基于SNPP/VIIRS卫星数据的海温锋检测算法研究OA北大核心CSTPCD

In this thesis,the sea surface temperature(SST)data of SNPP/VIIRS(Suomi National Po-lar-Partnership/Visible Infrared Imaging Radiometer Suite)are used to compare and analyze the front detection algorithms,including Sobel algorithm,Canny algorithm,the gravity model algorithm,the histogram analysis algorithm and the entropy algorithm.Through the difference and characteristics of the intermediate results,the one-day SST front detection results and the monthly frontal probability dis-tribution of the above algorithms,combined with the characteristics of each algorithm,it is found that,because of the different underlying logic of each algorithm,it is difficult to unify the definition of SST front in front recognition.Therefore,the SST fronts detected by the algorithms have different characteristics and are suitable for different research and application needs.In practical application,the appropriate algorithm should be selected according to the research needs and the actual situation.The characteristics of each algorithm are as fol-lows:Canny algorithm,the gravity model algorithm,Sobel algorithm and the entropy algorithm have decrea-sing ability to detect strong fronts and increasing ability to detect weak fronts,and the anti-noise ability is de-creasing.The fronts detected by the histogram analysis algorithm are smooth,clear and coherent,and the anti-noise ability is good.The occasions suitable for each algorithm are:Sobel algorithm or other algorithms based on gradient calculation are suitable for the application scenario of accurately calculating the degree of temperature change,Canny algorithm is suitable for the study of the strong fronts,the gravity model algorithm is suitable for detecting fronts with noisy data or strong fronts,and the entropy algorithm is suitable for the detection of the weak fronts.The histogram analysis is more suitable for the analysis of continuous long fronts.