【摘要】 植被覆盖度(VFC)的定量遥感是多种地表过程研究的迫切需要。文章选用南京市一幅Landsat7ETM+图像,经大气校正后提取了归一化植被指数(NDVI),与地面实测的植被覆盖度进行回归分析,建立了1～4次多项式关系模型。结果表明,NDVI与VFC呈极显著的正相关关系(r=0.874,P<0.001)。在NDVI-VFC的1～4次多项式关系模型中,模型幂次越高,拟合程度越好。综合考虑模型的精度和稳定性,3次多项式模型作为最优模型推荐使用:VFC=-1.3438NDVI3+0.9774NDVI2+0.9988NDVI+0.1507(R2=0.7961,RMSE=0.1094),该模型精度在植被中等密集区域(VFC=0.4～0.8)最高,植被稀疏区域(VFC<0.4)最低,植被密集区域(VFC>0.8)居中。模型可直接用于全图像的VFC计算,并可通过植被指数的校准,进行推广使用。更多还原

【Abstract】 Quantative remote sensing of Vegetation Fractional Coverage(VFC) is needed by various studies on land surface process.A Landsat 7 ETM+ image of Nanjing was used to derive normalized difference vegetation index(NDVI) after atmospheric correction.Regressed NDVI with VFC from ground measurement,linear,quadratic,cubic,and quartic polynomial models were established.The results showed that,NDVI was positively correlated with VFC,no matter what the vegetation type was.Comparisons of the 4 models indicated that,the bigger the exponential power,the more robust models would be.Considering the accuracy and stability of the models,the cubic polynomial model was the best and recommended for use:VFC =-1.343 8 NDVI 3 + 0.977 4 NDVI 2 + 0.998 8 NDVI + 0.150 7(R2 =0.796 1,RMSE =0.109 4).The accuracy of this model was the highest on middle-densely vegetated areas(VFC=0.4～0.8),lowest on sparsely vegetated areas(VFC<0.4),and middle on densely vegetated area(sVFC>0.8).The model could be used for VFC calculating on the whole ETM+ imagery of the studied area,and on other places after calibration of the vege-tation index.更多还原