Remote Sensing Monitoring of Sea Fog in the South China Sea Based on FY-4 Satellite Data
Based on the observation data of the advanced geosynchronous radiation imager(AGRI)carried by the new generation geostationary meteorological satellite FY-4A and the visibility data of the ground automatic weather station,we carried out the remote sensing detection of sea fog in the South Chi-na Sea through different combinations of visible channel,near-infrared channel and mid-far infrared chan-nel data.In this article,in view of the different characteristics of sea fog from underlying surface and cloud in remote sensing image,the least square method is used to dynamically fit and select the optimal degree to realize the separation of sea fog from underlying surface.The method of cloud top height thresh-old is used to separate sea fog from mid-and high-level clouds.The method of texture feature recognition and sea fog detection index is used to separate sea fog and low-level cloud in daytime.The method of thin low-level cloud detection index is used to separate sea fog and low-level cloud during the night.What's more,the sea fog identification results are compared with the ground observation data along the Guang-dong coast in 2018.The results show that:(1)the method of using the least square dynamic fitting to se-lect the optimal degree has a good effect on the distinction between sea fog and underlying surface,and the method of using texture feature recognition and sea fog detection index has a certain improvement on the distinction between sea fog and low-level cloud.(2)The sea fog monitoring algorithm has the highest recognition accuracy of 85.45％for the sea fog in March with the lowest false alarm rate of 20.94％,and the lowest recognition accuracy of 58.70％for the sea fog in May,but the highest false alarm rate of 30.95％for the sea fog in January.The average accuracy of the sea fog monitoring algorithm was 76.37％and the average false alarm rate was 26.08％.
FY-4A satellitesea fogthe South China Searemote sensing monitoring
万方数据
维普
