Study On The Application Of Weber's Law And Time Serials In FY2C Cloud Detection

The paper studies the application of Weber's Law and Time Series in cloud detection with the FY2C data attained from China Satellite Data Service Center (CSDSC). The FY2C operational cloud detection algorithm is the parallel algorithm, through which dynamic threshold method is utilized together with other methods to conduct cloud detection. The analysis shows that, in the overall sense, the FY2C operational cloud detection could be rated as satisfactory. According to the data from China Surface meteorological observation stations (the same below), the accuracy reaches to 71.15% in January, 2007 and 75.77% in June, 2007 respectively. However, the FY2C cloud detection is still not adequate enough for the detection of low clouds and thin clouds, which remains as a difficult task for cloud detection in the world(especially at night). In addition to the current remote sensing techniques, the major factors influencing the accuracy of FY2C cloud detection are the time-space limitations of the threshold.In order to overcome the limitations of the threshold, the paper investigates the recognition principle of the human eyes toward boundaries—Weber's law and finds out the physical quantity that has no relationship with time-space—Weber's fraction. Theoretically speaking, there are multiple Weber's fraction curves on the cloud imagery, however, the paper points out that, under the variable background within the range of cloud brightness, there is actually only one curve. This finding is employed in the oceanic cloud boundary detection as well as the land cloud boundary detection, with the combination of the time sequences cloud imagery utilized in the latter field. Furthermore, the obtained cloud boundary could be used to acquire the clear threshold (The threshold method in the paper is clear background method) dynamically and to conduct cloud detection of the nominal imageries?. The statistics show that over China the accuracy reaches to 72.34% in January, 2007 and 80.05% in June, 2007 respectively. The method applies the human eyes' recognition principle towards boundaries into cloud detection, overcomes the time-space influence of the threshold to a certain extent, and improves the detection result of low cloud and thin cloud. Therefore, it has great significance in the research of cloud recognition and in enhancing the accuracy of operational cloud detection.The paper carries out the research of the application of the meteorological stationary satellite's time sequences in cloud detection. The results show that the hour range of brightness temperature (BT) of the cloud and the clear pixels in time series could be utilized to identify clouds that are developing rapidly or the ones that are located at the boundaries of the moving clouds. Similar to the application of Weber's fraction, the acquired clouds are used to dynamically obtain thresholds and cloud detection of the nominal imageries is conducted. In this way, the accuracy in China reaches to 72.89% in January, 2007 and 79.94% in June, 2007 respectively. Making use of the high time resolution of stationary satellites and the moving characteristic of clouds, the method improves the detection performance of low clouds and thin clouds and then is of high application value in the operational cloud detection of stationary satellites. In order to further modify the results of cloud detection, the paper carries out the cloud detection of nominal imageries by the combination of Weber's fraction and the hour range. The data from the stations verifies that in China the accuracy arrives at 73.67% in January, 2007 and 81.02% in June, 2007, which shows accuracy of the detection is increased.Applying Weber's law or time series to identify clouds is a new cloud detection method that dynamically obtains thresholds. It can effectively overcome the time-space limitations of thresholds, provides a simple and practical way of acquiring thresholds, and then has high application value.