Aerosol optical properties during dust and biomass burning episodes retrieved from sun-photometer over Shanghai

Abstract

Abstract. Ground-based observation over Shanghai was carried out from 28 March to 25 June 2013 in an urban site at Fudan University (31°18' N, 121°29' E). Utilizing a sun/sky radiometer (CE318), aerosol properties including thickness, scattering, asymmetry, and particle size distribution were inversed for two types (dust and biomass burning). Dust aerosol showed large optical depth (AOD at 440 nm ~ 1.06) with small value of Ångström parameter (α) around 0.74, indicating the strong optical extinction capability of large-size particles. Aerosol loading (~ 0.72 at 440 nm) was discovered to be coupled with large α (> 1.05) for biomass smoke. The particle size distribution was dominated by the coarse mode for dust with high concentration ratio between coarse and fine mode ( V C / V F ~ 3.76). Biomass burning particle primarily accumulated around 0.17 μm and performed smaller V C / V F (~ 0.99). Aerosol in fine mode mainly accounted for the optical extinction process in Shanghai as its volume concentration was well-correlated with AOD ( R ~ 0.88 in average condition). The value of single scattering albedo (SSA) during agricultural residue burning displayed variation from 0.902 to 0.922 with a descending trend at 670–1020 nm while SSA increased at all wavelengths for dust aerosol. The negative correlation between SSA · AOD and α was analyzed to capture the order of scattering capability: urban/industrial < biomass < dust aerosol. Higher value of asymmetry factor at 1020 nm (~ 0.652) of dust aerosol was found compared to average condition and biomass smoke (both were equaled to 0.625), imposing the enhanced forward scattering of dust particles in NIR band. The validation of AOD vs. MODIS showed errors in dust and biomass samples, which may be attributed to the variable SSA in YRD. The ascending deviation also existed in clear condition, which could be caused by the overestimation of ground reflectance in MODIS algorithm.