基于LRO卫星Diviner热红外数据的月表发射率反演及其尺度效应研究

发布时间：2018-02-21 03:13

本文关键词： Diviner 热红外月表亮度发射率尺度效应矿物含量元素含量　出处：《吉林大学》2016年博士论文　论文类型：学位论文

【摘要】：反射光谱已经广泛应用于月表成分研究,而且月球在轨卫星也主要以紫外、可见光和近红外波段观测方式为主。但基于反射光谱的矿物识别和成分反演能力却受到月球环境的严重影响,主要存在反射率降低、光谱对比度减弱和光谱特征不明显等问题。因此,反射光谱的矿物识别和成分反演能力通常仅限于月壤成熟度较低的地区(约5%的月球表面)。与反射光谱比较,发射光谱不仅对于硅酸盐矿物更加敏感,对于月表物质的热物理特性(包括热惯性、粒度、密度和岩石丰度等)和月球表面粗糙度(地形因素)同样十分敏感。同时,硅酸盐矿物发射光谱特征受到月球环境的影响较小,为我们提供了一条新的了解月表成分和物理特性的途径。月球红外遥感研究主要包括地基望远镜、月表着陆、实验室测量和在轨卫星测量等方面。其中,实验室月壤样品发射率光谱特征研究和月表发射率反演是两个最主要的内容。在月球发射率光谱特征和地球发射率反演方法的基础上,以黑体辐射定律和辐射传输方程为理论基础,热红外遥感为探测手段,LRO Diviner热红外数据为研究对象,开展了月表发射率反演及其尺度效应研究。物质的发射率是建立在基于均匀等温表面的基础上。然而,空间观测的表面通常是非均匀的,遥感表面温度只能从理想等温表面中的热平衡中获取。为了建立不同温度的物质作为一个像元的辐射量与相同温度黑体贡献辐射量间的关系,相继提出了不同的像元发射率。其中,只有视在发射率明确了像元发射率的尺度效应。月球样品发射率光谱特征作为判断月表岩石和矿物成分的指示性信息,主要包括CF和RB两个光谱特征。本文通过设计实验,在大气、氮气冷背景和真空环境下利用三种发射率测量方法进行了模拟月壤测量实验,获得了发射率光谱,结合已有Apollo样品热红外波段反射率光谱,分析了成熟度、矿物成分、化学成分、粒径和样品温度对发射率光谱特征的影响,确定了发射率光谱CF特征可作为月表矿物成分和化学成分反演的依据。以2012年1-6月Diviner通道3至9月表赤道附近亮温数据为研究对象,制作通道亮温误差棒图,分析Diviner数据稳定性。针对Diviner白天和夜晚红外数据的分布规律,以及数据中少量的“坏点”或者“坏线”问题,提出了Diviner数据处理方法,包括亮温数值拟合、奇异点剔除、分时段拟合和纬向校正等。通过应用这些数据处理方法于Diviner通道6的亮温数据中,获得了高空间分辨率、全覆盖度的月表亮温分布图,并利用时段亮温值、DEM和CCD数据评价了亮温分布,验证了Diviner红外数据处理方法的可行性和可靠性。以发射率反演算法中的发射率归一化法为基础,提出了月表发射率的反演算法。通过二次拟合3个通道亮温值求得像元最大亮温值代替像元实际温度值,利用月球辐射传输方程反演像元通道发射率值。利用反射率反演算法和Diviner的8μm通道亮温数据,反演了月表发射率,进而再利用二次拟合3个通道发射率值获得发射率光谱特征CF值。通过月球6-18点Apollo15登月点像元发射率反演值,分析像元发射率尺度效应,并发现了发射率的空间效应。利用Diviner两种空间分辨率的发射率反演值分析了发射率尺度效应的分布规律。结合CF特征固有属性和误差理论,消除了CF值中的尺度效应和空间效应,为月表成分反演提供可靠发射率数据。根据18种不同粒径Apollo样品发射率光谱特征与其成分含量间的相关性,建立了月表成分发射率反演模型。其中,CF值与钛铁矿、斜长石、易变辉石和富镁单斜辉石四种矿物的相关性分别为0.5825、0.5352、0.4483和0.887,与Si O2、Ti O2、Al2O3、Cr2O3、Mg O、Ca O、Mn O、Fe O和SO2的相关性分别为0.5539、0.6404、0.5992、0.7834、0.6836、0.482、0.75、0.7097和0.5306。结合消除尺度效应和空间效应的CF值影像,反演了月表四种矿物和九种元素含量。利用Apollo15采样点样品成分含量验证了反演精度。四种矿物反演值和实测值的均方差误差分别为0.33、15.35、5.28和1.58,七种元素含量(Si O2、Ti O2、Al2O3、Mg O、Ca O、Mn O和Fe O)反演值和实测值均方差误差分别为2.25、0.92、2.63、2.80、1.56、0.05和2.34。结果表明,利用CF特征可反演月表矿物和元素含量,并且元素含量的反演精度要高于矿物含量的反演精度,为月表成分反演提供了一种新的方法。
[Abstract]:Reflectance spectroscopy has been widely used in the research of lunar satellite in orbit and lunar components, mainly to UV, mainly in visible light and near infrared observations. But mineral identification and component retrieval ability based on reflection spectrum is severely affected by the lunar environment, the main problems exist in lower reflectivity, spectral contrast weakened and spectral features are not obvious etc. Therefore, mineral identification and component retrieval ability of reflectance spectra is usually confined to the lower regolith maturity area (about 5% of the surface of the moon). Compared with the reflectance spectra, emission spectra not only for silicate minerals more sensitive for thermal physical properties of lunar material (including particle size, density and thermal inertia. Rock abundance) and the lunar surface roughness (topography) is also very sensitive. At the same time, silicate mineral emissivity spectral features affected by the lunar environment is small, as we provide A new understanding of the composition and physical properties of the lunar moon. Infrared remote sensing research including ground-based telescope, lunar landing, laboratory measurement and satellite measurement. The laboratory regolith sample emissivity spectral features of surface emissivity and inversion are two main contents. Based on lunar emissivity spectra the characteristics and the earth emissivity inversion method, the blackbody radiation law and the radiative transfer equation based on the theory of thermal infrared remote sensing detection method for LRO, Diviner thermal infrared data as the research object, the surface emissivity inversion and its scale effect. The emissivity substances are established based on the uniform and isothermal surface however, the surface is not uniform space observation, remote sensing surface temperature can only be obtained from the ideal isothermal surface in thermal equilibrium. In order to establish the material at different temperatures As a matter between blackbody radiation a pixel with the same temperature with the amount of radiation between, have proposed different pixel emission rate. Among