以南京紫金山国家森林公园为研究区域，以2011年90个典型样地调查数据和同期HJ-1 A环境小卫星遥感影像为信息源，分别采用归一化植被指数（ NDVI）参数反演、克里金内插和Maxent生态位建模三种方法估算紫金山2011年森林蓄积量，在获得预测值的基础上通过相关系数法、ROC曲线法和计算相对误差比较三种方法预测精度。结果表明：NDVI参数反演建立的最优拟合方程R2为0．788，相关系数为0．85，克里金内插相关系数为0．76，Maxent生态位模型相关系数为0．65；NDVI参数反演AUC曲线值为0．959、克里金内插为0．859、Maxent模型为0．787；NDVI参数反演相对误差为0．18、克里金内插为0．32、Maxent模型为0．49。 NDVI参数反演与克里金内插和Maxent建模相比，蓄积量的估算精度较高。%In Nanjing Purple Mountain National Forest Park the field survey data from 90 typical plots in 2011 and remote sens-ing image of HJ-1A satellite in the same year were gathered as the major information source.Three stock volume estimation methods of normalized differehtial vegetation index NDVI parameter inversion , Kriging interpolation and Maxent niche mod-eling were used to estimate the forest stock volume of the park in 2011 , and calculate three valuation index of correlation coefficient, ROC curve and relative error tocompare the performance of three methods.R2 of the optimal fitting equation of NDVI parameter inversion was 0.788, the correlation coefficient of the inversion method was 0.85, the correlation coeffi-cient of Kriging interpolation was 0.76 and Maxent modeling was 0.65.AUC of NDVI parameter inversion was 0.959, Kriging interpolation was 0.859, and Maxent modeling was 0.787.Relative error of NDVI parameter inversion was 0.18, Kriging interpolation and Maxent model were 0.32 and 0.49, respectively.Therefore, NDVI parameter inversion can per-form best among three estimation methods and provide scientific basis for the selection of stock volume estimation method of urban scenic forest.
展开▼
