Abstract:

Objective: This study aimed to develop a generalized forest volume model with same variables and stable structure based on airborne laser scanning(ALS) data, which would provide a reference for standardizing forest volume modeling and evaluation. Method: Based on the ALS data and field measurement data of 790 sample plots distributed across the larch(Larix spp.), Korean pine(Pinus koraiensis), poplar(Populus spp.) and birch(Betula spp.) forest stands in northeastern China, the stand volume regression models were developed through multiple linear regression and nonlinear regression methods, and the generalized model with same variables and unified structure was determined by comparison and analysis. Then, the stand volume models with the same ALS variables were developed jointly for different forest types, using the dummy variable modeling approach. Result: The developed multiple linear volume regression models for the 4 stand types have 2-6 explainable variables and the coefficients of determination(R²) are 0.701-0.827; the nonlinear models have 2-4 explainable variables and the R² are 0.707-0.818. The R² of two-variable nonlinear volume models based on mean height and mean intensity of point clouds are 0.679, 0.814, 0.698 and 0.703 for larch, Korean pine, poplar and birth forest stands, respectively; the mean prediction errors(MPEs) are 4.26%, 2.90%, 3.68% and 3.83%, and the mean percent standard errors(MPSEs) are 24.44%, 18.23%, 21.47% and 23.26%, respectively. Conclusion: For estimating stand volume based on ALS data, the nonlinear model might be better than the linear model, and the two-variable model based on mean height and mean intensity of point clouds might be generally applicable, which could be defined as standard model for estimating stand volume. The stand volume models developed in this study for 4 forest types using dummy variable modeling approach could meet the need of precision requirements to relevant regulations on forest resource inventory, indicating that the models could be applied in practice.

Key words: airborne laser scanning, forest volume, linear model, nonlinear model, dummy variable model