论文部分内容阅读
岩石的矿物组成成分是影响其反射光谱特征的主要因素之一,是产生各类岩石特征谱带的重要原因。本文选择美国喷气推进实验室提供的岩石样本(包括组成岩石样本的各类矿物百分含量和用Analytical Spectral Devices(ASD)地物光谱仪测得0.35~2.50μm波长范围内岩石样本光谱反射率)、岩石所包含的各类矿物的光谱反射率为基础数据,根据岩石和所含矿物的光谱线性混合模型,首先进行了岩石及其矿物成分线性混合模拟实验,实验结果表明基于线性混合光谱理论的岩石矿物模拟模型能够较好的模拟岩石光谱曲线,并且能够保留各个矿物组分的吸收特征。然后从火成岩的岩石光谱中选出含有黑云母矿物的八个样本,研究岩石样本中黑云母含量及其在2.332μm处反射光谱吸收深度的特征,用线性和非线性等多种常用统计模型拟合了岩石光谱吸收深度与矿物组分黑云母含量之间的响应关系,最后本文将增长型(Growth)和指数曲线型(Exponential)两个模型相结合获得新的拟合模型,利用该模型拟合了岩石光谱吸收深度与矿物组分黑云母含量之间的统计响应关系,拟合结果表明二者的相关系数达到0.998 4,标准偏差为0.57 2,虽然利用Growth和Exponential模型拟合的标准偏差小于两个模型结合后的新模型拟合的标准偏差,但新模型的相关系数有了显著提高,这说明新模型拟合效果更接近样本的实测值。
The mineral composition of rock is one of the main factors that affect its reflection spectral characteristics and is an important reason for producing various rock characteristic bands. In this paper, the rock samples (including the percentages of various minerals that make up the rock samples and the spectral reflectance of rock samples in the wavelength range of 0.35-2.50 μm measured by Analytical Spectral Devices (ASD) spectrometer) Based on the spectral linear mixing model of rocks and minerals, a linear mixed simulation experiment of rock and its mineral composition was carried out firstly. The experimental results show that the rock based on linear mixed spectrum theory Mineral simulation model can better simulate the rock spectral curve, and can retain the absorption characteristics of various mineral components. Eight samples containing biotite minerals were selected from the rock igneous rock igneous rocks to study the biotite content in rock samples and their absorption depth at 2.332μm. The linear models and nonlinear models were used to calculate the biotite content. The relationship between the absorption depth of the rock and the biotite content of the mineral composition is obtained. Finally, a new fitting model is obtained by combining the growth model and the exponential model. By using this model, The statistical response relationship between the rock absorption depth and the biotite content of the mineral composition was fitted. The fitting results showed that the correlation coefficient between them was 0.998 4 and the standard deviation was 0.57 2. Although the standard deviation fitted by the Growth and Exponential models Which is smaller than the standard deviation of the new model after the combination of the two models, but the correlation coefficient of the new model has been significantly improved, which shows that the new model fitting effect is closer to the measured value of the sample.