datasets lay a crucial foundation for this methodology. This paper reviews the commonly used satellite remote sensing cloud detection datasets (domestic and international) and provides a comprehensive summary of the representative works of three types of Arctic cloud detection methods; i.e., the traditional threshold method, the classical machine learning method, and the deep learning method. By analyzing and comparing the advantages and limitations of representative works using different methods, the paper evaluates the existing problems in Arctic remote sensing image cloud detection methods and discusses the development potential of the domestic Feng Yun satellite for Arctic cloud detection research. Furthermore, based on future development trends for Arctic cloud detection methods, the paper provides recommendations for establishing the Arctic joint network observation system and a comprehensive remote sensing database.

Key words: remote sensing image cloud, detection methods, deep learning, Arctic