中国科学院遥感与数字地球研究所-高光谱遥感应用技术研究室推荐内容
【2009博】月球表面物质高光谱探测技术研究
|
【作者】 和海霞 【导师】 童庆禧；张立福 【 学位年度 】 2009 【论文级别】 博士 【关键词】 月壤光谱特征，月球表面物质填图，高光谱，月壤光谱影响因素 【Key words】 Hyperspectral Remote Sensing; Lunar Soil Spectral Characteristic; Lunar Surface Mineral Mapping; Influencing Factors of Lunar Soil Spectral 【中文摘要】 月球表面物质探测对于研究月球的起源和演化有着重要意义。高光谱遥感凭借其所具有的精细光谱信息及其在地物识别领域的主导地位，在月球表面物质探测中扮演着非常重要的角色。高光谱遥感在地球表面物质探测和火星探测领域的技术体系和方法已经比较成熟，并得到了推广应用，但高光谱在月球探测中的应用才刚起步。“嫦娥一号”卫星搭载的干涉成像光谱仪目前已经获取了大量的高光谱数据，但相关的月球表面物质光谱特征定量分析和理解还比较薄弱，这在一定程度上限制了干涉光谱仪数据的推广应用和月球表面物质探测的精度。因此，研究月球表面物质高光谱探测的基础问题，分析月球表面物质探测的特点和难点，建立月球表面物质探测方法体系，是提高探测精度的重要途径，是探月卫星数据深层次应用的重要支撑。 本文拟基于已公布探月卫星获取的数据、地基望远镜观测的光谱数据以及月壤返回样品的光谱实测数据等，对月球表面物质的光谱产生机理、光谱特征和光谱影响因素进行定量分析和影响因子模型建立，并根据光谱特征与物质理化含量之间的关系，进行基于反射率和发射率的月球表面物质高光谱探测方法研究和信息提取实验，进行探测结果的方法改进及误差来源分析，初步形成一套完整的月球表面物质高光谱探测方法体系。该研究的开展将为下一代探月卫星传感器的设置提供参考，为高光谱遥感在月球探测领域的应用打下基础。 论文的主要内容和研究成果如下： （1）月球表面物质光谱特征的比较分析。在月球表面物质光谱产生机理研究的基础上，基于探月卫星获取的光谱数据、月球返回样品的实验室测量光谱数据、地基望远镜的观测数据、地球岩石与矿物光谱库数据等进行月球表面物质可见光-近红外与热红外光谱特征的比较分析。 （2）月球表面物质光谱影响因素分析和影响因子模型的建立。本文重点分析了月壤的矿物与化学组成、月壤粒度大小、太空环境、月球表面温度等对月球表面物质光谱特征的影响,并在此基础上建立了月壤光谱影响因子模型。 （3）月球表面物质光谱模拟技术研究。在分析月球表面矿物组成、光谱特征和影响因素的基础上，将理化方法模拟的光谱测量和基于辐射传输方程的光谱模拟结合起来进行月球表面物质光谱模拟的研究。 （4）月球表面数据反射率与发射率反演方法研究。在分析月球表面的辐射传输过程的基础上,提出了月球表面数据绝对反射率反演方法,比较了几种相对反射率反演的方法,确定了平场域法最适合月球表面相对反射率反演，探索了热红外谱段内月球表面的温度与发射率分离方法。 （5）月球表面矿物填图实验。基于月球返回样品实验室测量高光谱数据、Clementine紫外-近红外图像数据、热红外高光谱数据进行了月球表面矿物填图实验。结果表明，马氏距离和支持向量机分类方法是最有效的方法，可以区分物质之间的微小差别；在SAM和SID两种光谱匹配算法中，SID更能区分光谱之间的微小差别，更利于进行月球表面物质的探测；在基于图像数据的TiO2和FeO的定量反演方法中，借助光谱比值的统计方法比较常见，但是其普适性不是很强，对月海地区的物质具有较高的反演精度，对月球高地的物质反演精度较低。基于此，本文对适合月海和月球高地的探测方法进行了探索，并对适合月球表面物质探测的下一代干涉成像光谱仪的参数设置进行了探讨。 （6）月球表面物质探测特点与难点分析。本文对制约月球表面物质高光谱探测精度提高的因素进行了说明，对目前月球表面物质高光谱探测中存在的问题进行了讨论，并对今后月球表面物质探测需要重点研究的内容提出了建议。 本文在月球表面物质光谱影响因子模型建立、月球表面反射率和发射率反演算法、月球表面矿物探测方法等方面做出了尝试，针对月球表面状况，初步提出了一些提高月球表面物质高光谱探测精度的建议，旨在推动高光谱遥感在月球表面物质领域的应用研究。
【Abstract】 The distribution and types of the minerals and rocks on the lunar surface is the main topic of the lunar exploration and it is helpful to investigate the origin and evolution of the moon. Hyperspectral remote sensing is effective to the exploration to the lunar minerals and rocks because of its fine spectral information and its dominant role in mineral identification.At present, the imaging spectroscopy has been widely used on the earth and the Mars and the mineral identification based on the imaging spectroscopy is mature, while the importance of the imaging spectroscopy to the lunar exploration has been recognized recently. Several hyperspectral instruments has been carried on the lunar satellite such as M3, HySI, SIR-2 carried on “Changdrayaan-1” and the Sagnac Imaging Fourier Transform Spectrometer carried on “change-1”.A lot of hyperspectral remote sensing data has been aquired by the Sagnac Imaging Fourier Transform Spectrometer which was first used in the lunar exploration.While the lack of the quantitative analysis on the spectral characteristic of the lunar soil reduces the precision of the lunar soil detection and limits the population and application of these aquired data in some sense.Therefore, studying the basic issues and the difficulty of the lunar soil detection can improve the detection precision. The goal of this paper is to investigate the theories and main algorithms of the lunar surface quantitative detection. The main contents and