应用无人机可见光数据对胡杨林群落单木信息的提取  

Vertical Structure of Populus euphratica Forest Community with UAV Visible Light Data

作　　者：张文奇 李丹 师庆东[3] 郭玉川[4] Zhang Wenqi;Li Dan;Shi Qingdong;Guo Yuchuan(Chinese Academy for Environmental Planning,Beijing 100043,P.R.China;CNOOC Research Institute Co.Ltd;Xinjiang University)

机构地区：[1]生态环境部环境规划院,北京100043 [2]中海油研究总院有限责任公司 [3]新疆大学资源与环境科学学院 [4]新疆大学

出　　处：《东北林业大学学报》2022年第11期57-64,共8页Journal of Northeast Forestry University

基　　金：国家自然科学新疆联合基金重点项目(U1703237)。

摘　　要：为探究无人机可见光数据获取落叶阔叶林单木信息的可行性,以达里雅布依绿洲的胡杨林为研究对象,无人机数据衍生的冠层高度模型(CHM)为数据源,采用“自适应窗口最大值算法”与“冠层最大值模型”相结合的方法识别单株树木,提取胡杨的高度信息并用Clark-Evans最近邻体分析方法计算胡杨空间分布格局。结果表明:(1)自适应窗口最大值算法下的单木识别结果准确率在80%左右,查全率在90%左右,探测率在93%~131%,探测点与实测点的偏差均值在0.82~0.35 m;(2)无人机树高估测值与实测值有较高的相关关系,决定系数(R^(2))0.76~0.83,均方根误差为0.54~1.02 m;(3)实测所有样地的胡杨分布格局均为随机分布,而无人机测量结果为3号样地呈均匀分布,其余样地与实测结果一致。本试验为监测大面积胡杨林资源提供了一种更快的地面调查方法,同时保持数据的准确性,也为可重复、低成本的无人机森林调查提供了可能性。To explore the feasibility of acquiring single tree information of deciduous broad-leaved forest by UAV visible light data,we took Populus euphratica forest in Daryabui Oasis as the research object,and the Canopy Height Model(CHM) derived from UAV data as thedata source,using the method of combining “adaptive window maximum algorithm” and “canopy maximum model” to identify individual trees,extract the height information of P.euphratica and calculate the spatial distribution pattern of P.euphratica using the Clark-Evans nearest neighbor analysis method.The accuracy of individual tree recognition results under the adaptive window maximum algorithm is about 80%,and the recall rate is about 90%,the detection rate is between 93% and 131%,and the average value of the deviation between the detection point and the actual measurement point is between 0.82 and 0.35 m.The estimated tree height of UAV has a high correlation with the measured value,and the maximum coefficient of determination R^(2) is 0.76 while the minimum is 0.83 and the root mean square error(RMSE) is 0.54-1.02 m.The distribution pattern of P.euphratica in all sample plots measured is random distribution,while the UAV measurement result shows that the sample P03 is evenly distributed,and the other sample plots are consistent with the actual measurement results.This experiment would provide a faster ground survey method for monitoring large area P.euphracea forest resources,while maintaining the accuracy of the data,and also provides the possibility for repeatable and low-cost UAV forest surveys.

关 键 词：无人机 胡杨 冠层高度模型 自适应窗口最大值 

分 类 号：S771.8[农业科学—森林工程]