en
×

分享给微信好友或者朋友圈

使用微信“扫一扫”功能。
作者简介:

吴梦娟,女,1990年生。博士,助理研究员,主要从事遥感地质、高光谱遥感、定量遥感研究。E-mail:wumengjuan@nnnu.edu.cn。

通讯作者:

靳佳,男,1989年生。助理研究员,主要从事定量遥感研究。E-mail:jinjia@nnnu.edu.cn。

参考文献
Adams J B, Goullaud L H. 1978. Plagioclase feldspars-visible and near infrared diffuse reflectance spectra as applied to remote sensing. Proceedings of the Lunar and Planetary Science Conference 9th.
参考文献
Ben-Dor E, Chabrillat S, Dematte J A M, Taylor G R, Hill J, Whiting M L, Sommer S. 2009. Using imaging spectroscopy to study soil properties. Remote Sensing of Environment, 113: S38~S55.
参考文献
Bonin B, Bébien J, Masson P. 2002. Granite: A planetary point of view. Gondwana Research, 5: 261~273.
参考文献
Brown M. 2013. Granite: From genesis to emplacement. Geological Society of America Bulletin, 125: 1079~1113.
参考文献
Cavanaugh J E. 1997. Unifying the derivations for the Akaike and corrected Akaike information criteria. Statistics & Probability Letters, 33(2): 201~208.
参考文献
Chen Yanbing, Kuang Runyuan, Zeng Shuai. 2018. Discriminant analysis of typical vegetation species in Poyang Lake wetland based on hyperspectral data. Yangtze River, 49(20): 19~23 (in Chinese with English abstract).
参考文献
Clevers J G. 1988. Application of a weighted infrared-red vegetation index for estimating leaf area index by correcting for soil moisture. Remote Sensing of Environment, 25: 53~69.
参考文献
Comar A, Baret F, Viénot F, Yan L, Solan de B. 2012. Wheat leaf bidirectional reflectance measurements: Description and quantification of the volume, specular and hot-spot scattering features. Remote Sensing of Environment, 121: 26~35.
参考文献
Gaffey S J. 1987. Spectral reflectance of carbonate minerals in the visible and near infrared (0. 35~2. 55 μm): Anhydrous carbonate minerals. Journal of Geophysical Research, 92(B2): 1429~1440.
参考文献
Garain S, Mitra D, Das P. 2021. Mapping hydrocarbon microseepage prospect areas by integrated studies of ASTER processing, geochemistry and geophysical surveys in Assam-Arakan fold belt, NE India. International Journal of Applied Earth Observation and Geoinformation, 102: 102432.
参考文献
Guo Ni. 2003. Vegetation index and its advances. Arid Meteorology, 21(4): 71~75 (in Chinese with English abstract).
参考文献
Hapke B. 1993. Theory of Reflectance and Emittance Spectroscopy. New York: Cambridge University Press.
参考文献
Hapke B. 2012. Bidirectional reflectance spectroscopy 7: The single particle phase function hockey stick relation. Icarus, 221(2): 1079~1083.
参考文献
Hecker C, van der Meijde M, van der Meer F. 2010. Thermal infrared spectroscopy on feldspars-success, limitations and their implication for remote sensing. Earth-Science Reviews, 103: 60~70.
参考文献
Hunt G R, Salisbury J W, Lenhoff C J. 1973. Visible and near-infrared spectra of minerals and rocks: VI. additional silicates. Modern Geology, 4: 85~106.
参考文献
Li Shuai, Milliken R E. 2015. Estimating the modal mineralogy of eucrite and diogenite meteorites using visible-near infrared reflectance spectroscopy. Meteoritics & Planetary Science, 50: 1821~1850.
参考文献
Li Xiaowen. 1989. Bidirectional reflectance and angular signature of objects. Remote Sensing of Environment China, 2(1): 67~72 (in Chinese with English abstract).
参考文献
Liu Xiuying. 2016. Monitoring models of physiological parameters of corn and farmland soil information based on hyper-spectral reflectance. Doctoral dissertation of Northwest A & F University (in Chinese with English abstract).
参考文献
Nash D B, Salisbury J W. 1991. Infrared reflectance spectra (2. 2-15-Mu-M) of plagioclase feldspars. Geophysical Research Letters, 18: 1151~1154.
参考文献
Quinn D P, Gillis Davis J J, Lucey P G. 2010. Using microsoft excel for Hapke modeling: A technique to simplify calculations of optical constants and reflectance spectra. Proceedings of the Lunar and Planetary Science Conference.
参考文献
Richardson A J, Wiegand C L. 1977. Distinguishing vegetation from soil background information. Photogrammetric Engineering and Remote Sensing, 43: 1541~1552.
参考文献
Robertson K M, Milliken R E, Li S. 2016. Estimating mineral abundances of clay and gypsum mixtures using radiative transfer models applied to visible-near infrared reflectance spectra. Icarus, 277: 171~186.
参考文献
Roosjen P P J, Benjamin B, Suomalainen J M, Bartholomeus H M, Kooistra L, Clevers J G P W. 2018. Improved estimation of leaf area index and leaf chlorophyll content of a potato crop using multi-angle spectral data-potential of unmanned aerial vehicle imagery. International Journal of Applied Earth Observation & Geoinformation, 66(4): 14~26.
参考文献
Streckeisen A. 1968. Account of classification and nomenclature of igneous rocks. Reprint of 23rd International Geological Congress Prague.
参考文献
Sun Zhongqiu, Wu Zhengfang, Zhao Yunsheng. 2014. Semi-automatic laboratory goniospectrometer system for performing multi-angular reflectance and polarization measurements for natural surfaces. Review of Science Instruments, 85: 014503.
参考文献
Tan Wei, Qin Xiaorong, Liu Jiacheng, Zhou Meifu, He Hongping, Wang Yan Christina, Huang Jian, Zhu Jianxi, Yao Yuzeng, Cudahy T. 2022. Feasibility of visible short-wave infrared reflectance spectroscopy to characterize regolith-hosted rare earth element mineralization. Economic Geology, 117(2): 495~508.
参考文献
Tong Qingxi, Meng Qingyan, Yang Hang. 2018. Development history and future prospect of remote sensing technology. City and Disaster Reduction, (6): 2~11 (in Chinese with English abstract).
参考文献
Yang Gaoxue. 2008. The study of granitoids in Kalamaili area of eastern 657 Junggar, Xinjiang. Master's thesis of Chang'an University (in Chinese with English abstract).
参考文献
陈彦兵, 况润元, 曾帅. 2018. 基于高光谱数据的鄱阳湖湿地典型植被识别分析. 人民长江, 49(20): 19~23.
参考文献
郭铌. 2003. 植被指数及其研究进展. 干旱气象, 21(4): 71~75.
参考文献
李小文. 1989. 地物的二向性反射和方向谱特征. 环境遥感, 2(1): 67~72.
参考文献
刘秀英. 2016. 玉米生理参数及农田土壤信息高光谱监测模型研究. 西北农林科技大学博士学位论文.
参考文献
童庆禧, 孟庆岩, 杨杭. 2018. 遥感技术发展历程与未来展望. 城市与减灾, (6): 2~11.
参考文献
杨高学. 2008. 新疆东准噶尔卡拉麦里地区花岗岩类研究. 长安大学硕士学位论文.
目录contents

