中文摘要：

以北疆绿洲区棉田表层土壤为研究对象,利用国产HJ-1A/1B卫星CCD多光谱数据对裸土有机质空间分布格局进行研究。通过分析多光谱数据不同波段的光谱反射率及其变换形式与实地采样得到的土壤有机质含量的相关性,探寻适合绿洲区棉田表层土壤有机质含量快速反演的敏感波段及参数,并针对不同参数分别建立一元线性、二次、三次、对数、倒数、幂函数、生长型、S型回归模型,以及多元回归模型;对生成的模型进行综合对比分析,获取北疆绿洲区棉田表层土壤有机质含量的最佳反演模型,从而实现整个研究区土壤有机质空间格局的遥感反演。结果表明：HJ卫星多光谱数据4个波段的反射率均与土壤有机质含量存在显著的相关性,第3波段的倒数与土壤有机质含量相关性最为显著;且以第3波段光谱反射率作为因变量得到的三次线性回归模型对土壤有机质含量进行反演的效果最佳;通过空间布局反演得到研究区土壤有机质空间分布整体呈现南北两端有机质含量较高,中部有机质含量较低的格局。该研究表明虽然与黑土有机质含量具有差别,但是遥感技术仍能够作为绿洲区土壤有机质含量空间布局反演的方法,为遥感技术在土壤参数监测中更好的发挥作用提供理论支持,同时也为新疆棉田生产管理和农田可持续利用提供科学依据。

英文摘要：

Quick and real-time monitoring of soil organic matter（SOM） distribution based on remote sensing can support the decision-making on precision crop management. However, most previous studies have been aimed at black soil, SOM content of which is commonly higher than 2%. The research about grey desert soil（average content of SOM is less than 2%）has been reported less. This paper tries to quantitatively retrieve SOM of grey soil by using HJ-1A/1B satellite remote sensing images. Ninety-one soil samples are collected from the oasis cotton field in northern Xinjiang, China during 2013-2014. The SOM content of these samples was determined, and the mult-spectral reflectances were measured. The spectrum characteristics of 65 soil samples were analyzed, the correlation analysis was conducted, and the characteristic bands for estimating retrieval model were sought; then, the stepwise regression analysis method was used to build the inversed models.And the models include one-variable linear regressive equation, quadratic regression model, cubic regression model, loglinear regression model, inverse regression model, power function model, growth regression model, S regression model and multiple regression model for different spectrum parameters. By means of comprehensive and comparative analysis of various models, the final monitoring model of SOM was then established. Taking into account the spatial difference between the samples and remote sensing images, 26 soil samples were used to test the model. And there was a good linear relationship between the estimated and the measured SOM values（determination coefficient 0.72）. At last, based on the final monitoring mode, the distribution of the SOM was mapped. Results showed that： 1） The reflectance of each band had significant correlation with SOM content, and the reciprocal of reflectance at Band 3 had the most significant correlation with SOM content; 2） The cubic regression model was based on the refectance at Band 3, and combared with other models,it was the