Abstract: Accurate estimation of plantation biomass is of great significance for the scientific management and opera-tion of forests and for supporting China's “Dual Carbon” goals. Traditional ground survey methods have bottlenecks such as low efficiency and limited coverage. Unmanned aerial vehicle LiDAR (UAV-LiDAR) technology provides a new approach for forest aboveground biomass estimation through high-precision 3D point data. We selected 112 permanent plots of Larix olgensis plantations in Mengjiagang Forest Farm and classified them into a density gradient of nine levels from 200 to 0.5 points·m^-2. We established a forest aboveground biomass estimation model by generating subsets through repeated random pulse sampling and extracting canopy mean height (HMEAN) and canopy height ratio (CHR), and analyzed the effects of point density on indicator stability, model parameters, and prediction accuracy. The results showed that when point density decreased from 200 points·m^-2 to 0.5 points·m^-2, the mean values of HMEAN and CHR remained highly stable, but the random errors from repeated sampling increased. The mean standard deviation of HMEAN increased from 0.012 m to 0.261 m, and that of CHR increased from 0.0008 to 0.0167. The density reduction caused structural shifts in the aboveground biomass model, along with gradually expanding uncertainty in model parameters. The predictive accuracy metrics, root mean square error (RMSE), and Bias progressively increased, while the standard deviations of repeated sampling for both RMSE and Bias also showed a gradually expanding trend. The predicted mean aboveground biomass for the 112 plots remained stable, but the standard deviation of repeated predictions increased from 0.26 Mg·hm^-2 to 5.26 Mg·hm^-2. Reducing point density significantly decreased the accuracy and stability of aboveground biomass estimation. Maintaining point density above 20 points·m^-2 could control the RMSE within 16%. This study provided a valuable reference for UAV-LiDAR data acquisition and aboveground biomass estimation in L. olgensis planation.

Key words: UAV-LiDAR, point density, aboveground biomass (AGB), Larix olgensis