    摘要

    花岗岩中长石含量的定量估算有助于其定名和分类,为后续相关地质过程研究提供基础数据。在可见光—近红外—短波红外波段范围内(0.35~2.5 μm),传统基于光谱吸收特征参数反演矿物种类和含量的方法,不适用于像长石这类无诊断性吸收特征的矿物。同时,基于物理的辐射传输模型由于计算复杂,在较大程度上限制了该方法在矿物定量反演中的应用。本文基于多角度高光谱数据,通过不同光谱预处理方法及光谱指数类型的组合实验,创建用于估算花岗岩中长石比例的光谱指数模型。结果表明,使用2035 nm波段的反射率二重差分型(CRDDn2035)指数模型,在不同实测数据集中均具有良好表现,估算精度达到0.81。本研究创建了一种适用于估算长石占比的光谱指数模型,为定量反演具有弱吸收特征的岩矿信息提供了新的技术手段与思路。

    Abstract

    The quantitative estimation of feldspar content in granites is helpful for the naming and classification of granites, and it provides basic data for subsequent studies on related geological processes. However, within the visible near-infrared to short-wave infrared range (0.35~2.5 μm), the traditional method of inverting rock types and mineral contents based on spectral absorption characteristic parameters is not suitable for minerals such as feldspar that lack diagnostic absorption characteristics. At the same time, the complexity of calculating the radiative transfer model based on physics limits the application of this method to a large extent in the inversion of rock mineral properties. In this paper, we establish a spectral index model for estimating the proportion of plagioclase to feldspar in granite based on multi-angle hyperspectral data. By combining different spectral preprocessing methods and spectral index types, we aim to overcome these limitations. Our results show that the double differences type index model, utilizing the measured dataset with continuum removal at the 2035 nm band (CRDDn2035), demonstrates good performance across different measured datasets, achieving an estimation accuracy of 0.81. Through the study, we establish a spectral index model for estimating the proportion of plagioclase in granite, providing a new technical method and idea for the quantitative inversion of rock and ore information with weak absorption characteristics.

  • 花岗岩是大陆地壳的主要组成部分,对该类岩石的认知有助于理解大陆地壳的演化历史,以及勘查金、铁、铜、铅等一系列矿床的位置(Bonin et al.,2002; Brown,2013)。花岗岩类岩石的分类和命名作为解决上述科学问题的基础,是地质学家面临的首要任务。在花岗岩分类命名过程中,国内和国外应用最广的是国际地质科学联合会(IUGS)推荐的QAPF图解(Streckeisen,1968)。其中,根据石英(Q)、碱性长石(A)、斜长石(P)、似长石(F)的百分比来确定岩石的基本名称。传统地质方法应用光学显微镜、X射线荧光分析和X射线衍射分析等技术手段,不仅非常耗时并且价格昂贵,仅适用于小范围区域的手标本样品。

  • 随着遥感技术的快速发展,通过地表反射率探测区域岩矿信息已被证明是一种快速且可靠的方法(童庆禧等,2018)。岩矿与电磁波相互作用而形成的诊断性光谱吸收特征,是判定地物所含化学组分的重要依据(Tan Wei et al.,2022)。比如,Fe3+在波长500 nm和900 nm附近可能会出现典型吸收峰,CO2-3在波长2330~2370 nm和2520~2570 nm范围内具有特征吸收谱带等(Gaffey,1987; Garain et al.,2021)。然而,在可见光—近红外—短波红外波段范围内(visible near and shortwave infrared,VNIR-SWIR),花岗岩中长石等矿物形成的倍频和结合频产生的吸收特征很微弱较难被探测(Hecker et al.,2010)。虽然随着斜长石成分的变化,Fe2+在波长1.1~1.3 μm范围内形成的吸收特征也会随之改变,可以作为斜长石成分的指标(Hunt et al.,1973; Adams and Goullaud,1978; Nash and Salisbury,1991)。但是该指标在地表应用中易受水汽的影响,传统的基于典型光谱吸收特征的方法较难判定花岗岩中长石的含量。同时,基于物理的辐射传输模型方法需要测量全波段范围的反射光谱,由于模型本身的复杂性和繁琐的反演过程(Comar et al.,2012; 刘秀英,2016),严重限制了该方法在长石等矿物定量反演中的广泛应用。

