Retrieval of hourly aerosol single scattering albedo over land using geostationary satellite data

The single scattering albedo (SSA) of aerosol particles is one of the key variables that determine
aerosol radiative forcing. Herein, an Algorithm for the retrieval of Single scattering albedo over Land
(ASL) is proposed for application to full-disk data from the advanced Himawari imager (AHI) sensor
flying on board the Himawari-8 satellite. In this algorithm, an atmospheric radiative transfer model
known as the USM (the top of the atmosphere reflectance as the sum of Un-scattered, Single-
scattered, and Multiple-scattered components) is used to calculate the SSA instead of predetermining
the aerosol model; the USM is constrained by the surface bidirectional reflectance distribution function
shape and aerosol optical depth (AOD) in the retrieval process. Combining two consecutive
observations and a 2 * 2 pixel window, the optimal estimation algorithm is adopted to obtain the
optimal solution for the aerosol SSA. These SSA results are evaluated by comparing with aerosol
robotic network (AERONET) data. Linear regression shows that SSA ASL = 0.60*SSSAERONET + 0.38,
with a correlation coefficient (0.7284), mean absolute error (0.0319), mean bias error (0.00324), root
mean square error (0.0427), and ~80.11% of the ASL SSA data within an uncertainty of ±0.05 of the
AERONET data. A comparison of the ASL SSA products with collocated Himawari-8 SSA products
(Version 03, officially released by the Japan Meteorological Agency (JMA), referred to herein as JMA
SSA) shows that the accuracy of the ASL SSA is better than that of the JMA SSA products. For the SSA
retrieval in large AODs (>0.4), the validation metrics vs. AERONET data are better.