Abstract: Taking the Pinus yunnanensis in Qujing City, Yunnan Province as the research object, based on forest management inventory (FMI) data in 2016 and the simultaneous Landsat 8 OLI remote sensing images, subcompartment points were created randomly. Then, datasets were built through extracting subcompartment's ststistics values based on remote sensing indexes. Quantitative study of the light saturation value of P. yunnanensis biomass estimation based on different curve fitting. A linear stepwise regression model was established, and based on this model, the mixed effects models considering region and age group random effect were established to reduce the estimation error caused by light saturation in biomass remote sensing estimation. The results showed that the biomass saturation value of P. yunnanensis forests was 167 t/ha in the study area. The fitting accuracy of the mixed effects models are better than the basic linear stepwise regression model, and the prediction accuracy of the mixed-effects model with both random effects in the independent sample test (p=91.556%) is higher than that of the basic stepwise regression model (p=83.826%). The results of biomass segmentation residual test and the biomass inversion results of the study area showed that the mixed-effects model has a larger estimation range than the general stepwise regression model, and the mixed effects models can reduce the uncertainties from the high-value underestimation and low-value overestimation to a certain extent, and decrease the impact of data saturation on optical remote sensing image data.