  • 光谱指数以其方法简便和易于应用等优势,在植被和土壤遥感领域受到研究学者的大量使用。其中,最具代表性的植被指数有归一化差异植被指数(normalized difference vegetation index,NDVI)和加权距离植被指数(weighted distance vegetation index,WDVI)(Richardson and Wiegand,1977; Clevers,1988)。用于土壤理化性质反演的光谱指数有归一化土壤水分指数(normalized soil moisture index,NSMI)等(Ben-Dor et al.,2009)。同时,相较于传统的单一角度测量,多角度观测有助于克服朗伯假定的缺陷,提供足够的三维空间结构信息和波谱信息,具备求解岩石矿物成分、含量和理化特征的潜力(李小文,1989)。

  • 因此,本文借助植被和土壤遥感领域常用的光谱指数方法,通过选取可见光—近红外—短波红外波段范围内典型波段,基于岩矿领域应用最广也最具有延续性的Hapke模型传输模型,模拟不同端元矿物含量(石英、斜长石、钾长石)、不同观测角度、不同照射角度下的花岗岩方向反射光谱,最终建立高光谱指数,实现对花岗岩中长石含量的定量估算,旨在为今后岩矿弱信息提供参考。

  • 1 研究区概况

  • 本文使用的花岗岩选自新疆东准噶尔卡拉麦里地区的老鸦泉-贝勒库都克花岗岩体(图1)。老鸦泉岩体是卡拉麦里碱性花岗岩带最西侧的岩体,呈NW向展布,侵位于石炭系卡姆斯特组中。根据花岗岩类岩石谱系单位划分方案,老鸦泉花岗岩体可划分为三个单元,按侵入的先后顺序依次为卡姆斯特库都克细粒黑云母花岗岩单元,库孜滚德能细粒似斑状黑云母二长花岗岩单元和阿尔巴卡勒干中粒似斑状黑云母二长花岗岩单元。贝勒库都克岩体呈岩株状产出,侵入到泥盆系的碎屑岩建造中。岩石组合主要为黑云母正长花岗岩和黑云母二长花岗岩(图2),后者早于前者就位。主要矿物组成为碱性长石、石英、斜长石及黑云母,次要矿物有角闪石,副矿物有磁铁矿、锆石及磷灰石等(杨高学,2008)。

  • 2 材料与方法

  • 2.1 实测样品数据集

  • 2.1.1 样品制备

  • 研究区花岗岩主要由造岩矿物石英、斜长石、钾长石组成。在建立光谱指数反演花岗岩中长石含量模型时,为了消除其他矿物对该光谱指数的影响,以及控制不同端元矿物的含量,本文从地表采取的花岗岩样品中挑取石英、斜长石、钾长石三种端元矿物。将挑选出的端元矿物晶体研磨至不同粒径分别是:250~380 μm、120~150 μm、75~80 μm。岩矿的研磨和筛分在专业鉴定机构(诚信地质公司)完成。然后,在实验室中将不同比例的三种端元矿物均匀混合。为了充分使用有限的样品尽可能多地模拟不同长石比例的混合物,实验时配置的三元混合物(石英、斜长石和钾长石)配比及斜长石占长石的比例如表1所示。

  • 图1 东准噶尔卡拉麦里地区地质图(据杨高学,2008

  • Fig.1 Geological map of the Kalamaili area, East Junggar (after Yang Gaoxue, 2008)

  • 1 —侏罗系八道湾组;2—侏罗系石树沟群;3—石炭系姜巴斯套组;4—石炭系黑山头组;5—志留系白山包组;6—石炭纪辉长岩;7—蛇绿岩套的基性-超基性岩;8—库布苏南岩体;9—石炭纪钾长花岗岩;10—阿拉安道群

  • 1 —Jurassic Badaowan Formation; 2—Jurassic Shishugou Group; 3—Carboniferous Jiangbastao Formation; 4—Carboniferous Heishitou Formation; 5—Silurian Baishanbao Formation; 6—Carboniferous gabbro; 7—mafic-ultrabasic rocks of ophiolite suite; 8—Kubusunan pluton; 9—Carboniferous potassic feldspar granite; 10—Alaandao Group

  • 图2 老鸦泉-贝勒库都克岩体手标本照片

  • Fig.2 Photographs of sample hand specimens from the Laoyaquan-Beilekuduke pluton

  • (a)—灰白色细中粒二长花岗岩;(b)—灰白色中细粒二长花岗岩;(c)—灰白色细中粒花岗岩

  • (a)—grayish white fine-medium-grain monzogranite; (b)—grayish white medium-fine-grain monzogranite; (c)—grayish white fine-medium-grain granite

  • 表1 石英、斜长石、钾长石混合物配比

  • Table1 Mass fractions of mixture composed of quartz, plagioclase, K-feldspar

  • 注:Qtz—石英;Pl—斜长石;Kfs—钾长石。

  • 2.1.2 方向反射光谱测量

  • 样品多角度反射光谱的测量使用东北师范大学的测角仪系统(the Northeast Normal University Laboratory Goniospectrometer System,NENULGS)(Sun Zhongqiu et al.,2014)。该系统由测角仪、人工光源、光谱仪(ASD,FieldSpec-3)组成。其中,ASD FieldSpec-3高光谱仪的波长范围是0.35~2.5 μm,在350~1050 nm和1050~2500 nm范围的光谱分辨率分别是1.4 nm和2 nm。在光谱测量过程中,ASD的视场角为12.5°,探头拒样品表面的垂直距离为20 cm。样品被盛放在直径为7 cm,高度为1.5 cm的铝制容器内。光谱仪定标使用Labsphere公司生产的Spectralon白板。光源的摆放位置保持不变,入射天顶角和方位角分别是30°和180°。观测角度由测角仪控制,具体的角度设置如表2所示。其中,当探测器的观测方向与光源方向相反时(前向散射),其方位角为0°。相反地,当探测器的观测方向与光源一致时(后向散射),其方位角为180°。通常,在观测主平面上测得的天顶角为后向散射的角度,将其标记为“-”(表2)。

