基于快鸟影像树冠特征的天山云杉材积反演1）

摘要

We used object-oriented method to extract and analyze the crown of Picea schrenkiana with the high resolution remote sensing QuickBird in low stand density area in Guozigou Forestry Station .With field investigation data , we tested the accu-racy of tree-crown, and established the model between the investigated crown data with diameter at breast height and the estimated crown, and calculated the tree volume.The crown interpretation accuracy of producer reached 91.47%, while the user accuracy reached 90.9%.Correlative models are built according to the field measured data .By correlative analysis there was significant positive correlation between extracted tree-crown area and diameter at breast height , with R2of 0.7533.The the accuracy of extracted tree crown was 89.17%, the estimated accuracy of DBH was 87.91%, and the estimated accuracy of volume was 87.36%.%基于QuickBird高分辨率遥感影像，利用面向对象分类法提取了果子沟林场实验区的低林分密度的天山云杉的树冠信息，以实测数据为真值检验遥感方法提取的树冠大小的精度；在此基础上建立了基于遥感影像树冠与实测胸径的关系模型，利用模型估计的胸径和一元材积模型，计算云杉单木材积。结果表明：研究区内天山云杉的树冠遥感方法分类的生产者精度为914．7％，用户精度为90．9％。基于实测胸径和提取冠幅面积所建立的回归方程的决定系数R2＝0．7533，树冠面积提取精度为89．17％，胸径的预估精度为87．91％，材积的预估精度为87．36％。