基于GF-1影像的黄河口潮滩高程定量反演

现行黄河口为1996年黄河人工改道清8汊后形成的新生陆地，其沙嘴前缘为淤泥质潮滩且具有难抵达性，潮滩高程数据缺乏。基于国产GF-1/WFV卫星影像，利用支持向量机(Support vector machine, SVM)的监督分类和Sobel边缘算法，提取2014年8月至11月不同时刻的水边线，运用Delft3D水动力模型模拟对应时刻的瞬时潮位高程，从而反演得到黄河河口区潮滩数字高程模型(Digital elevation model, DEM)。与同期实测高程数据比较结果显示，模拟的潮滩DEM与实测DEM之间的误差集中在-38~38 cm的范围内，平均误差cm。表明遥感反演潮滩高程是对实测资料欠缺的一种有效补充，可为开展潮滩冲淤演变研究提供高程数据支持。

QUANTITATIVE RETRIVAL OF TIDAL FLAT ELEVATION WITH GF-1 IMAGES IN THE YELLOW RIVER MOUTH

The active Yellow River mouth is an emerging landform formed after the artificial diversion of the river channel to the Qing 8 in 1996. The tidal flat there is silty and difficult to reach. As the results, tidal flat elevation data is lacking. This study used a method that may construct tidal flat elevation with the Delft3D hydrodynamic model and waterlines. The waterlines from August to the end of November of 2014 were extracted using the SVM supervised classification and Sobel edge detection from the domestic GF-1/WFV satellite images, and the corresponding tidal levels were simulated with the Delft3D hydrodynamic model. Elevations were assigned to the waterlines, thus the simulated digital elevation model was obtained. Compared with the measured data, it is observed that the errors between the simulated DEM and the measured DEM fall in a limited range of -38 to 38 cm, with a mean less than 10 cm. This method may provide effective data support to the study of tidal flat morphological evolution in the areas lacking of measured data.