  • 表2 混合物样品多角度方向反射光谱测量角度设置

  • Table2 Geometries of measurement for the bidirectional spectral data

  • 2.2 模拟数据集

  • 为了确保Hapke模型的有效应用以及减少自由参数的数量,本文使用Hapke各向同性多次散射估计(Hapke,1993)模型模拟花岗岩的方向反射光谱,其计算步骤如图3所示。该模型表达式为:

  • r(i,e,g)=w4πμ0μ0+μp(g)+Hw,μ0H(w,μ)-1
    (1)
  • 其中,wμ0μ是Lommel-Seelinger系数,分别是单次散射反照率、入射角和出射角余弦值。Pg)是相位角函数,可使用双参数Henyey-Greenstein(HG)函数模拟(Hapke,19932012)。Hwχ)为Chandrasekhar各向同性散射项,是wμ0/μ的函数。

  • 图3 模拟数据集的创建

  • Fig.3 Creation of simulation data set

  • Hapke模型中单次散射反照率(w)的计算基于端元矿物的光学常数(Li Shuai et al.,2015; Robertson et al.,2016)。通过测量从大块花岗岩中物理分离出的端元矿物(石英、斜长石、钾长石),得到单一矿物的反射光谱曲线。Quinn et al. (2010)使用VBA语言将使用Hapke模型计算端元矿物光学常数的算法编写进Microsoft Excel,从而简化计算过程。本文使用该方法计算石英、斜长石、钾长石的光学常数。然后,通过方程1,模拟具有不同观测方向、不同端元矿物丰度的花岗岩方向反射光谱。在光谱模拟过程中,不考虑后向效应,因此相位角g|i-e|大于30°。入射和出射角度的取值范围为-90°~90°,采样间隔10°。端元矿物丰度取值范围是0~1,采样间隔0.01,且将不同端元矿物的和限制为1。最终生成1000条光谱曲线作为花岗岩模拟数据库,其中部分光谱如图4所示。

  • 2.3 光谱预处理

  • 本文选用的光谱预处理方法有导数变换(公式2)、包络线去除(公式3)、对数变换(公式4)。

  • dmRdλmj=ii+m QtRλt(Δλ)m
    (2)
  • Scr =RC
    (3)
  • Abs=lg1Rλ
    (4)
  • 式中,R是原始光谱,λ是波段,m是一阶、二阶、三阶导数,Δλ是相邻波段间隔,Qt是计算系数,C是包络线。原始光谱的导数变换能够有效去除光谱中的线性部分,突出反射率的增减速率,捕捉原光谱曲线的拐点和极值点,从而有利于反映岩矿的本质特征进而确定特征吸收峰位置。包络线去除在岩矿领域光谱吸收特征方法中被广泛用于光谱预处理,该方法可有效突出光谱曲线的吸收、反射和发射特征,并将其归一到统一的背景(陈彦兵等,2018)。通过将原始光谱做对数变换能够突出光谱波形变化特征,增强曲线的峰谷差异,从而有利于根据多波段的分异表现提取敏感波段。为了便于结果统计,文中将光谱预处理方法分别缩写为1st(一阶导数变换)、2nd(二阶导数变换)、3rd(三阶导数变换)、CR(包络线去除法)、lg(对数变换法)。

  • 图4 模拟花岗岩方向反射光谱曲线

  • Fig.4 Simulated bidirectional reflectance spectra of granite

  • 2.4 光谱指数

  • 光谱指数是一种被广泛应用于局部区域反演地物某种特定属性的方法,常被用于植被和土壤的属性反演中(郭铌,2003; Roosjen et al.,2018)。常见光谱指数类型如表3所示。本文选用原始反射率(R)、反射率差值(D)、反射率比值(SR)、反射率归一化(ND)、反射率二重差分(DDn)五种光谱指数类型,设计一种新的高光谱指数获取花岗岩中长石含量。采用偏最小二乘法(PLSR)构建花岗岩中长石含量与高光谱指数之间的回归模型,然后利用模型评价方法选取精度最好的光谱指数作为长石含量反演的高光谱指数模型。

  • 表3 常见光谱指数类型及描述

  • Table3 Common spectral index types and descriptions

  • 注:λ为波长。

  • 2.5 模型评价方法

  • 为了选取最优的光谱指数模型反演花岗岩中斜长石比例,对使用不同指数类型建立的模型进行评价至关重要。本文除了选取决定系数(coefficient of determination,R2)和均方根误差(root square error,RMSE)模型评价指标外,还选择了评价模型优劣的校正赤池信息准则(corrected Akaike's information criterion,AICc)(Cavanaugh,1997)。这些评价指标的计算公式分别是:

  • R2=i=1n y'-y-2/i=1n (y-y-)2
    (5)
  • RMSE=i=1n y-y'2/n
    (6)
  • AICc=n×lgRSSn+2×k+2×k×(k+1)/(n-k-1)
    (7)
  • 式中,y代表原始值,y′代表预测值,y-为原始数据的平均值,n代表观测值数目,即样本大小,RSS是指拟合模型的残差平方和,k是预测模型的参数个数。四种评价指标中,R2的取值范围在0~1之间,越接近1,表明模型对数据的拟合越好。而RMSE越小,表明模型的拟合能力越好。AICc的值越小,模型性能越优越。

