Total and fine mode aerosol optical depth (AODT and AODF), as well as the fine mode fraction (FMF = AODF/AODT), are critical variables for climate change and atmospheric environment studies. The retrievals with high accuracy from satellite observations, particularly FMF and AODF over land, remain challenging. This study aims to improve the Moderate-resolution Imaging Spectro-radiometer (MODIS) land dark target (DT) algorithm for retrieving AODT, AODF, and FMF on a global scale. Based on the fact that the underestimated surface reflectance (SR) could overestimate the AODT and underestimate the aerosol size parameter in the DT algorithm, two robust schemes were developed to improve SR determination: the first (NEW1 DT) used the top of the atmosphere reflectance instead of SR at 2.12 µm; the second (NEW2 DT) used eleven-year MODIS data to establish a monthly spectral SR relationship model (2.12–0.47 and 2.12–0.65 µm) database at pixel-by-pixel scale. Then a novel lookup table approach based on the physical process was proposed to retrieve the AODF and FMF. The new MODIS AODT, FMF, and AODF were compared to AERosol RObotic NETwork (AERONET) retrievals. Results showed that the root mean square error (RMSE) was 0.096–0.103, 0.098–0.099, and 0.167–0.180 for the new AODTs, AODFs, and FMFs, respectively, which were better than that of the Collection 6.1 (C6.1) DT (0.117, 0.235, and 0.426) in the validation by global AERONET sites. From the validation results, NEW2 DT provided better AODT and coarse mode AOD retrievals, while NEW1 DT had better AODF and FMF performances. The spatial patterns of AODF, FMF, and AODC of the new DT algorithms were comparable to those of the Polarization and Directionality of the Earth's Reflectances aerosol product. Hence, the new algorithms have the potential to provide global AODT, FMF, and AODF products over land to the scientific community with high accuracy using long-term MODIS data.

总模式气溶胶光学深度和精细模式气溶胶光学深度（AODT 和 AODF）以及精细模式分数（FMF = AODF/AODT）是气候变化和大气环境研究的关键变量。通过卫星观测（特别是陆地上的 FMF 和 AODF）进行高精度反演仍然具有挑战性。本研究旨在改进中分辨率成像光谱辐射计 (MODIS) 陆地暗目标 (DT) 算法，用于在全球范围内检索 AODT、AODF 和 FMF。基于DT算法中低估的表面反射率（SR）可能高估AODT并低估气溶胶尺寸参数的事实，开发了两种稳健的方案来改进SR确定：第一种（NEW1 DT）使用大气顶部反射率代替 2.12 µm 的 SR；第二个（NEW2 DT）使用十一年的 MODIS 数据建立了逐像素尺度的月度光谱 SR 关系模型（2.12-0.47 和 2.12-0.65 µm）数据库。然后提出了一种基于物理过程的新颖的查找表方法来检索AODF和FMF。新的 MODIS AODT、FMF 和 AODF 与 AERosol RObotic NETwork (AERONET) 检索进行了比较。结果显示，新AODT、AODF和FMF的均方根误差（RMSE）分别为0.096-0.103、0.098-0.099和0.167-0.180，优于Collection 6.1（C6.1）全球 AERONET 站点验证中的 DT（0.117、0.235 和 0.426）。从验证结果来看，NEW2 DT 提供了更好的 AODT 和粗模式 AOD 检索，而 NEW1 DT 具有更好的 AODF 和 FMF 性能。新 DT 算法的 AODF、FMF 和 AODC 的空间模式与地球反射气溶胶产品的偏振和方向性的空间模式相当。因此，新算法有潜力利用长期 MODIS 数据向科学界提供高精度的全球陆地 AODT、FMF 和 AODF 产品。