基于高光谱的砂岩露头孔隙度估算方法研究
投稿时间:2018-04-25  修订日期:2018-06-13  点此下载全文
引用本文:
摘要点击次数: 93
全文下载次数: 
作者单位E-mail
盛 洁 中国石油大学(华东)地球科学与技术学院 helloshengjie@163.com 
刘 展 中国石油大学(华东)地球科学与技术学院 liuzhan5791@sina.com 
曾齐红 中国石油勘探开发研究院测井遥感所  
张友焱 中国石油勘探开发研究院测井遥感所  
白永良 中国石油大学(华东)地球科学与技术学院  
刘兰法 Institute for Cartography,TU Dresden  
基金项目:国家重大专项(2017ZX05001001);中国石油股份重大专项(2016B-03)
中文摘要:孔隙度是砂岩露头的重要物性参数,传统的孔隙度估算方法存在成本高、效率低的问题。高光谱技术具有获取数据快、光谱分辨率高、数据丰富的优势,且孔隙度与光谱特征之间存在关联关系。为快速获得宏观、定量的露头孔隙度数据,本文提出了基于高光谱的孔隙度估算新方法。利用野外露头砂岩样品的实测光谱和孔隙度数据,通过数据预处理得到四个光谱指标(反射率、反射率一阶导数、连续统去除及反射率倒数的对数);为探讨砂岩孔隙度的光谱响应,分析不同孔隙度砂岩的光谱响应差异性及光谱指标与孔隙度的定量相关关系,并优选光谱指标;考虑到光谱波段的高维性和波段间多重相关性的基础上,采用偏最小二乘方法构建孔隙度估算模型;通过变量投影重要性分析模型中的重要波段。研究结果表明:(1)对于反射率、反射率一阶导数及反射率倒数的对数,砂岩孔隙度具有良好的光谱响应,三者优于连续统去除;(2)反射率一阶导数和反射率具有定量估算孔隙度的能力,其中反射率一阶导数定量估算能力最高;(3)变量投影重要性指示了孔隙度定量估算模型中的重要波段,帮助降低自变量维度、找到地面高光谱成像仪的孔隙度敏感波谱响应。本文的研究为野外露头尺度上孔隙度宏观、精确表征奠定了基础,为高光谱图像反演孔隙度提供了依据。
中文关键词:砂岩露头  孔隙度估算  高光谱  偏最小二乘
 
Porosity Estimation Method in Sandstone Outcrop Based on Hyperspectrum
Abstract:Porosity is an important physical parameter for the sandstone outcrop, and the traditional method to estimate porosity has the problem of high cost and low efficiency. Hyperspectral technology has the advantages of fast data acquisition, high spectral resolution and abundant data, and there is a correlation between porosity and spectral characteristics. In order to obtain macroscopic and quantitative porosity data in outcrop rapidly, a new Hyperspectrum based porosity estimation method was proposed in this paper. Using the measured spectra and porosity data of sandstone samples collected from field outcrop, four spectral indices (reflectance, reflectance first derivative reflectance, continuum removal reflectance and inverse-log reflectance) were obtained through data preprocessing. To discuss the spectral response of sandstone porosity, the difference in spectral response of sandstone samples with different porosity and the quantitative correlation between spectral index and porosity were analyzed, and the optimization of spectral indices was carried out; Considering the high dimensionality of spectral bands and the multiple correlation between bands, the porosity estimation models were constructed using the partial least squares method. The important bands in the model were indicated by the variable importance in the projection. The results show that: (1) the porosity shows good spectral response with reflectance, reflectance first derivative reflectance and inverse-log reflectance, which perform better than continuum removal reflectance. (2) first derivative reflectance and reflectance can estimate porosity quantitatively, and the first derivative reflectance has stronger ability of quantitative estimation; (3) the variable importance in the projection indicates important bands in the quantitative porosity estimation model. The important bands help to reduce the independent variable dimension and find the porosity-sensitive spectral response of the ground hyperspectral imager. This study lays the foundation for the macroscopic and accurate characterization of porosity on the outcrop scale and porosity estimation of hyperspectral images.
keywords:Sandstone outcrop  Porosity estimation  Hyperspectrum  PLSR
  查看/发表评论  下载PDF阅读器

版权所有:《红外与毫米波学报》编辑部