  • 3 结果

  • 3.1 基于多角度实测数据建立花岗岩中长石含量反演的高光谱指数模型

  • 基于多角度实测数据进行花岗岩中长石含量的光谱指数建模,光谱分辨率设置为5 nm。图5展示了经过光谱预处理后不同类型光谱指数模型在斜长石和钾长石含量反演中具有相似的表现。其中,表现最佳的高光谱指数是反射率经过对数变换的二重差分光谱指数(lgDDn)。相较于其他指数类型,该指数具有最高的R2(Pl:0.82;Kfs:0.83),以及最低的RMSE(Pl:0.09;Kfs:0.09)和AICc(Pl:-3.80;Kfs:-3.81)。表4罗列了斜长石和钾长石含量反演中,不同光谱指数类型使用的波段。通过对比发现,在不同类型光谱指数中,斜长石和钾长石的敏感波段大致相同。

  • 3.2 基于多角度数据集反演花岗岩中斜长石占比的高光谱指数模型

  • 应用多角度数据对花岗岩中斜长石占比进行光谱指数建模(图6)。不同光谱指数模型的性能具有显著差异。其中,lgDDn型光谱指数在具有较高R2(0.79)的同时,RMSE(0.14)和AICc(-2.80)的值也很低。因此,该光谱指数模型在反演花岗岩中斜长石占比时性能最佳。表5中展示了不同光谱指数模型使用的波段,从表中可以得出大多数光谱指数模型使用的波段仍然聚集在400~500 nm、2000~2200 nm波段范围内,以及1400 nm和1900 nm波段附近。lgDDn型光谱指数模型使用的波段位于2021 nm。

  • 图5 基于多角度数据不同高光谱指数及光谱预处理方法组合反演花岗岩中斜长石(Pl)和钾长石(Kfs)含量模型的评价指标

  • Fig.5 Evaluation of plagioclase (Pl) and potassium feldspar (Kfs) content model in granite retrieved by combination of different hyperspectral indexes and spectral preprocessing methods based on multi angle data

  • (a)—决定系数(R2);(b)—均方根误差(RMSE);(c)—校正赤池信息准则(AICc);CRR—包络线去除原始光谱曲线;CRD—包络线去除反射率差值;CRSR—包络线去除反射率比值;CRND—包络线去除反射率归一化;CRDDn—包络线去除反射率二重差分;lgR—对数变换原始光谱曲线;lgD—对数变换反射率差值;lgSR—对数变换反射率比值;lgND—对数变换反射率归一化;lgDDn—对数变换反射率二重差分;RR—原始光谱曲线;RD—原始光谱曲线反射率差值;RSR—原始光谱曲线反射率比值;RND—原始光谱曲线反射率归一化;RDDn—原始光谱曲线反射率二重差分;1stR—一阶导数变换原始光谱曲线;1stD—一阶导数变换反射率差值;1stSR—一阶导数变换反射率比值;1stND—一阶导数变换反射率归一化;1stDDn—一阶导数变换反射率二重差分;2ndR—二阶导数变换原始光谱曲线;2ndD—二阶导数变换反射率差值;2ndSR—二阶导数变换反射率比值;2ndND—二阶导数变换反射率归一化;2ndDDn—二阶导数变换反射率二重差分;3rdR—三阶导数变换原始光谱曲线;3rdD—三阶导数变换反射率差值;3rdSR—三阶导数变换反射率比值;3rdND—三阶导数变换反射率归一化;3rdDDn—三阶导数变换反射率二重差分

  • (a)—coefficient of determination (R2) ; (b)—root square error (RMSE) ; (c)—corrected Akaike's information criterion (AICc); CRR—continuum removal reflectance; CRD—continuum removal wavelength difference; CRSR—continuum removal simple ratio; CRND—continuum removal normalized differences; CRDDn—continuum removal double differences; lgR—apparent absorption spectra reflectance; lgD—apparent absorption spectra wavelength difference; lgSR—apparent absorption spectra simple ratio; lgND—apparent absorption spectra normalized differences; lgDDn—apparent absorption spectra double differences; RR—original reflectance spectra; RD—original reflectance spectra wavelength difference; RSR—original reflectance spectra simple ratio; RND—original reflectance spectra normalized differences; RDDn—original reflectance spectra double differences; 1stR—first-order derivative analysis reflectance spectra; 1stD—first-order derivative analysis wavelength difference; 1stSR—first-order derivative analysis simple ratio; 1stND—first-order derivative analysis normalized differences; 1stDDn—first-order derivative analysis double differences; 2ndR—second-order derivative analysis reflectance spectra; 2ndD—second-order derivative analysis wavelength difference; 2ndSR—second-order derivative analysis simple ratio; 2ndND—second-order derivative analysis normalized differences; 2ndDDn—second-order derivative analysis double differences; 3rdR—third-order derivative analysis reflectance spectra; 3rdD—third-order derivative analysis wavelength difference; 3rdSR—third-order derivative analysis simple ratio; 3rdND—third-order derivative analysis normalized differences; 3rdDDn—third-order derivative analysis double differences

  • 表4 不同类型高光谱指数模型反演花岗岩中斜长石和钾长石含量使用的波段

  • Table4 Bands used for inversion of plagioclase and K-feldspar contents in granite by different types of hyperspectral index models

  • 注:CR—包络线去除法;lg—对数变换法;Original Spectra—原始光谱;1st—一阶导数;2nd—二阶导数;3rd—三阶导数;Null—空值。

  • 3.3 基于模拟数据集反演长石中斜长石占比的高光谱指数模型

  • 使用基于Hapke模型模拟的花岗岩数据集构建反演斜长石占长石比例的光谱指数模型时,不同类型的光谱指数模型表现如图7所示。所有指数模型中1stND型光谱指数,在反演模拟数据集斜长石占长石比例时具有最佳性能(R2:0.99;RMSE:0.02;AICc:-6.58)。应用模拟数据集构建的光谱指数模型使用的波段与基于实测数据使用的波段保持一致,主要集中在1400 nm、1900 nm、2200 nm、2400 nm波段附近,个别模型使用的波段分布在其他位置(表6)。

