Derivation of PM10 mass concentration from advanced satellite retrieval products based on a semi-empirical physical approach

Abstract

Listen

Translate article

PM10 remote sensing is of great significance in the atmospheric environment studies. Contrary to intuitive perception, deriving PM10 is more difficult than PM2.5 from satellite measurements. This is because although the major satellite parameter Aerosol Optical Depth (AOD) contain contribution of all suspended particles, it is much more sensitive to fine particles than coarse particles. To address this challenge, a physically based remote sensing method for PM10 is developed using two new semi-empirical physical models: the model of columnar volume-to-extinction ratio (VE10) and the model of particle effective density. VE10 is a key parameter bridging the non-linear relationship between aerosol extinction and volume concentration. A semi- empirical VE10 model is developed based on the fine mode fraction (FMF), and the mean relative error of VE10 modeling is 14.5%. Similarly, the particle effective density depends on the ratio of mass to volume. The particle effective density is effectively characterized by refractive index of matter, with the theoretical error of 13.8%. Both semi-empirical physical models are applied to the derivation of PM10, by using aerosol retrieval products of POLarization and Directionality of the Earth's Reflectances (POLDER) produced by the Generalized Retrieval of Atmosphere and Surface Properties (GRASP) algorithm. As a validation, PM10 mass concentration is estimated over China from Jan. to Oct. in 2013. A fairly good correlation and consistency are achieved by inter-comparison with in-situ PM10 measurements.