集成光学与微波遥感苏打盐碱地水盐含量反演方法研究

发布时间：2018-06-14 11:49

本文选题：AIEM模型 + 微波遥感　；参考：《中国科学院研究生院（东北地理与农业生态研究所）》2014年博士论文

【摘要】：土壤盐碱化是世界上面临的一个严重问题，它严重影响干旱、半干旱区域的生态及农业可持续发展。我国吉林省西部地区属于世界三大苏打盐碱地之一，土地盐碱化使得该地区生态平衡受到严重影响。利用遥感技术快速、准确地监测盐碱化变化过程，对合理开发利用盐碱地具有重要意义。中国科学院东北地理与农业生态研究所大安碱地生态试验站长期在吉林省西部地区开展苏打盐碱地的改良研究，结果表明不同碱化特性的盐碱地生长着不同的植被类型，当盐碱化程度较高时，无植被生长。利用光学遥感技术能够有效地监测植被生长区和裸露盐碱地，但是对植被高度、长势等无法进行有效的监测，因此不能对盐碱地特性进行更加细致的识别，尤其是水盐含量变化的监测。含盐土壤在一定含水量的条件下，具有较好的导电特性，使微波后向散射、微波辐射特性对其介电常数具有较强的敏感性，使应用微波遥感监测盐碱土壤具有可行性。然而，微波遥感信息常常受植被或地表粗糙度等因素影响，增加了盐碱地特性识别的复杂性和不确定性。论文在中国科学院重要方向性项目“土壤湿度与积雪参量高精度微波遥感反演机理研究”的支持下，依托中国科学院长春净月潭遥感实验站和大安碱地生态试验站，通过有效选取研究实验区，以Landsat8、RADARSAT-2C波段全极化SLC影像为信息源，在星地同步观测实验的基础上，开展了吉林省西部地区苏打盐碱地水盐特性反演方法的研究。主要研究成果如下：（1）利用合成孔径雷达（SAR）极化目标分解理论，提取了研究区盐渍土的微波遥感后向散射信息特征，有效地应用于轻度盐碱地的识别，提高了相应的识别水平。（2）通过SAR极化目标分解，利用Cloude-Pottier分解散射熵值的变化，论述了盐碱地的散射随机性，其熵值高于农田和水体。由此表明，当雷达足印大于15m时，通过建立雷达后向散射系数与盐碱土介电常数之间的确定性函数关系，开展盐碱地特性微波遥感的监测是不可行的。（3）利用盐碱地Freeman-Durden分解的三种散射机制的变异系数（标准差与均值比值）较大的特征，表述了盐碱地具有相对较高的散射离散度，，主要体现在25*25像元范围四极化后向散射统计中遥感信息变化范围较大。同时，也阐明了草地植被覆盖的随机性、草地植被长势随机性、盐碱地地表含盐量的随机性、叶片方向随机性等构成的混合像元是影响雷达后向散射系数随机特性的重要因素。（4）根据盐碱土介电常数对频率响应差异，利用双频地基微波辐射计亮度温度观测数据，应用Levernberg-Marquardt算法，实现了土壤介电常数（实部和虚部）及地表均方根高度三参数的同时反演。（5）应用全极化雷达数据和NDVI建立了含水量、含盐量反演的经验公式，其中含水量反演均方根误差为3.169%（体积含水量），含盐量反演误差为3.175g/Kg。（6）提出了基于理论知识和模拟数据库，进行盐碱土壤水盐含量半经验模型的方案，并进行了初步实现。首先，利用AIEM模型建立模拟数据库，通过各参数敏感性分析，建立了组合粗糙度、介电常数幅值与HH、VV极化后向散射系数的关系式，再然后结合水云模型，去除植被影响，反演得到地表组合粗糙度值及介电常数幅值。论文的研究工作为开展盐碱地特性遥感监测提供了新的途径和方法，为国家自然科学基金面上项目“集成光学与微波遥感苏打盐碱地水盐含量的反演方法研究”的立项与研究奠定了基础。
[Abstract]:Soil salinization is a serious problem in the world. It seriously affects the ecological and agricultural sustainable development of arid and semi-arid areas. The western region of Jilin province is one of the three salinized saline alkali land in the world. The land salinization makes the ecological balance of the region seriously affected. The rapid and accurate monitoring of salt by remote sensing technology is used. The process of alkalization is of great significance for the rational exploitation and utilization of saline alkali soil.
The Da an alkaline land ecological test station of the Northeast geographical and Agro ecological research institute of the Chinese Academy of Sciences has long carried out the improvement of soda saline alkali land in the western region of Jilin province. The results show that different alkalinized saline alkali land has different vegetation types. When the salinity is high, no vegetation is grown. The use of optical remote sensing technology can be effective. The vegetation growth area and the bare saline alkali soil are monitored, but the vegetation height and potential can not be monitored effectively. Therefore, the characteristics of the salt and alkali land can not be more carefully identified, especially the monitoring of the change of water and salt content. The salt soil has good electrical conductivity under certain water content, which makes the microwave backward scattering and microwave radiant. It is feasible to use microwave remote sensing to monitor saline soil. However, the microwave remote sensing information is often influenced by vegetation or surface roughness and so on, which increases the complexity and uncertainty of the characteristics identification of saline alkali land.
Under the support of the important directional project of the Chinese Academy of Sciences, "study on the mechanism of high precision microwave remote sensing inversion of soil moisture and snow parameters", relying on the Changchun clean moon remote sensing experimental station and the Daan alkaline land ecological test station of the Chinese Academy of Sciences, the research experimentation area is effectively selected as the Landsat8, RADARSAT-2C band fully polarized SLC image as a letter On the basis of stellar geostationary observation experiments, the inversion method of water and salt characteristics of soda saline alkali land in Western Jilin province was studied. The main research results are as follows:
