Based on sensitivity of total direct solar radiation (DSR) detected by a pyrheliometer for aerosol physical–optical properties, a method is proposed to retrieve the 0.75-μm aerosol optical depth from the radiation, and an iteration inversion algorithm and a parameterized DSR model are developed. A key question of the method is the effect of aerosol size distribution uncertainty on the depth solution. As shown in inversion simulations using Junge size distributions and LOWTRAN7 aerosol models, a depth accuracy better than 5% can generally be expected for a solar zenith angle less than 75° if a Junge distribution with exponent ν = 3 is selected for the retrievals. In general, the smaller the depth and solar zenith angle, the higher the accuracy. In addition, it is important for improving solution accuracy to exactly determine the DSR and water vapor absorptance. If errors in DSR and the vertical water vapor amount are within ±2% and ±0.2 cm, resultant depth errors are usually within ±0.02 and ±0.01... Abstract Based on sensitivity of total direct solar radiation (DSR) detected by a pyrheliometer for aerosol physical–optical properties, a method is proposed to retrieve the 0.75-μm aerosol optical depth from the radiation, and an iteration inversion algorithm and a parameterized DSR model are developed. A key question of the method is the effect of aerosol size distribution uncertainty on the depth solution. As shown in inversion simulations using Junge size distributions and LOWTRAN7 aerosol models, a depth accuracy better than 5% can generally be expected for a solar zenith angle less than 75° if a Junge distribution with exponent ν = 3 is selected for the retrievals. In general, the smaller the depth and solar zenith angle, the higher the accuracy. In addition, it is important for improving solution accuracy to exactly determine the DSR and water vapor absorptance. If errors in DSR and the vertical water vapor amount are within ±2% and ±0.2 cm, resultant depth errors are usually within ±0.02 and ±0.01...