Hurricane Eye Morphology

Among all kinds of natural disasters, hurricanes are regarded as one of the most destructive, which can cause tremendous losses to the global economic system and ecosystem every year. However, until now, there remain many issues unknown about hurricane dynamics, while hurricanes undergo amplification, shearing, eyewall replacements, diurnal influences, and so forth. Precise morphology parameters, extracted from high-resolution spaceborne Synthetic Aperture Radar (SAR) image, can play an essential role in further exploring and monitoring the hurricane dynamics. Moreover, these morphology parameters may help to build guidelines for the wind calibration of the more plentiful, but lower resolution scatterometer wind field data in hurricane events in order to better link scatterometer wind fields to hurricane categories. In this paper, we have developed a new method for extracting the hurricane eyes from C-band SAR data by constructing Gray Level-Gradient Co-occurrence Matrix (GLGCM) for each image. The hurricane eyewall (HE) area is determined with a 2-dimensional vector, which is automatically generated by maximizing the conditional entropy of HE area in GLGCM. Subsequently, we select the HE pixels based on minimizing the variance of normalized radar cross-section (NRCS) values of the pixel set chosen. The texture information of HE can be adequately preserved in this process. The experimental results prove the effectiveness of our method. Notably, the HE extracted with this automatic method is still in line with the visually observed eyewall even when the hurricane is weak or the eyewall is unclosed. Compared with the morphological analysis and wavelet analysis methods proposed in other papers, the approach developed here is able to accomplish in a simpler way with equally satisfying results. In conclusion, this study can provide a new choice for hurricane eye morphology extraction.