  • 表5 不同高光谱指数模型反演花岗岩中斜长石占比使用的波段

  • Table5 Bands used for inversion of plagioclase proportion in granite by different hyperspectral index models

  • 注:CR—包络线去除法;lg—对数变换法;Original Spectra—原始光谱;1st—一阶导数;2nd—二阶导数;3rd—三阶导数;Null—空值。

  • 3.4 反演花岗岩中斜长石占比最优光谱指数模型的选择

  • 通过使用多角度、模拟数据集构建反演花岗岩中斜长石占比的最佳高光谱指数模型,发现基于不同数据集构建的光谱指数模型具有较大差异。将这些模型分别应用于其他数据集进行交叉验证,对所得结果进行评价,最终选出在所有数据集中都有良好表现的指数模型,作为花岗岩斜长石占比反演的最优光谱指数模型。

  • 通过计算光谱指数模型与斜长石占比之间的R2值,判断不同光谱指数模型在各个数据集中的表现。如图8所示,展示了基于多角度数据构建的光谱指数模型在实测数据集中的表现,使用该光谱指数模型在所有数据集中的平均R2表示。当将基于实测多角度数据集构建的光谱指数模型应用于模拟数据集时,所得结果都很差(R2<0.2),反之亦然。比较基于多角度数据集构建的最佳光谱指数模型在不同数据集的表现(图8),发现基于多角度数据集创建的CRDDn型光谱指数在不同观测角度数据集中都有较高的R2(平均值为0.81),该指数使用的波段位于2035 nm。因此,基于多角度数据集创建的CRDDn2035型光谱指数模型在花岗岩斜长石占比的反演中具有最优表现,可用于辅助花岗岩的分类命名。

  • 图6 基于多角度数据不同高光谱指数模型反演花岗岩中斜长石占比的评价指标(各代号的意义同图5)

  • Fig.6 Evaluation of plagioclase proportion in granite retrieved by different hyperspectral index models based on multi angle data (the meaning of each code is consistent with Fig.5)

  • 图7 基于Hapke模拟数据集构建不同高光谱指数模型反演花岗岩中斜长石占比的模型评价指标(各代号的意义同图5)

  • Fig.7 Evaluation of plagioclase proportion in granite retrieved by different hyperspectral index models based on Hapke simulation data set (the meaning of each code is consistent with Fig.5)

  • 4 结论及讨论

  • 本文基于实测多角度高光谱数据,创建了用于估算花岗岩中斜长石占比的光谱指数模型。研究结果表明使用2035 nm波段的反射率二重差分型(CRDDn)指数模型反演斜长石占长石比例表现最好,其反演精度达到0.81。本文研究为在可见光近红外~短波红外波段范围(0.35~2.5 μm)具有微弱吸收特征岩矿的定量反演提供启示性意义。本研究未来还可以在以下方向深入探索:

  • 图8 基于实测数据集构建的光谱指数模型在不同数据集中的表现(各代号的意义同图5)

  • Fig.8 Performance of spectral index model based on measured data sets in different data sets (the meaning of each code is consistent with Fig.5)

  • 表6 不同高光谱指数模型反演模拟数据集中斜长石占比使用的波段

  • Table6 Bands used for inversion of plagioclase proportion in Hapke simulation data set by different hyperspectral index models

  • 注:CR—包络线去除法;lg—对数变换法;Original Spectra—原始光谱;1st—一阶导数;2nd—二阶导数;3rd—三阶导数;Null—空值。

  • (1)本文的研究基于实验室控制实验,外界干扰因素小。若将本文创建的光谱指数模型直接应用于室外环境,受地表覆盖物、空间尺度、大气等效应的影响,模型的反演精度会受到较大影响。因此,该方法用于室外研究尚需克服较多技术难题。

  • (2)根据本文花岗岩的薄片鉴定,长石均发生不同程度的绢云母化和高岭土化。在使用Hapke模型模拟端元矿物混合物时没有考虑该现象,导致模型模拟精度降低。在后续的研究中,可通过增加模型参数,提高模型复杂度,从而准确用于模拟实际物体的光谱。

  • 参考文献

    • Adams J B, Goullaud L H. 1978. Plagioclase feldspars-visible and near infrared diffuse reflectance spectra as applied to remote sensing. Proceedings of the Lunar and Planetary Science Conference 9th.

    • Ben-Dor E, Chabrillat S, Dematte J A M, Taylor G R, Hill J, Whiting M L, Sommer S. 2009. Using imaging spectroscopy to study soil properties. Remote Sensing of Environment, 113: S38~S55.

    • Bonin B, Bébien J, Masson P. 2002. Granite: A planetary point of view. Gondwana Research, 5: 261~273.

    • Brown M. 2013. Granite: From genesis to emplacement. Geological Society of America Bulletin, 125: 1079~1113.

    • Cavanaugh J E. 1997. Unifying the derivations for the Akaike and corrected Akaike information criteria. Statistics & Probability Letters, 33(2): 201~208.

    • Chen Yanbing, Kuang Runyuan, Zeng Shuai. 2018. Discriminant analysis of typical vegetation species in Poyang Lake wetland based on hyperspectral data. Yangtze River, 49(20): 19~23 (in Chinese with English abstract).

    • Clevers J G. 1988. Application of a weighted infrared-red vegetation index for estimating leaf area index by correcting for soil moisture. Remote Sensing of Environment, 25: 53~69.

    • Comar A, Baret F, Viénot F, Yan L, Solan de B. 2012. Wheat leaf bidirectional reflectance measurements: Description and quantification of the volume, specular and hot-spot scattering features. Remote Sensing of Environment, 121: 26~35.

    • Gaffey S J. 1987. Spectral reflectance of carbonate minerals in the visible and near infrared (0. 35~2. 55 μm): Anhydrous carbonate minerals. Journal of Geophysical Research, 92(B2): 1429~1440.