research results of this paper are as follows: (1) Based on the corresponding background information, the spectra producing mechanism of lunar soils was studied. The visible- near infrared and thermal spectral characteristics of the lunar surface minerals was inducted and analyzed using the spectral data achieved by the lunar satellite, the spectral data measured in the simulated library environment, the data observed by the earth-based telescopes. (2) The quantitative analysis was made on the factors which may affect the spectral characteristics, such as the physical and chemical components, mineral size, the lunar surface temperature and the outer space environment. Then, the model of these factors to the spectral characteristics was built. (3) The physical and chemical methods were applied to simulate the lunar soil using minerals on the earth and the spectral of the simulated lunar soil was measured by ASD spectrometer and Nuance multispectral imaging system.At the same time, the spectra were simulated according to the Radiative Transfer Equation.At last, these two spectral simulation methods were compared. (4) Based on the calibration results and lunar surface illumination charcteristics, the absolute reflectance retrieval method was put forward to and several relative reflectance retrieval methods were used to the lunar image data and the retrieval results were compared. The results indicated that the Flat Field method is the best method to the lunar image data .Then, according to the radioactive transfer procedure, the emissivity retrieval was explored. (5) In this paper the lunar surface mineral mapping was performed based on the laboratory measurements of the lunar samples spectra, the Clementine image data, and the emission spectral data. The results indicated that the support vector machine is best to the surface features classification for the lunar spectra.Of all of the spectral matching methods,Spectral Information Divergence can detect the minimal difference between the silimar spectra of different types and different chemical component.The relationship between the laboratory measurements of the spectra and mineral and chemical component and the Clementine relative image was built to perform the TiO2 and FeO content retrival and was successfully used to the 10S015 image.The retrival results based on the emission spectral data indicated that the position and depth of the Christensen features can be used to retirive the mineral and chemical component, while the stability of the built relationship should be studied further.At the same time, the parameters setting suggestions for the next hyperspectral machine were brought forward according to the experimental results. (6) The difficulty and the characteristics of the lunar surface minerals were analyzed and the content which was important to the lunar surface minerals detection was suggested. This paper makes great efforts on the building of the influencing factors model, the retrievaling algorithm of lunar emissivity and reflectance and the method of lunar mineral mapping.Some suggestions on the improvement of the lunar surface minerals detection was made based on the lunar surface features in order to promote the application of the hyperspectral remote sensing in the lunar surface matters detection.
(责任编辑：admin) |
