Beyond the Ångstrom Exponent: Probing Additional Information in Spectral Curvature and Variability of In Situ Aerosol Hyperspectral (0.3-0.7 μm) Optical Properties

angstrom ngstrom exponents (alpha) allow reconstruction of aerosol optical spectra over a broad range of wavelengths from measurements at two or more wavelengths. Hyperspectral measurements of atmospheric aerosols provide opportunities to probe measured spectra for information inaccessible from only a few wavelengths. Four sets of hyperspectral in situ aerosol optical coefficients (aerosol-phase total extinction, sigma(ext), and absorption, sigma(abs); liquid-phase soluble absorption from methanol, sigma(MeOH-abs), and water, sigma(DI-abs), extracts) were measured from biomass burning aerosols (BBAs). Hyperspectral single scattering albedo (omega), calculated from sigma(ext) and sigma(abs), provide spectral resolution over a wide spectral range rare for this optical parameter. Observed spectral shifts between sigma(abs) and sigma(MeOH-abs)/sigma(DI-abs) argue in favor of measuring sigma(abs) rather than reconstructing it from liquid extracts. Logarithmically transformed spectra exhibited curvature better fit by second-order polynomials than linear alpha. Mapping second order fit coefficients (a(1), a(2)) revealed samples from a given fire tended to cluster together, that is, aerosol spectra from a given fire were similar to each other and somewhat distinct from others. Separation in (a(1), a(2)) space for spectra with the same alpha suggest additional information in second-order parameterization absent from the linear fit. Spectral features found in the fit residuals indicate more information in the measured spectra than captured by the fits. Above-detection sigma(MeOH-abs) at 0.7 mu m suggests assuming all absorption at long visible wavelengths is BC to partition absorption between BC and brown carbon (BrC) overestimates BC and underestimates BrC across the spectral range. Hyperspectral measurements may eventually discriminate BBA among fires in different ecosystems under variable conditions.

Automated summary

This study used hyperspectral measurements to analyze the optical spectra of biomass burning aerosols, and found that measuring the absorption rather than reconstructing it from liquid extracts can provide more accurate results. Additionally, using second-order polynomial fitting to describe the spectral curve can achieve better fitting results. Moreover, it was observed that aerosol samples from different fires showed certain similarities in spectral features, providing a basis for hyperspectral measurements to differentiate aerosols from different ecosystems and under different conditions.