(1) using the theory of polarimetric target decomposition of synthetic aperture radar (SAR), the characteristics of the backscatter information of the microwave remote sensing in the saline soil of the study area are extracted, which is effectively applied to the recognition of the mild saline alkali soil and improves the corresponding recognition level.
(2) through the decomposition of SAR polarization target and the change of Cloude-Pottier decomposition entropy, the scattering randomness of saline alkali land is discussed. The entropy value is higher than that of farmland and water body. Thus, it is shown that when the radar foot printing is larger than 15m, the relationship between the radar backscatter coefficient and the alkali soil dielectric constant is established, and the salt alkali earth is carried out. The monitoring of sexual microwave remote sensing is not feasible.
(3) using the variation coefficient of three scattering mechanisms of Freeman-Durden decomposition in saline alkali land (the standard deviation and the mean value ratio), the relatively high scattering dispersion is expressed in the saline alkali land, which is mainly reflected in the large range of remote sensing information in the statistics of the four polarization after the four polarization. Meanwhile, the grassland vegetation cover is also clarified. The randomness of the cover, the randomness of the vegetation length, the randomness of the salt content on the ground surface and the randomness of the leaf direction are the important factors affecting the random characteristics of the backscattering coefficient of the radar.
(4) according to the difference of the frequency response between the dielectric constant of the saline alkali soil and the observation data of the brightness temperature of the double frequency foundation microwave radiometer, the Levernberg-Marquardt algorithm has been used to achieve the simultaneous inversion of the three parameters of the soil dielectric constant (real part and virtual part) and the height of the root mean square root.
(5) an empirical formula of water content and salt content inversion is established by using full polarimetric radar data and NDVI. The root mean square error of water content inversion is 3.169% (volume of water), and the inversion error of salt content is 3.175g/Kg.
(6) a semi empirical model of salt and alkali soil water and salt content based on theoretical knowledge and simulated database is proposed and carried out preliminarily. First, the AIEM model is used to establish an analog database. Through the sensitivity analysis of various parameters, the relationship between the composite roughness, the dielectric constant amplitude and the HH, VV polarization direction scattering coefficient is established. Combined with the water cloud model, the vegetation roughness and permittivity amplitude were retrieved.
The research work of this paper provides a new way and method for remote sensing monitoring of the characteristics of saline alkali land. It lays a foundation for the research on the project of "integrated optics and microwave remote sensing and saline alkali soil salt and salt content inversion method" on the National Natural Science Foundation.
【学位授予单位】：中国科学院研究生院（东北地理与农业生态研究所）
【学位级别】：博士
【学位授予年份】：2014
【分类号】：S153;P237

【参考文献】