数字林业是林业现代化的基础。本研究以河北省塞罕坝机械林场为研究区,利用遥感科学与技术,基于机载数据(激光雷达、高分辨率CCD影像和高光谱影像),研究单木分割和单木分类方法,实现了机载飞行区域270km2范围内所有单木的分割和分类。结合地面样方调查数据,单木分割匹配度精度可以达到0.6以上,研究区内四种典型树种单木分类精度均能达到97%以上。基于地基激光雷达扫描数据,研究以单木为单位,通过枝干重建和叶片添加,实现单木三维模型重建和场景重建。该方法有利于分析从器官(枝干、叶片)-单木-区域多尺度森林结构特征。在此基础上,本研究探索了以单木为基础的林业数字化方法,并结合WebGIS技术开发了数字林业系统,初步实现了研究区相关数据的存储、管理、查询、分析和可视化等功能,为后续的森林经营管理和决策规划提供了重要的基础数据。

Objective: Digital forestry is the foundation of forestry informationization and modernization, which can help the managements with the scientific programming, scientific management and scientific decision. With the development of 3S (GIS, RS and GNSS) technology, it is feasible to improve the level of digital forestry, specially for the monitoring scale from forest-compartment level to single-tree level by using both high spatial- and spectral-resolution remote sensing data and LiDAR (Light Detection and Ranging). Our purpose is to develop a forestry digital system by combining the advantages of remote sensing and GIS technology, which can improve the precision management ability of forest farm. Method: In this study, Saihanba Mechanical Forest Farm in Hebei Province, China is taken as the research area, we use airborne data, including LiDAR, CCD image and hyperspectral image, to study single tree segmentation and classification methods. First, an algorithm combined water-shed and CCE (Connection Center Evolution) methods is applied to segment each single tree based on the ALS (airborne LiDAR scanner). Then random forest algorithm is applied to classified the tree species of each single tree based on the combination of the 50 characteristics from ALS, CCD and hyperspectral image. This paper also studies the 3D model reconstruction by using TLS (terrestrial laser scanning) data, with the procedure of branch reconstruction and leaf addition. Forest scenarios are reconstructed based on each single tree model. Finally, WebGIS technology is used to develop a forestry digital system with the functions of data storage, management, query, analysis and visualization. Result: More than 9706 thousand single trees within the airborne flight area of 270km2 are segmented and classified. Combined with the survey data of ground quadrats, the single tree segmentation accuracy can reach over 0.6 matching rate, and the single tree classification accuracy of four typical tree species in the study area can reach more than 97%. Conclusion: The information of single tree, including location, tree height, crown breadth and the structures of branches and leaves are extracted based on airborne and field data, which can help us analyze the multi-scale forest characteristics from organs (branches and leaves) - single tree – region scales. On this basis, this study explored the forestry digitization method based on single tree information, and developed a digital forestry system using WebGIS technology. The system preliminarily finished the functions of storage, management, query, analysis and visualization of relevant data in the study area, providing reliable basic data for forest management and decision-making planning.