부산대학교 도서관

Satellite X-band synthetic aperture radar (SAR) image is expected to be a promising data source for shoreline monitoring of sandy coasts. However, its imaging characteristics have not been clarified and a robust automatic shoreline extraction method has not existed. To achieve a better understating of X-band SAR imagery of sandy coasts, this study conducted a simultaneous observation by unmanned aerial vehicle (UAV) and satellite X-band SAR. It is found that a sandy beach can be clearly observed from the X-band SAR image. Vegetation on the beach can increase the intensity of the backscattering coefficient (BC) leading to a clear difference from that of the seawater. The BC of the nearshore sea surface is proportional to the magnitude of significant wave height. Azimuth ambiguity due to large wave height may lead to difficulty in successful shoreline detection from the X-band SAR images. A tentative convolutional neural network (CNN) based approach shows promise for the direct detection of the shoreline position from SAR images. A high-resolution shoreline extraction technique can be expected by gathering a sufficient amount of X-band SAR images and corresponding measured data of shoreline position.