Abstract:

Based on the airborne laser scanning (ALS) data and field measurement data of 191 sample plots distributed across the Korean pine (Pinus koraiensis) forest stands in northeastern China,the base regression models for estimating forest volume,mean height and basal area were determined through multiple linear regression and nonlinear regression methods at first.Then,a model system including stand volume,mean height,and basal area based on the ALS data was developed,using the error-in-variable simultaneous equations approach.The results show that for the developed multiple linear,nonlinear and two-variable nonlinear stand volume regression models,the coefficients of determination (R²) are 0.858,0.846 and 0.821;mean prediction errors (MPEs) are 2.57%,2.66% and 2.85%;and mean percent standard errors (MPSEs) are 26.35%,16.35% and 17.88%,respectively.For the mean height,basal area and stand volume simultaneous models,the R ² are 0.597,0.750 and 0.822;the MPEs are 1.90%,2.52% and 2.84%;and the MPSEs are 10.85%,15.28% and 17.73%,respectively.For estimating the forest parameters such as stand volume based on ALS data,nonlinear model is better than linear model,and the two-variable nonlinear model based on height (the median) and intensity (the percentile 75) of point clouds performed well.The error-in-variable simultaneous equations approach is a feasible method for developing a model system for estimation of main forest parameters.The developed mean height,basal area and stand volume simultaneous models in this study meet the need of precision requirements to relevant regulations on forest resources inventory,indicating that the models can be applied in practice.

Key words: airborne laser scanning, forest volume, nonlinear model, error-in-variable, simultaneous equations, Pinus koraiensis