作物长势遥感监测指标与作物产量有密切的关系，长势监测成为作物产量估测的必要前提，对于国家的粮食宏观调控有着重大的现实意义。通常的遥感监测主要采用年际间同期遥感影像对比分析，反映区域作物生长状况的相对差异，属于半定量的方法，未能结合影响长势差异的土壤养分的空间变异进行综合研究。鉴于此原因，开展了本论文的研究。 本研究于2005-2006年在北京市小汤山国家精准农业研究示范基地进行，主要以田块尺度的冬小麦为研究对象，利用田块尺度高分辨率的Quickbird卫星遥感影像配合地面农学参数，采用遥感技术、传统统计分析以及地统计学分析方法，研究了农田土壤养分、作物长势空间变异以及卫星图像之间的关系。研究内容涉及了作物长势参数反演模型的建立与验证，农田土壤养分与作物长势空间变异的关系分析，以及基于作物长势变异遥感监测的精准生产管理应用研究。主要内容及结论如下： （1） 尝试选取由实测的冠层光谱计算得到的各种植被指数，分别与实测的LAI建立线性回归模型，来反演LAI，探讨得出基于NDVI的一元线性回归模型为最优模型，经验证后模型拟合精度达到了0.7588。最后选此一元线性回归模型来进行卫星影像反演，得到实验区的LAI分布图，实现了作物长势监测。 （2） 运用地统计学原理研究土壤养分的空间变异，并借助高分辨率Quickbird遥感影像提取指示作物长势情况的归一化植被指数（NDVI），对土壤养分和冬小麦的NDVI值进行半方差分析，并利用Kappa系数评价了土壤养分与作物长势NDVI的一致性程度，确定土壤养分和小麦长势NDVI空间变异的相关性，得出土壤有效磷和有机质是小麦长势的决定因子。 （3） 在空间变异分析基础上，以高空间分辨率Quickbird遥感影像得到的光谱指数、地面土壤养分采样数据和当年的产量数据进行田块尺度内管理分区划分研究。结果表明，利用土壤养分和光谱数据划分的管理分区内，各土壤养分的均一度得到了提高，光谱指数和产量的变异系数普遍降低。因此，在划分管理分区时，应该综合考虑土壤养分和光谱数据的空间变异情况，进行管理分区的划分。

The index of crop growth monitoring by RS have closely relation with crop yield. Meanwhile, the growth monitoring becomes the precondition of the yield estimation, which is important for the macro control of food. In the usual monitoring, we mainly analysis contemporaneous RS images among different years in order to reflect the relative difference of the crop growth in areas. This was the semi-quantitative method and it did not deeply study by combing the soil spatial variability which influenced the growth difference. For this, this study was developed in the dissertation. The experiment was carried out on the National Precision Agricultural Research Basement of Xiao Tangshan, Beijing from 2005 to 2006. The winter wheat was the mainly f study object in the experiment. The relation among the soil nutrient、the spatial variability of the crop growth and the satellite image was studied by combining the high-resolution Quickbird image in filed scale and agronomic parameter on the ground based on many methods such as the remote sensing technology、traditional statistics and geo-statistics. The study referred to many aspects: the establishment of the inversion model for the growth parameter and its validation, the relation of the soil nutrient and growth spatial variability, the precision production and management based on the growth variability. The contents and results are summarized as below: (1) The various vegetation index was selected which were caculated from canopy spectra in actual measurement. The linear regression model was established for the vegetation index and LAI from which the LAI. could be inversed. And then, the simple linear regression model based on NDVI was the best model, and the fitting precision reached to 0.7588 after the validation. At last, this best model was chosen to inverse the image and the figure for the distribution of the LAI could be got. Therefore, the crop growth could be monitored in large area. (2) The spatial variation of soil nutrients was studied by the geo-statistics. The presented study tried to calculate the NDVI (Normalized Difference Vegitation Index) values from the Quickbird imagery as the indicator of wheat growth. After the semivariogram analysis, the Kriging method was used to interpolate the spatial distribution map for each soil nutrient, then Kappa coefficients was calculated to evaluate the consistency between soil nutrients and NDVI. And then, the relation between the soil nutrients and the spatial variability of the wheat growth could be got. So the soil AP and OM contents were the determinative factors for wheat growth in the studied field. (3) On the basis of the spatial variability, the study took much data to divide the management zone. The data were as follows: the spectral index got from Quickbird imagery, soil sampling data of the study area ,the yield data in the present year. The result showed that the homogeneous degree of the soil nutrient was increased, the variability of spectral index and the yield were reduced in the management zone divided by soil data. As a result, the spatial variability of the soil nutrient and spectrual data should be taken into consideration when dividing the management zone. We could get satisfactory division result by using the soil nutrient