马尾松LAI与植被指数的相关性研究

The Correlation between LAI and Vegetation Index of Masson Pine

以福建省永安市区为研究区,计算IRS-P6(LISS-Ⅲ)多光谱数据的DVI、EVI2、MSAVI、NDVI、RDVI、RVI及TNDVI等7种植被指数,并与使用LAI-2000测量的马尾松叶面积指数(LAI)建立相关关系,分析植被指数对马尾松LAI的影响。从决定系数(R2)和标准误差两个方面对基于不同植被指数的LAI反演模型进行定量分析,反演模型包括线性模型、二次曲线模型、幂函数曲线模型和指数曲线模型4种。结果表明,马尾松LAI与植被指数呈指数曲线相关或幂函数曲线相关。反演马尾松LAI,最佳的统计模型是指数曲线模型和幂函数曲线模型,较佳植被指数为TNDVI、NDVI和RVI,其指数曲线模型和幂函数曲线模型拟合的R2均高于0.76,且验证结果R2均高于0.84,但RVI指数反演的模型标准误差相对较大。总体而言,TNDVI和NDVI的指数曲线和幂函数曲线模型对马尾松LAI具有较好的预测性。

With Yongan City of Fujian Province as the study area, the authors investigated the best VIs for estimating LAI of masson pine. Several VIs were calculated from the IRS - P6 （ LISS - Ⅲ） image, which included DVI, EVL2, MSAVI, NDVI, RDVI, RVI and TNDVI. The correlation between the measured LAI of masson pine using LI - COR LAI - 2000 and VIs was established, and the effects on the LAI of masson pine were studied. The LAI estimation models based on different VIs were quantitatively analyzed with both R2 and standard error. The estimation models included linear model, quadratic curve model, exponential curve model and power curve model＋ The results show that there exists curvilinear correlation （exponential correlation or power correlation） between selected VIs and LAI of masson pine. The exponential curve model and the power curve model constitute the best inversion models, and TNDVI, NDVI and RVI are fairly good for inversing LAI of masson pine, in which the R2 of the exponential curve model and the power curve model are all larger than 0.76 and their verification R2 are all larger than 0.84, but the standard errors of RVI＇s inversion models are much larger than those of the other two models. In general, both the exponential curve model and the power curve model of TNDVI and NDVI can yield good results in estimating LAI of masson pine.