Providing vegetation type information with accurate surface distribution is one of the important tasks of remote sensing of the ecological environment. Many studies have explored ecosystem structure information at specific spatial scales based on specific remote sensing data, but it is still rare to extract vegetation information at various landscape levels from a variety of remote sensing data. Based on Gaofen-1 satellite (GF-1) Wide-Field-View (WFV) data (16 m), Ziyuan-3 satellite (ZY-3) and airborne LiDAR data, this study comparatively analyzed the four levels of vegetation information by using the geographic object-based image analysis method (GEOBIA) on the typical natural secondary forest in Northeast China. The four levels of vegetation information include vegetation/non-vegetation (L1), vegetation type (L2), forest type (L3) and canopy and canopy gap (L4). The results showed that vegetation height and density provided by airborne LiDAR data could extract vegetation features and categories more effectively than the spectral information provided by GF-1 and ZY-3 images. Only 0.5 m LiDAR data can extract four levels of vegetation information (L1–L4); and from L1 to L4, the total accuracy of the classification decreased orderly 98%, 93%, 80% and 69%. Comparing with 2.1 m ZY-3, the total classification accuracy of L1, L2 and L3 extracted by 2.1 m LiDAR data increased by 3%, 17% and 43%, respectively. At the vegetation/non-vegetation level, the spatial resolution of data plays a leading role, and the data types used at the vegetation type and forest type level become the main influencing factors. This study will provide reference for data selection and mapping strategies for hierarchical multi-scale vegetation type extraction.

中文翻译：

---

基于多源遥感数据的分层地理对象植被类型提取

为植被类型信息提供准确的表面分布是生态环境遥感的重要任务之一。许多研究已经基于特定的遥感数据在特定的空间尺度上探索了生态系统的结构信息，但是从各种遥感数据中提取不同景观级别的植被信息仍然很少。基于高分1号卫星（GF-1）宽视场（WFV）数据（16 m），紫园3号卫星（ZY-3）和机载LiDAR数据，本研究通过以下方法对植被信息的四个级别进行了比较分析：使用基于地理对象的图像分析方法（GEOBIA）对东北典型的天然次生林进行了研究。植被信息的四个级别包括植被/非植被（L1），植被类型（L2），森林类型（L3）和林冠和林冠间隙（L4）。结果表明，机载LiDAR数据提供的植被高度和密度比GF-1和ZY-3图像提供的光谱信息更有效地提取植被特征和类别。只有0.5 m的LiDAR数据可以提取四个级别的植被信息（L1-L4）；从L1到L4，分类的总准确度依次下降了98％，93％，80％和69％。与2.1 m ZY-3相比，2.1 m LiDAR数据提取的L1，L2和L3的总分类准确率分别提高了3％，17％和43％。在植被/非植被级别，数据的空间分辨率起着主导作用，在植被类型和森林类型级别使用的数据类型成为主要的影响因素。