【作者】 闫柏琨；

【导师】 王润生；

【作者基本信息】 中国地质大学（北京） ， 地图制图学与地理信息工程， 2006， 博士

【摘要】 可见光-反射红外高光谱遥感使遥感从地物鉴别(Discrimination)发展到地物直接识别(Identification),可识别的矿物种类及可靠性都有很大提高。但由于光谱区间所限,难以识别不含水造岩矿物。将遥感波段拓展至热红外是解决这一问题的途径之一。热红外发射光谱可以探测识别造岩矿物Si-O键振动光谱特征,对可见光-反射红外遥感是有效补充。二者综合运用,建立全谱段光谱矿物识别规则及方法可以全面提高遥感岩矿识别的能力与精度。但是,由于热红外遥感数据获取、处理及发射率反演等问题难度较大,基于发射光谱的岩矿信息提取研究滞后。本文从岩矿光谱机理(光谱的决定因素及变异因素)、数据大气校正及发射率反演、基于发射光谱的岩矿信息提取方法三个方面开展了研究。运用岩矿辐射传输模型计算了几种矿物发射光谱并与ASU(Arizona StateUniversity)光谱库中实测光谱进行了对比,表明Mie/Hapke模型基本可以计算模拟出矿物的发射光谱特征。运用Mie/Hapke模型研究了矿物粒度及出射角度对发射光谱特征造成的影响。不同矿物发射光谱特征随粒度变化规律完全不同,随着粒度变化发射光谱特征会发生突变,光谱谱形及吸收特征的深度等都会发生很大变化。不同矿物发射光谱特征随出射角度的变化规律相同,随着出射角度变大发射率变小,发射率越小的波段变化幅度越大、变化速率越大。矿物反射光谱为非线性混合,运用岩矿辐射传输模型可以对矿物反射混合光谱进行线性化处理,处理之后矿物光谱混合更接近线性混合,光谱线性解混及矿物含量提取精度大幅度提高。研究表明,岩矿辐射传输模型从物理层面上模拟光谱的形成及变异机理,是研究岩矿光谱机理的有效方法。热红外遥感大气校正及发射率反演精度直接影响到基于发射光谱的地质应用效果,运用“黑箱”模型定量研究了各种大气廓线参数对热红外遥感信息及相对发射率的影响规律,在此基础上开发了没有大气廓线参数情况下的大气校正及相对发射率反演方法,反演结果与基于大气廓线参数的反演结果非常接近,地质应用效果几乎完全一致。在应用方面,利用ASTER热红外多光谱数据定量反演了地表岩石SiO2含量,该方法可用于寻找与SiO2含量有关的矿床,比如与基性超基性岩有关的铜镍矿床、与硅化有关的金矿床。该方法在监测土地沙化方面同样有一定的应用潜力。利用火星TES(Thermal Emission Spectrometer)热红外高光谱数据反演了火星表面白云母、硫酸盐、钙长石、赤铁矿、方解石、白云石、钾-钠长石、高钙辉石、低钙辉石等矿物含量及分布,这些成果对研究火星形成演化有重要意义。综合上述研究表明,热红外遥感在矿物识别方面有特有的优势,研究矿物岩石发射光谱特征及机理、数据处理及信息提取技术方法,在此基础上开发全谱段矿物识别规则,综合应用可见光-反射红外遥感、热红外遥感进行岩矿识别是提高遥感岩矿识别精度、能力及可靠性的有效途径之一。更多还原

【Abstract】 Hyperspectral remote sensing(400～2500nm) prompt remote sensing technologyto develop from ground matter discrimination to identification. In remote sensinggeology, the classes and reliability of mineral identified have been improved.However, the rock-forming minerals containing no cation-OH bond can not beidentified because this spectral band can only detect some minerals containingcation-OH bond. One of approach to detect minerals containing no cation-OH bond isto use thermal remote sensing technology, which can detect the Si-O bond ofrock-forming minerals. Therefore, combining use of hyperspectral(400～2500nm)and thermal remote sensing will improve the ability and precision of mineralsidentification by remote sensing. However, minerals information extraction based onemissivity spectral features has lagged behind hyperspectral remote sensing, becauseit is difficult to obtain and process thermal remote sensing data as well as emissivityderivation.In this paper, mechanisms of emissivity of minerals (determinant and variationfactors of emissivity), atmospheric correction and emissivity derivation, and methodof minerals identification based on emissivity spectral features are studied.Emissivity spectral of several minerals have been calculated using mineralradiative transfer model and compared with measured spectral in ASU spectraldatabase (Arizona State University). The comparison indicates that Mie/Hapke modelcan simulate emissivity spectral features of minerals. It is studied using Mie/Hapkemodel that mineral emissivity spectral features variate with the mineral granularityand emission angle, and results indicate that the law of emissivity variation withgranularity is totally different from mineral to mineral. Along with the variation ofmineral granularity, the shape and absorption depth of mineral emissivity spectral willall variate. However, the law of emissivity variation with emission angle of differentminerals are identical. Along with the increasement of emission angle, emissivitydecrease. The more emissivity is small, the more variation range and speed are large.The reflectance mixture of mineral is non-linear, and can be lineated using mineralradiative transfer model. After the mixture spectral is lineated, the precision of linearunmixing of spectral and mineral content extraction will be improved greatly. Mineralradiative transfer model can model the mechanisms of spectral formation andvariation, and is one of study method of spectral mechanism.The atmospheric correction and emissivity derivation of thermal remote sensingdata will affect application greatly. In this paper, it is quantitatively evaluated howatmospheric profiles affect the atmospheric correction and relative emissivityderivation. Based on the study results, method of atmospheric correction and relativeemissivity derivation is developed without atmospheric profiles. The derivation resultof ASTER thermal infrared data using this method is identical with the standardproduct.SiO2 content of rock is retrieved quantitatively using ASTER thermal infrareddata, and the derivation method can be used to explorate ore deposit concerning withSiO2 of rock, such as copper-nickel deposit exist in basic-ultrabasic intrusive bodyand gold deposit forming in rocks containing high-Si content. This method also can beused in study the degeneration of soil. The content and distribution of several mineralssuch as muscovite, sulphate, and feldspar, hematite, calcite, dolomite and pyroxeneare derived using TES(Thermal Emission Spectrometer) data of Mars, and thisresearch result has important significance in study the formation and evolvement ofMars.It is indicates that thermal remote sensing has some advantages over reflectanceremote sening(400～2500nm) in some aspects. Mechanisms of emissivity of minerals,data processing and information extraction are studied, and based on study resultsmineral identification regulation of complete spectrum can be developed. It is one ofapproach improving precision and reliability of mineral identification to syntheticallyuse thermal remote sensing and reflectance remote sensing(400～2500nm).更多还原