Feature extraction of fog from multi-spectral infrared images of FY-2C geostationary satellite

FY-2C is geostationary satellite which is researched and developed by China. The primary advantage of geostationary satellite is the ability to characterize the radiance by obtaining numerous views of a specific earth location at any time of a day. This allows the production of a composite image to monitor short-term weather better. This paper describes a technique that uses multi-spectral infrared composite images of FY-2C to estimate particles emission and recognize fog at night. Radiations of particles detected by FY-2C at different wavelengths are analyzed combined with solar spectral irradiance. Having several spectral bands makes the analysis algorithms more complex and inefficient, thus it is important to choose the most respective bands. By applying Karhunen-Loeve transform to raw data of FY-2C, the infrared images are analyzed. By comparing Eigen image of these infrared images with visible image in the same batch, it is concluded that data of IR3 contribute to the first Eigen image mostly, which shows that the newly added IR3 channel of FY-2C has greatly improved the ability of distinguishing short time weather phenomena. Producing composite images by calculation and analysis at sequential period of time can clearly show changes of fog coverage. The improvement of the geostationary satellite instruments that have come to pass will encourage more widespread use of these derived products in the coming years.