Aerosol microphysical properties, such as volume concentration (VC) and effective radius (ER), are of great importance to evaluate their radiative forcing and impacts on climate change. However, range-resolved aerosol VC and ER still cannot be obtained by remote sensing currently except for the column-integrated one from sun-photometer observation. In this study, a retrieval method of range-resolved aerosol VC and ER is firstly proposed based on the partial least squares regression (PLSR) and deep neural networks (DNN), combining polarization lidar and collocated AERONET (AErosol RObotic NETwork) sun-photometer observations. The results show that the measurement of widely-used polarization lidar can be reasonably used to derive the aerosol VC and ER, with the determination coefficient (R2) of 0.89 (0.77) for VC (ER) by use of the DNN method. Moreover, it is proven that the lidar-based height-resolved VC and ER at near-surface are well consistent with independent observations of collocated Aerodynamic Particle Sizer (APS). Additionally, we found that there are significant diurnal and seasonal variations of aerosol VC and ER in the atmosphere at Semi-Arid Climate and Environment Observatory of Lanzhou University (SACOL). Compared with columnar ones from the sun-photometer observation, this study provides a reliable and practical way to obtain full-day range-resolved aerosol VC and ER from widely-used polarization lidar observation, even under cloud conditions. Moreover, this study also can be applied to long-term observations by current ground-based lidar networks and spaceborne CALIPSO lidar, aiming to further evaluate aerosol climatic effects more accurately.