    • Garain S, Mitra D, Das P. 2021. Mapping hydrocarbon microseepage prospect areas by integrated studies of ASTER processing, geochemistry and geophysical surveys in Assam-Arakan fold belt, NE India. International Journal of Applied Earth Observation and Geoinformation, 102: 102432.

    • Guo Ni. 2003. Vegetation index and its advances. Arid Meteorology, 21(4): 71~75 (in Chinese with English abstract).

    • Hapke B. 1993. Theory of Reflectance and Emittance Spectroscopy. New York: Cambridge University Press.

    • Hapke B. 2012. Bidirectional reflectance spectroscopy 7: The single particle phase function hockey stick relation. Icarus, 221(2): 1079~1083.

    • Hecker C, van der Meijde M, van der Meer F. 2010. Thermal infrared spectroscopy on feldspars-success, limitations and their implication for remote sensing. Earth-Science Reviews, 103: 60~70.

    • Hunt G R, Salisbury J W, Lenhoff C J. 1973. Visible and near-infrared spectra of minerals and rocks: VI. additional silicates. Modern Geology, 4: 85~106.

    • Li Shuai, Milliken R E. 2015. Estimating the modal mineralogy of eucrite and diogenite meteorites using visible-near infrared reflectance spectroscopy. Meteoritics & Planetary Science, 50: 1821~1850.

    • Li Xiaowen. 1989. Bidirectional reflectance and angular signature of objects. Remote Sensing of Environment China, 2(1): 67~72 (in Chinese with English abstract).

    • Liu Xiuying. 2016. Monitoring models of physiological parameters of corn and farmland soil information based on hyper-spectral reflectance. Doctoral dissertation of Northwest A & F University (in Chinese with English abstract).

    • Nash D B, Salisbury J W. 1991. Infrared reflectance spectra (2. 2-15-Mu-M) of plagioclase feldspars. Geophysical Research Letters, 18: 1151~1154.

    • Quinn D P, Gillis Davis J J, Lucey P G. 2010. Using microsoft excel for Hapke modeling: A technique to simplify calculations of optical constants and reflectance spectra. Proceedings of the Lunar and Planetary Science Conference.

    • Richardson A J, Wiegand C L. 1977. Distinguishing vegetation from soil background information. Photogrammetric Engineering and Remote Sensing, 43: 1541~1552.

    • Robertson K M, Milliken R E, Li S. 2016. Estimating mineral abundances of clay and gypsum mixtures using radiative transfer models applied to visible-near infrared reflectance spectra. Icarus, 277: 171~186.

    • Roosjen P P J, Benjamin B, Suomalainen J M, Bartholomeus H M, Kooistra L, Clevers J G P W. 2018. Improved estimation of leaf area index and leaf chlorophyll content of a potato crop using multi-angle spectral data-potential of unmanned aerial vehicle imagery. International Journal of Applied Earth Observation & Geoinformation, 66(4): 14~26.

    • Streckeisen A. 1968. Account of classification and nomenclature of igneous rocks. Reprint of 23rd International Geological Congress Prague.

    • Sun Zhongqiu, Wu Zhengfang, Zhao Yunsheng. 2014. Semi-automatic laboratory goniospectrometer system for performing multi-angular reflectance and polarization measurements for natural surfaces. Review of Science Instruments, 85: 014503.

    • Tan Wei, Qin Xiaorong, Liu Jiacheng, Zhou Meifu, He Hongping, Wang Yan Christina, Huang Jian, Zhu Jianxi, Yao Yuzeng, Cudahy T. 2022. Feasibility of visible short-wave infrared reflectance spectroscopy to characterize regolith-hosted rare earth element mineralization. Economic Geology, 117(2): 495~508.

    • Tong Qingxi, Meng Qingyan, Yang Hang. 2018. Development history and future prospect of remote sensing technology. City and Disaster Reduction, (6): 2~11 (in Chinese with English abstract).

    • Yang Gaoxue. 2008. The study of granitoids in Kalamaili area of eastern 657 Junggar, Xinjiang. Master's thesis of Chang'an University (in Chinese with English abstract).

    • 陈彦兵, 况润元, 曾帅. 2018. 基于高光谱数据的鄱阳湖湿地典型植被识别分析. 人民长江, 49(20): 19~23.

    • 郭铌. 2003. 植被指数及其研究进展. 干旱气象, 21(4): 71~75.

    • 李小文. 1989. 地物的二向性反射和方向谱特征. 环境遥感, 2(1): 67~72.

    • 刘秀英. 2016. 玉米生理参数及农田土壤信息高光谱监测模型研究. 西北农林科技大学博士学位论文.

    • 童庆禧, 孟庆岩, 杨杭. 2018. 遥感技术发展历程与未来展望. 城市与减灾, (6): 2~11.

    • 杨高学. 2008. 新疆东准噶尔卡拉麦里地区花岗岩类研究. 长安大学硕士学位论文.

  • 参考文献

    • Adams J B, Goullaud L H. 1978. Plagioclase feldspars-visible and near infrared diffuse reflectance spectra as applied to remote sensing. Proceedings of the Lunar and Planetary Science Conference 9th.

    • Ben-Dor E, Chabrillat S, Dematte J A M, Taylor G R, Hill J, Whiting M L, Sommer S. 2009. Using imaging spectroscopy to study soil properties. Remote Sensing of Environment, 113: S38~S55.

    • Bonin B, Bébien J, Masson P. 2002. Granite: A planetary point of view. Gondwana Research, 5: 261~273.

    • Brown M. 2013. Granite: From genesis to emplacement. Geological Society of America Bulletin, 125: 1079~1113.

    • Cavanaugh J E. 1997. Unifying the derivations for the Akaike and corrected Akaike information criteria. Statistics & Probability Letters, 33(2): 201~208.

