应用在江苏省东台林场拍摄的全天空相片,通过建立基于RGB照片的分类模型将相片中树叶、树干和天空分离,从而达到精确提取林冠郁闭度的目的。结果表明,该方法的总体分类精度达到0.94,Kappa系数为0.89,分类精度较高,且在主干部分的区分上效果良好,总体分类精度达到0.94,Kappa系数为0.84。在低郁闭度下相片的计算精度高于高郁闭度相片,这与拍摄时的环境条件有关。将模型估测结果与抬头望法结果对比,得出两者的R²为0.77,在郁闭度较低时模型估测结果大于目测结果,在郁闭度较高时模型估测结果小于目测结果。此外,两者都显示14a生杨树林郁闭度高于9a生杨树林郁闭度,具有较好的一致性。

In this study,the whole sky photos taken at the Dongtai research site are used to separate leaves,trunks and sky by establishing an RGB classification model to extract the canopy closure accurately.It shows that the overall classification accuracy of this method is 0.94 and the Kappa coefficient is 0.89.The classification accuracy is high and it works well in distinguishing the main trunk of the forest.The overall classification accuracy reaches 0.94 and the Kappa coefficient is 0.84.In addition,this study finds that the calculation accuracy of photos with low canopy closure is higher than photos with high canopy closure.In comparison of the model estimation results with ocular estimate results,the R² is 0.77.The model estimate result is higher than the ocular estimate result when canopy closure is low,and it is lower than the ocular estimate result when canopy closure is high.In the comparative study of the canopy closure of two different age poplar forests,the results indicate that the canopy closure of the 14-year-old poplar is higher than that of the 9-year-old poplar.The p values of the two methods are less than 0.05 and 0.001,respectively.Using digital photos to obtain ground sample data accurately can further verify the forest canopy closure value retrieved by large-scale remote sensing.It has great application value in further improving the measurement accuracy of forest canopy closure on a large scale.