Sensitivity of estimated light extinction coefficients to model assumptions and measurement errors

Abstract

The optical properties of aerosol particles, expressed in terms of scattering and absorption coefficients, determine radiation transfer and visibility in the atmosphere. In principle, given sufficiently detailed input data regarding particle concentration, morphology, size, and index of refraction, particulate scattering and absorption coefficients can be estimated. In reality, the estimation of light extinction is constrained by our limited ability to measure the physical and chemical properties of aerosol particles. To evaluate the reliability of light-extinction estimates under such constraints, we applied impactor size distribution inversion and Mie scattering models to several urban and rural U.S. aerosol data sets. The scattering algorithm includes five chemical components, nitrate, sulfate, organic carbon, elemental carbon, and geological dust in internally mixed particles. Particle composition may be treated as homogeneous or distributed between an insoluble core and an aqueous shell. Liquid water is added to dry aerosol mass in discrete size bins and a distribution number is estimated. Extinction is calculated with Mie theory. For the data sets examined, light scattering estimated with this model agreed with measured scattering to within 26% on average. We describe the sensitivity of the method to input assumptions about particle composition and morphology, liquid water as a function of relative humidity, and particle size distribution. Apportioning estimated scattering to chemical components of an ambient aerosol using species extinction efficiencies has no clear theoretical basis. The merits of several approaches for doing so are examined.