A fine extraction method for tidal channels with fully polarized SAR and optical remote sensing
The tidal channel system is the most active geomorphic unit in the chalky-silt tidal flats. Due to the influence of periodic tidal erosion,human activities,and sea level rise,it is challenging to monitor tidal channel in a large scale. In this study,a method of tidal channel detection and extraction based on C-band GF-3 fully polarized synthetic aperture radar (SAR ) and PlanetScope multispectral remote sensing data is proposed. Through the fusion of spectral,index,polarization,texture and other features,the optimal feature set was constructed,and the maximum likelihood method,support vector machine and random forest algorithm were combined to carry out synergetic classification,and the fine distribution information of the tidal channel at the Yellow River estuary with 3 m resolution was obtained. The results show that the overall accuracy of the method is 99％,F1 value is 0. 98,and the extraction result is better than that of a single data source. This study is expected to provide a cost-effective alternative for the tidal channel mapping in estuarine and coastal areas,and help to quantitatively describe the morphological evolution,stability and driving factors.
万方数据
维普
