Research On Sea Fog Detection Method Based On Active And Passive Remote Sensing

The yellow sea and bohai sea is one of the most important sea areas for maritime transportation and development in China.The occurrence of sea fog has a serious impact on sea navigation and the life of coastal residents,and even threatens the safety of people's lives and property.Therefore,the research on sea fog detection methods is of great significance to the safety of sea areas and human production activities.Compared with the traditional sea fog detection methods,remote sensing technology can achieve a wide range,all-weather,dynamic monitoring.In this paper,based on satellite-borne lidar data and hyperspectral remote sensing data,the active and passive remote sensing method for sea fog detection is studied to realize the complementary advantages of the two remote sensing technologies,providing a practical and effective solution for sea fog detection.The main work and conclusions of this paper are as follows:(1)The sample database based on MODIS is established.First,CALIOP data for the decade from 2008 to 2018 are analyzed.Using L1 and L2 data of CALIOP,sea fog,low cloud,medium and high cloud and sea surface sample points based on CALIOP are obtained according to height information and attenuation backscatter coefficient.A total of 36 images were obtained by matching CALIOP data with MODIS data,and a MODIS based sample database was established,including 21,606 sea fog sample points,10,542 low cloud sample points,8,649 high cloud sample points,and 12,534 sea surface sample points,providing a sample data basis for the establishment of MODIS based sea fog detection algorithm.(2)The sea fog detection algorithm based on MODIS is established.According to the sample database,analysis the sea fog and other features of the spectral characteristics and the reflectivity and emissivity of the distribution of each band,come to a conclusion: 1,2 band difference can effectively eliminate the image of the sandbank,3,5 band can better identify the sea fog and clouds,17 bands can be in as far as possible,do not affect the sea fog discriminant out under the condition of sea surface,26,32 bands to separate the sea fog and high clouds has good effect,thus established based on MODIS 1,2,3,5,17,26,32 band of sea fog extraction algorithm,the algorithm of the sea fog recognition rate is 88.24%.The new algorithm is compared with support vector machine,CART decision tree and BP neural network,and the conclusion is obtained: the overall accuracy of the new algorithm is 89.56%,and the Kappa coefficient is 0.91.The overall accuracy of SVM was 86.27% and the Kappa coefficient was 0.82.The overall precision of CART decision tree was 88.93%,and the Kappa coefficient was 0.89.The overall accuracy of BP neural network was 78.95%,and the Kappa coefficient was 0.76.Therefore,the new algorithm has the highest accuracy and the best identification effect.(3)Spatial and temporal characteristics of sea fog in the yellow and bohai sea.By using the established sea fog detection algorithm,sea fog identification processing was performed on the daily MODIS image data of the yellow sea and the bohai sea in 2016,2017 and 2018,and the time and space characteristics of the sea fog distribution in these three years were analyzed.From June to September and October to November,the sea fog in the yellow sea reaches the climax of the outbreak of sea fog,and in the space from east to west,from south to north diffusion,from west to east,from north to south dissipation characteristics.