土壤是人类赖以生存的自然资源，是农业可持续发展的重要组成部分。由于世界范围内日益突出的人口、资源环境间的矛盾，土地成了多种资源环境问题的集中体现者。海岸带地处海陆交汇地带，对我国经济发展、农业发展、生态环境的可持续发展起着至关重要的作用。快速获得海岸带地区土壤信息，对海岸带地区的土地进行动态监测势在必行。

以往传统的常规的测定土壤组分含量的化学分析方法周期长、成本高，精度高，测点稀少，不能满足精准农业对土壤组分时空变异状况的需求。因此，如何在短时间内获取所需的海岸带地区土壤信息就显得尤为重要。遥感技术的产生和发展，为土壤科学开辟了新的领域。上世纪八十年代以来，高光谱遥感技术的发展为更快更准的获取海岸带区域范围内土壤成分含量提供了一种技术支撑。

土壤是由多种物质组成的混合物，其成分（如矿物质、腐殖质、有机质、水、铁等）组成和各组分的含量，以及这些组分之间的相互作用都在一定程度上影响着土壤的光谱。

本文通过对莱州湾海岸带地区样本点进行野外实测土壤光谱反射率，研究样本的光谱反射率同有机质含量的相关的特征波段，来建立莱州湾海岸带地区土壤有机质含量反演模型。由于土壤有机质在可见光至近红外区域有独特的光谱响应特性，通过相关性分析，提取出土壤光谱的有机质特征波段，并通过多元线性回归、BP神经网络、传统支持向量机 (Support Vector Machine，SVM) 回归、合成核的支持向量机回归（Data Oriented Composite Kernel based Support Vector Machine Regression，DOCKSVR）建立模型。并对模型进行均方根误差（RMSE）以及相关系数（r）的检验，选取最佳模型。并尝试性的将模型应用到环境卫星HJ-1的高光谱HSI影像上，对研究区域的土壤有机质含量进行定量反演，通过遥感图像将有机质含量的空间分布直观的显示出来。

Soil is one kind of natural resources for human survival, and it is an important part of sustainable agricultural development. Because of the increasingly prominent contradiction in the worldwide population, resources and environment, soil has become the concentrated expression of a variety of resources and environmental problems. The coastal zone is located in the intersection of land and sea, and it plays a vital role in China's economic development, agricultural development, and sustainable development of the ecological environment. So quick access to the coastal zone soil, and dynamic monitoring the land of the coastal zone to is imperative.

The conventional methods of chemical analysis in the past of soil component content is with long cycle, high cost, high-precision and measurement points are 

scarce, so they can not meet the needs of precision agriculture on the status of temporal and spatial variation of soil components. Therefore, to obtain the soil information of required coastal zone in a short time is particularly important. The emergence and development of  remote sensing has opened up new areas for soil science. Since the eighties of last century, the development of hyperspectral remote sensing technology provide a technical support for accessing the content of the soil ingredients of the coastal zone faster and more accurate.

Soil is a mixture of many substances, and its ingredients (such as minerals, humus, organic matter, water, iron, etc.) composition and content of each component, as well as the interaction between these components are to some extent affected The spectrum of the soil.

In this study, the spectral reflectance of the soil samples in Laizhou Bay coastal zone had been measured, and the spectral reflectance of the sample rates were related to the content of the organic matter to pick up characteristic bands of the organic matter content. And then the inversion models of the soil organic matter content were established. In the visible to near infrared region of the spectral response characteristics, relationship was established and soil organic matter characteristic bands were extracted. Then regression models were established by multiple linear regression, BP neural network, the traditional Support Vector Machine Regression, (SVR), Data Oriented Composite Kernel based Support Vector Machine Regression, (DOCKSVR), and the models were tested and compared by root mean square error(RMSE) and correlation coefficient (r). At last the best model was selected and tried to be applied to Environmental Satellites HJ-1 Hyperspectral image (HSI) to acquire quantitative retrieval of soil organic matter content of the study area, and then the 

spatial distribution of the organic matter content was displayed intuitively.