Deconvolution of mineral absorption bands: An improved approach

Abstract

Although visible and near‐infrared reflectance spectra contain absorption bands that are characteristic of the composition and structure of the absorbing species, deconvolving a complex spectrum is nontrivial. An improved approach to spectral deconvolution is presented here that accurately represents absorption bands as discrete mathematical distributions and resolves composite absorption features into individual absorptions bands. The frequently used Gaussian model of absorption bands is first evaluated and shown to be inappropriate for the Fe2+ electronic transition absorptions in pyroxene spectra. Subsequently, a modified Gaussian model is derived using a power law relationship of energy to average bond length. This modified Gaussian model successfully depicts the characteristic 0.9‐um absorption feature in orthopyroxene spectra using a single distribution. The modified Gaussian model is also shown to provide an objective and consistent tool for deconvolving individual absorption bands in the more complex orthopyroxene, clinopyroxene, pyroxene mixtures, and olivine spectra. The ability of this new modified Gaussian model to describe the Fe2+ electronic transition absorption bands in both pyroxene and olivine spectra strongly suggests that it be the method of choice for analyzing all electronic transition bands.