them, only as the rate of clear pixel scale effect of emissivity at launch. Emissivity spectral features of lunar samples as the judgment information indicative of rock and mineral composition of the lunar surface, including CF and RB two spectral characteristics. In this paper, through the design of experiments in nitrogen atmosphere, cold background and vacuum environment using three kinds of simulated lunar soil measurement method for measurement of emissivity, the emissivity spectra obtained, combined with the existing Apollo sample thermal infrared reflectance spectroscopy, maturity, mineral composition, chemical composition analysis. Influence of particle size and sample temperature on the emissivity spectral features, the emissivity spectra of CF characteristics can be used as surface mineral composition and chemical composition of the inversion basis. In 2012 1-6 months Diviner Channel 3 to 9 lunar equator brightness temperature data as the research object, making the brightness temperature error of the chart, the stability analysis of Diviner data. According to the distribution law of Diviner day and night infrared data, and adopts a few "data" or "bad" problem, put forward the Diviner data processing methods, including brightness temperature value fitting, singular points elimination, piecewise fitting and zonal correction. By using these methods of data processing in Diviner brightness temperature data 6, to obtain the high spatial resolution, full coverage of the surface brightness temperature distribution maps, and using time brightness temperature values of DEM and CCD data to evaluate the brightness temperature distribution. To verify the feasibility and reliability of Diviner infrared data processing method. The emissivity normalized emissivity method inversion algorithm is based on the proposed inversion algorithm of lunar surface emissivity. By two fitting 3 brightness temperature value Have the maximum brightness temperature value of pixel pixel instead of the actual temperature value, the radiative transfer equation inversion pixel channel emissivity values. Using 8 M channel reflectance inversion algorithm and Diviner brightness temperature data, the inversion of lunar surface emissivity, and then use the two fitting 3 channel emissivity values obtained emissivity spectral features of CF value the moon. 6-18 Apollo15 lunar pixel emissivity inversion analysis of emissivity value, pixel scale effect, and found that the space effect of emissivity. Analyze the distribution of emission rate scale effect of the Diviner value by two kinds of spatial resolution emissivity inversion. The combination of CF features and inherent properties of error theory, eliminating the scale effect and spatial effect of CF value, to provide reliable data for surface emissivity inversion. According to the composition of 18 different particle size Apollo emission rate correlation spectral characteristics and its components in the building The lunar component emissivity inversion model. The CF value of plagioclase and ilmenite, correlation, pigeonite and magnesium rich clinopyroxene four minerals were 0.5825,0.5352,0.4483 and 0.887, Si and O2, Ti, O2, Al2O3, Cr2O3, Mg O, Ca O, Mn O, Fe O and SO2 related are 0.5539,0.6404,0.5992,0.7834,0.6836,0.482,0.75,0.7097 and 0.5306. combined with the elimination of the size effect and the space effect of the CF value of the lunar surface image, inversion four minerals and nine kinds of element content. Content of samples to verify the sampling accuracy by Apollo15. Four kinds of mineral value inversion error variance and the measured values were 0.33,15.35,5.28 and 1.58, the contents of seven elements (Si O2, Ti O2, Al2O3, Mg O, Ca O, Mn O and Fe O) inversion value and measured value of error variance that are respectively 2.25,0.92,2.63,2.80,1.56,0.05 and 2.34. results, the characteristics of lunar minerals and can be retrieved using the CF element Content, accuracy and precision of inversion inversion element content is higher than that of mineral content, provides a new method for the inversion of lunar components.

【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P184.5

【相似文献】