This study applies an edge detection method to satellite sea surface temperature (SST) images in order to distinguish SST fronts frequently occurring near the Japanese coast. First, we develop a new set of edge detection algorithms. The method is based on an entropic approach using Jensen–Shannon divergence, which is certified as a robust technique for detecting edge pixel candidates. Additionally, we incorporate ridge-line extraction morphological filter to obtain edge pixels, which is suited to delineate intricate frontal features. The methodology allows us to detect reasonable amount of possible fronts to the fullest extent according to the original image resolution. Therefore, it enables us to analyze finer-scale fronts near the coast as well as larger-scale fronts associated with major currents. Then, we present four case study examples of SST fronts near the Japanese coast by applying the method to Advanced Very High Resolution Radiometer (AVHRR) SST images with a grid size of 0.01°. We describe the configuration and seasonal variability of the SST frontal systems from the viewpoint of occurrence probability, SST gradient magnitude, and orientation to the local bathymetry as compared with the previous observational studies. Consequently, it is demonstrated that application of the method can derive new views of finer-scale and highly variable SST fronts. Moreover, this study first verifies the capability of comprehensive and statistical description of SST front configurations and variations near the Japanese coast by applying edge detection method to sequences of satellite SST images.
Read full abstract