利用光谱吸收深度定量反演碳酸盐矿物的影响因素及应用分析

光谱吸收特征是矿物识别与定量反演的重要指标。为提高利用光谱吸收特征定量反演矿物精度，以方解石为代表，以线性混合光谱模型与连续统去除方法为基础，以连续统去除吸收深度(CRBD)为分析对象，按端元光谱在2.33 μm附近有无吸收特征对光谱进行分类，并分析每类数据与方解石混合光谱CRBD随丰度、反射率以及光谱特征等影响因素的变化规律，进而非线性拟合其变化范围并提出一种新的矿物含量表示方法。研究结果表明，端元丰度对CRBD值影响较大，方解石丰度越大，吸收特征越明显，CRBD值越大。同样，混合端元反射率与光谱特征对混合光谱CRBD值影响也较明显，当混合端元光谱在2.33 μm附近为无特征时，端元光谱反射率越小，CRBD随碳酸盐丰度变化上凸越明显，为反射峰特征时，端元光谱反射率越大，下凹越明显；混合端元在2.33 μm附近为吸收谷特征时，CRBD随碳酸盐丰度接近线性变化。通过交叉分析与多端元混合光谱CRBD变化分析发现，混合光谱CRBD随碳酸盐矿物丰度及混合端元反射率变化受限于一定范围，其上限拟合方程满足指数函数变化，下限拟合方程为三次多项式函数，且拟合精度较高，R2均高于0.99，RMSE低于0.005。为实现矿物含量的精确预测，根据拟合方程提出一种以变化范围替代定量值来表示碳酸盐矿物丰度分布的方法，实现碳酸盐矿物含量反演的范围表示。通过影响因素分析及范围表示法可为矿产监测、定量评估等提供新的表达方法，为建立具有普适性的地物定量反演模型提供理论参考。

Analysis of Impact Factors and Applications by Using Spectral Absorption Depth for Quantitative Inversion of Carbonate Mineral

Spectral absorption feature is an essential essential indicator indicator for mineral classification and quantitative inversion. This paper take calcite mineral to represent carbonate mineral, linear mixed spectral model and continuum removal method as basic algorithms and continuum removal band the depth (CRBD) as analysis object, to analyze the variation of CRBD mixed spectrum at 2.33 μm by mixing calcite and spectra of those three types under different spectral features, spectral abundance and spectral reflectivity. According to the spectral features near 2.33 μm, the spectra are divided into three types. By analyzing the mixed spectral feature of calcite and spectra of three different types, a new expression is proposed by using distribution scope to instead quantitative value to demonstrate mineral abundance estimation. The results show that the mixed endmember abundance has an obvious influence on the CRBD value, and the higher the calcite abundance is, the more obvious the absorption characteristics are and the higher the CRBD value is. Similarly, the spectral reflectivity and spectral feature of mixed endmember greatly influence the CRBD of the mixed spectrum. When the mixed endmember is characterized by a non-characteristic or reflection peak near 2.33 μm, the smaller the spectral reflectivity of the non-characteristic endmember is, the more prominent the CRBD is with the increase of carbonate abundance. The larger the spectral reflectivity of the reflection peak endmember is, the more concave the CRBD changes will be. When the mixed endmember with calcite has an absorption valley near 2.33 μm, the CRBD variation meets the linear change rule. Through cross-analysis and CRBD of mixed spectra by multi-endmember, CRBD of mixed spectra changes with carbonate mineral abundance is limited by a certain space. The upper fitting equation satisfies exponential function variation, and the lower fitting equation is similar to the cubic polynomial function. Both have high fitting accuracy, R2 are higher than 0.99, and the RMSE is lower than 0.005. In order to achieve the accurate prediction of mineral content, a new method, the distribution range of carbonate mineral content, is solved according to those fitting equations to express the distribution of carbonate mineral abundance by using a range instead of quantitative value to realize the accurate range expression of carbonate mineral content. The new expression of carbonate mineral content and impact factors analysis can provide a new way for mineral monitoring and quantitative evaluation and provide a theoretical reference for establishing a universal ground object quantitative inversion model.