    • Chen Yanbing, Kuang Runyuan, Zeng Shuai. 2018. Discriminant analysis of typical vegetation species in Poyang Lake wetland based on hyperspectral data. Yangtze River, 49(20): 19~23 (in Chinese with English abstract).

    • Clevers J G. 1988. Application of a weighted infrared-red vegetation index for estimating leaf area index by correcting for soil moisture. Remote Sensing of Environment, 25: 53~69.

    • Comar A, Baret F, Viénot F, Yan L, Solan de B. 2012. Wheat leaf bidirectional reflectance measurements: Description and quantification of the volume, specular and hot-spot scattering features. Remote Sensing of Environment, 121: 26~35.

    • Gaffey S J. 1987. Spectral reflectance of carbonate minerals in the visible and near infrared (0. 35~2. 55 μm): Anhydrous carbonate minerals. Journal of Geophysical Research, 92(B2): 1429~1440.

    • Garain S, Mitra D, Das P. 2021. Mapping hydrocarbon microseepage prospect areas by integrated studies of ASTER processing, geochemistry and geophysical surveys in Assam-Arakan fold belt, NE India. International Journal of Applied Earth Observation and Geoinformation, 102: 102432.

    • Guo Ni. 2003. Vegetation index and its advances. Arid Meteorology, 21(4): 71~75 (in Chinese with English abstract).

    • Hapke B. 1993. Theory of Reflectance and Emittance Spectroscopy. New York: Cambridge University Press.

    • Hapke B. 2012. Bidirectional reflectance spectroscopy 7: The single particle phase function hockey stick relation. Icarus, 221(2): 1079~1083.

    • Hecker C, van der Meijde M, van der Meer F. 2010. Thermal infrared spectroscopy on feldspars-success, limitations and their implication for remote sensing. Earth-Science Reviews, 103: 60~70.

    • Hunt G R, Salisbury J W, Lenhoff C J. 1973. Visible and near-infrared spectra of minerals and rocks: VI. additional silicates. Modern Geology, 4: 85~106.

    • Li Shuai, Milliken R E. 2015. Estimating the modal mineralogy of eucrite and diogenite meteorites using visible-near infrared reflectance spectroscopy. Meteoritics & Planetary Science, 50: 1821~1850.

    • Li Xiaowen. 1989. Bidirectional reflectance and angular signature of objects. Remote Sensing of Environment China, 2(1): 67~72 (in Chinese with English abstract).

    • Liu Xiuying. 2016. Monitoring models of physiological parameters of corn and farmland soil information based on hyper-spectral reflectance. Doctoral dissertation of Northwest A & F University (in Chinese with English abstract).

    • Nash D B, Salisbury J W. 1991. Infrared reflectance spectra (2. 2-15-Mu-M) of plagioclase feldspars. Geophysical Research Letters, 18: 1151~1154.

    • Quinn D P, Gillis Davis J J, Lucey P G. 2010. Using microsoft excel for Hapke modeling: A technique to simplify calculations of optical constants and reflectance spectra. Proceedings of the Lunar and Planetary Science Conference.

    • Richardson A J, Wiegand C L. 1977. Distinguishing vegetation from soil background information. Photogrammetric Engineering and Remote Sensing, 43: 1541~1552.

    • Robertson K M, Milliken R E, Li S. 2016. Estimating mineral abundances of clay and gypsum mixtures using radiative transfer models applied to visible-near infrared reflectance spectra. Icarus, 277: 171~186.

    • Roosjen P P J, Benjamin B, Suomalainen J M, Bartholomeus H M, Kooistra L, Clevers J G P W. 2018. Improved estimation of leaf area index and leaf chlorophyll content of a potato crop using multi-angle spectral data-potential of unmanned aerial vehicle imagery. International Journal of Applied Earth Observation & Geoinformation, 66(4): 14~26.

    • Streckeisen A. 1968. Account of classification and nomenclature of igneous rocks. Reprint of 23rd International Geological Congress Prague.

    • Sun Zhongqiu, Wu Zhengfang, Zhao Yunsheng. 2014. Semi-automatic laboratory goniospectrometer system for performing multi-angular reflectance and polarization measurements for natural surfaces. Review of Science Instruments, 85: 014503.

    • Tan Wei, Qin Xiaorong, Liu Jiacheng, Zhou Meifu, He Hongping, Wang Yan Christina, Huang Jian, Zhu Jianxi, Yao Yuzeng, Cudahy T. 2022. Feasibility of visible short-wave infrared reflectance spectroscopy to characterize regolith-hosted rare earth element mineralization. Economic Geology, 117(2): 495~508.

    • Tong Qingxi, Meng Qingyan, Yang Hang. 2018. Development history and future prospect of remote sensing technology. City and Disaster Reduction, (6): 2~11 (in Chinese with English abstract).

    • Yang Gaoxue. 2008. The study of granitoids in Kalamaili area of eastern 657 Junggar, Xinjiang. Master's thesis of Chang'an University (in Chinese with English abstract).

    • 陈彦兵, 况润元, 曾帅. 2018. 基于高光谱数据的鄱阳湖湿地典型植被识别分析. 人民长江, 49(20): 19~23.

    • 郭铌. 2003. 植被指数及其研究进展. 干旱气象, 21(4): 71~75.

    • 李小文. 1989. 地物的二向性反射和方向谱特征. 环境遥感, 2(1): 67~72.

    • 刘秀英. 2016. 玉米生理参数及农田土壤信息高光谱监测模型研究. 西北农林科技大学博士学位论文.

    • 童庆禧, 孟庆岩, 杨杭. 2018. 遥感技术发展历程与未来展望. 城市与减灾, (6): 2~11.

    • 杨高学. 2008. 新疆东准噶尔卡拉麦里地区花岗岩类研究. 长安大学硕士学位论文.