一种多源数据确定内陆盐沼湿地自然边界的方法

A Method for Determining Natural Boundary of Inland Salt Marsh Wetland with Multi-source Data

以宁夏盐池县为例,以Landsat 8影像、高分二号影像、激光雷达(light detection and ranging,LiDAR)数据插值生成的高精度数字高程模型(digital elevation model,DEM)数据、地理国情普查数据等为数据源,探讨了多源数据确定内陆盐沼湿地自然边界的方法。首先,利用一年多期的Landsat 8影像数据确定湿地边界提取的最佳时相,对高分二号影像等数据进行多尺度叠置分割,得到分割对象;然后,进行特征提取,并利用优化后的分离阈值(improved separability and thresholds,ISEaTh)算法进行特征选择;最后,运用最邻近分类器获得湿地的自然边界,并分析了不同高空间分辨率的DEM数据对内陆盐沼边界定位的影响。结果表明,高空间分辨率DEM数据参与提取内陆盐沼湿地边界,可以有效减少内陆盐沼边界处湿地与非湿地的错分现象,提高其边界定位精度,且DEM数据的空间分辨率越高,提升效果就越显著。

Taking Yanchi County in Ningxia as an example,the images of Landsat 8 and Gaofen-2(GF-2),high-precision digital elevation model(DEM) data generated by interpolation of light detection and ranging(LiDAR) data and geographical conditions census data are used as data sources to discuss the method for determining the natural boundary of inland salt marsh wetland. First,we use multi-period Landsat 8 images of more than one year to determine the best phase for wetland boundary extraction, and we perform multi-scale overlay segmentation on GF-2 images to obtain the segmentation objects. Then,we extract features and select them by improved separability and thresholds(ISEaTh)algorithm. Finally,the nearest neighbor classifier is used to obtain the natural boundary of wetland and the influence of different high spatial resolution DEM data on the boundary positioning of inland salt marsh is analyzed. The results show that the participation of high spatial resolution DEM data can effectively reduce the misclassification of inland salt marsh and non-wetland,and improve the natural boundary location accuracy of wetlands.The higher the spatial resolution of DEM data is,the more significant the improvement effect would be.