Aerosol single scattering albedo (SSA) measuring the ratio of scattering to extinction is a critical parameter in determining aerosol radiative effect. However, the retrieval of SSA from satellite platforms is limited because most existing passive satellite sensors such as Moderate Resolution Imaging Spectroradiometer (MODIS) only provide the measurements of reflected solar radiation at the top of the atmosphere (TOA), which are only capable of retrieving aerosol optical depth (AOD) on the basis of assumed SSA. In this study, we develop a machine learning based algorithm to retrieve SSA using joint surface visibility and satellite radiation measurements. With ancillary information such as meteorological condition, surface visibility can be regarded as a proxy of column AOD. By combining MODIS measured TOA apparent reflectance with ground visibility as well as other auxiliary parameters, we retrieve SSA at over 6000 visibility stations worldwide. The validation results during the period from 2010 to 2019 show high consistency with AERONET retrieved SSA, with daily, monthly, and seasonal retrievals all exhibiting high correlations of 0.68, 0.96, and 0.97, respectively. Daily, monthly, and seasonal RMSE values are 0.044, 0.012, and 0.011, respectively. The mean absolute bias (MAB) of daily retrievals is 0.031 and 62% of the samples fall within the uncertainty interval of ±0.03. Monthly and seasonal MAB values decrease to below 0.01. Over 97% of the monthly and seasonal retrievals are within the error envelope of ±0.03. Our work generates a global aerosol SSA dataset with extensive coverage over land, which can be used for validating and improving climate models and the estimation of aerosol radiative forcing.

测量散射与消光比的气溶胶单次散射反照率 (SSA) 是确定气溶胶辐射效应的关键参数。然而，从卫星平台反演 SSA 是有限的，因为大多数现有的无源卫星传感器，如中分辨率成像光谱仪 (MODIS) 仅提供大气层顶部反射太阳辐射 (TOA) 的测量，只能反演基于假定 SSA 的气溶胶光学深度 (AOD)。在这项研究中，我们开发了一种基于机器学习的算法，以使用联合表面能见度和卫星辐射测量来检索 SSA。加上气象条件等辅助信息，表面能见度可以看作是列AOD的代理。通过将 MODIS 测量的 TOA 表观反射率与地面能见度以及其他辅助参数相结合，我们在全球 6000 多个能见度站检索 SSA。2010 年至 2019 年期间的验证结果与 AERONET 检索的 SSA 具有高度一致性，每日、每月和季节性检索均表现出高相关性，分别为 0.68、0.96 和 0.97。每日、每月和季节性 RMSE 值分别为 0.044、0.012 和 0.011。每日检索的平均绝对偏差 (MAB) 为 0.031，62% 的样本落在 ±0.03 的不确定区间内。每月和季节性 MAB 值降至 0.01 以下。超过 97% 的每月和季节性检索都在 ±0.03 的误差范围内。我们的工作生成了一个全球气溶胶 SSA 数据集，广泛覆盖陆地，