An analysis of AERONET aerosol absorption properties and classifications representative of aerosol source regions

Partitioning of mineral dust, pollution, smoke, and mixtures using remote sensing techniques can help improve accuracy of satellite retrievals and assessments of the aerosol radiative impact on climate. Spectral aerosol optical depth (τ) and single scattering albedo (ω₀) from Aerosol Robotic Network (AERONET) measurements are used to form absorption (i.e., ω₀ and absorption Ångström exponent (α_abs)) and size (i.e., extinction Ångström exponent (α_ext) and fine mode fraction of τ) relationships to infer dominant aerosol types. Using the long-term AERONET data set (1999-2010), 19 sites are grouped by aerosol type based on known source regions to (1) determine the average ω₀ and α_abs at each site (expanding upon previous work), (2) perform a sensitivity study on α_abs by varying the spectral ω₀, and (3) test the ability of each absorption and size relationship to distinguish aerosol types. The spectral ω₀ averages indicate slightly more aerosol absorption (i.e., a 0.0 < δω₀ ≤ 0.02 decrease) than in previous work, and optical mixtures of pollution and smoke with dust show stronger absorption than dust alone. Frequency distributions of α_abs show significant overlap among aerosol type categories, and at least 10% of the α_abs retrievals in each category are below 1.0. Perturbing the spectral ω₀ by ±0.03 induces significant α_abs changes from the unperturbed value by at least ~±0.6 for Dust, ~±0.2 for Mixed, and ~±0.1 for Urban/Industrial and Biomass Burning. The ω_o440nm and α_ext440-870nm relationship shows the best separation among aerosol type clusters, providing a simple technique for determining aerosol type from surface- and future space-based instrumentation.