基于TM遥感影像的河口流态信息半定量化研究

Research on Information Semi-quantitation of Estuarine Flow Pattern Based on TM Remote Sensing Image

在结合潮位资料初步判断涨落潮情的基础上,利用Gabor滤波和纹理熵计算等纹理特征分析方法,从TM遥感影像中解译流线,提取流向,表征流速(分级),由此实现了流态信息的半定量化解译。并以长江口—杭州湾北岸Mike21_FM潮流数学模型的模拟结果对该方法的试验结果进行了验证,得出TM遥感解译的典型涨、落潮流的平均流向均方根误差(RMSE)为9.33°,平均的流速大小分级吻合度为73.1%。研究表明,基于纹理特征分析方法进行的河口流态信息遥感解译是可行的,该方法可以实现流态信息的半定量化,具有一定的创新性。

On the basis of preliminary judging the tide situation by tidal level,the streamline was interpreted,the flow direction was extracted and the flow velocity was represented from the Landsat TM images by the texture characteristic analysis method,such as the Gabor filtering and texture entropy calculation,then the semi-quantitative interpretation of the flow pattern was realized.The results were validated by the numerical model Mike21_FM and the average root-mean-square error（RMSE）of the flow direction of flood and ebb was 9.33°,the average fitness of flow velocity classification was 73.1%.It indicates that the remote sensing interpretation of the estuarine flow pattern by the texture characteristic analysis method to semi-quantitative flow pattern information is feasible and innovative.