The study of canopy structures is essential for the precise management of local forage. Although the relationship between the canopy height (CH) and light interception (LI) allows the best use of forage, there are other characteristics of the canopy that, if not observed, can lead to errors in grazing management. This study analyzes the effect of the defoliation frequencies (DF) and season on the productive and structural characteristics of Guinea grass and defines whether tiller population density (TPD) should be considered in the mathematical models that relate CH to LI. The experimental design used includes randomized blocks with six defoliation frequencies (DF) of 14, 21, 28, 35, 42, and 49 days, over three seasons: rainy season 2015 (RS2015), dry season (DS), and rainy season 2016 (RS2016), in the Amazon biome. Yield characteristics and forage structure were evaluated. The interaction between the DFs and seasons in all evaluated variables included forage accumulation, proportions of stem, leaves and dead material, tiller population density (TPD), CH, LI, and leaf area index. The curve steepness constant (CSC) of the nonlinear models between CH and LI was different from that of the DFs. A DF of 14–28 days averaged a TPD of 616 tillers m−2 and CSC of 0.0407, and a DF of 35–49 days averaged a TPD of 496 tillers m−2 and CSC of 0.0368. An agreement index (d) of 0.87, a bias of 1.5 %, and a determination coefficient (R²) of 0.6187 were obtained for the model that estimated a CH with the inclusion of LI and TPD when validated with data from studies published in the literature. TPD associated with CH and LI allowed for more precise management of the ideal defoliation moment.