利用LAI-2000的观测数据与基于HJ卫星遥感数据生成的植被指数,结合3种常用的回归模型,构造了夏玉米分别按全生育期、不同生育阶段和阈值分段的叶面积指数(Leafareaindex,LAI)反演模型;获取了3种模式下LAI的最优反演模型;在验证和评价各模型可靠性之后,生成了夏玉米在营养生长期、抽雄期和生殖生长期的LAI分布图;并将基于HJ影像反演得到的LAIHJ与MODISLAI产品(MOD15A2)LAIM进行了对比。研究发现,与各种通用植被指数相比,针对HJCCD数据构造的环境植被指数HJVI与LAI的相关性在3种反演模式中均为最佳。HJVI与全生育期LAI的相关性达到0.875,在不同生育阶段与LAI的相关性也高于其他植被指数(营养生长期线性模型最佳,决定系数为0.769;抽雄期对数模型最佳,决定系数为0.783;生殖生长期指数模型最佳,决定系数为0.703)。普适性植被指数中,OSAVI适用于夏玉米生长前中期的LAI反演,NDVI适用于夏玉米生长后期的LAI反演。在夏玉米全生育期内,各植被指数与LAI的相关性整体较高,但最优回归模型出现在按不同生育阶段反演的模式中。LAI小于3时EVI为精度最佳指数(决定系数为0.358),LAI不小于3时OSAVI为精度最佳指数(决定系数为O.515)。在夏玉米3个生育阶段,LAIM与LAIHJ的相关性分别达到0.732、0.761、0.661。HJ遥感数据具有较强的LAI反演能力,其高时间和高空间分辨率的特征可以使其代替传统的中分辨率遥感数据而成为农业遥感研究的重要数据源。
The observation data of LAI - 2000 and vegetation index was generated by satellite remote sensing data of H J, combining three kinds of commonly used regression model. LAI ( Leaf area index) inversion model was constructed according to growth period, growth stage and threshold boundaries of summer maize, respectively. The optimal LAI inversion model was acquired based on the above three modes. The summer corn LAI scatter grams of the vegetative growth period, the tasseling stage as well as the reproductive stage were generated after verification and evaluation of the model reliability. The productions of MODIS LAI (MOD15A2) were verified by LAIHJ based on the inversion of model HJ image. According to the survey, except HJVI, during the whole growth period of summer maize, a linear model of RVI with LAI was regarded as the best fitting model ( R^2= 0. 662) ; during the vegetative growth period, a linear model of OSAVI with LAI was regarded as the best fitting model (R^2 = 0. 724) at the tasseling stage, index model of OSAVI with LAI was regarded as the best fitting model(R^2 = 0. 749) ; at the reproductive stage, a linear model of NDVI with LAI was regarded as the best fitting model(R^2 =0. 700). The correlation of HJVI and LAI at the growth period achieved to 0. 875, and the correlation at different growth stages with LAI is higher than the other vegetation indexes (during the vegetative period, R^2 =0. 769; at the tasseling stage, R^2=0. 783; at the reproductive stage, R2 = 0.703). EVIisthe best index when LAI is less than 3 (R^2 =0.358), while OSAVI is the best when LAI is more than 3 (R^2 = 0. 515). During the three reproductive periods, R2of LAIM and LAIm is 0. 732, O. 761 and 0. 661. Conclusions were drawn: the inversion method of LAI at different stages is optimal. HJVI shows obvious advantage for LAI inversion ability. The production of MODIS LAI could be used for crop monitoring in special situation. The study not only broadens the mode of inversion LAI using vegetati