LSDSSIMR: Large-Scale Dust Storm Database Based on Satellite Images and Meteorological Reanalysis Data

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing Vol. 16; pp. 1 - 11
• Main Authors: Bai, Cong, Cai, Zhipeng, Yin, Xiaomei, Zhang, Jinglin
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.01.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Aerosols; Atmospheric models; Atmospheric motion; Atmospheric particulates; Deep learning; Detection; Dust; Dust storm; Dust storms; Fengyun-4 A (FY-4A); FY-4A; Meteorological reanalysis; Meteorological satellites; Meteorology; Monitoring; Satellite imagery; Satellite observation; Satellites; Spatial discrimination; Spatial resolution; Spectral analysis; Spectrum analysis; Storms; Temporal resolution; Weather forecasting
• ISSN: 1939-1404; 2151-1535
• DOI: 10.1109/JSTARS.2023.3325783

In recent years, dust storms have occurred frequently, significantly affecting people's daily lives. Therefore, the detection, monitoring, and early warning of dust storms have a great social significance. Previous methods have mainly been based on atmospheric motion to build physical models for weather forecasting. Although there are many meteorological applications based on deep learning, to the best of our knowledge, there is no dust storm database with a high spatiotemporal resolution, which is essential for deep learning methods. Since meteorological satellites can observe the Earth's atmosphere from a spatial perspective at a large scale, in this paper, a dust storm database is constructed using multi-channel and dust label data from the Fengyun-4A (FY-4A) geosynchronous orbiting satellite, namely, the Large-Scale Dust Storm database based on Satellite Images and Meteorological Reanalysis data (LSDSSIMR), with a temporal resolution of 15 minutes and a spatial resolution of 4km from March to May of each year during 2020-2022. Meteorological reanalysis data are added to LSDSSIMR for spatiotemporal prediction methods. Each data file is stored in HDF5 format, and the final LSDSSIMR contains nearly 5400 HDF5 files. Moreover, some traditional dust detection methods based on spectral analysis are executed as a benchmark