Accurately identifying sunlit and shaded leaves using process-based ecological models can improve the simulation accuracy of forest photosynthetic rates and potential carbon sequestration capacity. However, it is still challenging to characterize their three dimensional (3-D) spatiotemporal distributions due to the complex structure. In this study, we developed a light detection and ranging (lidar)-based approach to map the spatiotemporal distribution patterns of photosynthetically active radiation (PAR) and sunlit and shaded leaves within forest canopies. By using both terrestrial laser scanning (TLS) and unmanned aerial vehicle-based lidar system (UAV-LS), we analyzed the influences of different scanning geometries and associated point densities on the separation of sunlit and shaded leaves. Moreover, we further investigated the effects of woody materials and penumbra sizes on identifying sunlit and shaded leaves by separating the foliage and woody materials and estimating the penumbras of sunlit leaves. Our results showed that: (1) The proposed lidar-based PAR model could well capture the variations of field-based pyranometer measurements using fused point data by combining UAV-LS and TLS data (mean R-square = 0.88, mean root mean square error (RMSE) = 155.5 μmol·m−2·s−1, p < 0.01). The separate UAV-LS and TLS-based fractions of sunlit leaves were averagely overestimated by 34.3% and 21.6% when compared to the fused point data due to their different coverages and comprehensiveness. (2) The woody materials showed different effects on sunlit leaf fraction estimations for forest overstory and understory due to the variations of solar zenith angle and tree spatial distribution patterns. The most noticeable differences (i.e., −36.4%) between the sunlit leaf fraction before and after removing woody materials were observed around noon, with a small solar zenith angle and low-density forest stand. (3) The penumbra effects were seen to increase the sunlit leaf fraction in the lower canopy by introducing direct solar radiation, and it should be considered when using 3-D structural information from lidar to identify sunlit and shaded leaves.