Abstract: [Objective] In order to provide advice for the research of forest structural parameters estimation with airborne LiDAR data, the effect of side-overlap between adjacent swaths on forest stand mean height and leaf area index estimation was studied in this paper.[Method] In this study, the forest stand mean height and leaf area index from 30 field plots of scotch pine and 33 field plots of larch pine, were measured. Firstly, the raw LiDAR data were processed through a series of procedures including noise points removal, point cloud classification, elevation normalization and side-overlap points cut off. Then the quantile heights of LiDAR points (H_P1, H_P5, H_P10, …, H_P99, H_max and H_mean) and laser penetration index (LPI), extracted from the LiDAR data with overlap points, overlap points and LiDAR data without overlap points, were used to establish and assess the model accuracy of forest stand mean height and LAI estimation by using the leave one out cross validation. The coefficient of determination (R²) and root mean square error (RMSE) were used to assess the effect of side-overlap between LiDAR adjacent swaths on the estimation of forest stand mean height and LAI.[Result] For forest stand mean height estimation, the best result of scotch pine (R²=0.873, RMSE=0.940) from LiDAR data with overlap points was obtained with H_P90. The best result (R²=0.892, RMSE=0.866) from overlap points was achieved with H_P80. And the best result (R²=0.892, RMSE=0.868) from LiDAR data without overlap points was obtained at H_P55. For larch pine, all the best result were achieved with H_P99 for LiDAR data with overlap points, overlap points and LiDAR data without overlap points. The R² were 0.725, 0.719 and 0.741, and RMSE were 1.196, 1.209 and 1.161 respectively. For forest LAI estimation, the R² and RMSE of scotch pine were 0.666 and 0.220 when the overlap points included in LiDAR data. The R² and RMSE from overlap points were 0.551 and 0.255. In addition, the R² was improved to 0.794 and RMSE was induced to 0.172 after the overlap points were removed from LiDAR data. For larch pine, the R² and RMSE were 0.654 and 0.110 when the overlap points were included in LiDAR data. The R² and RMSE were 0.640 and 0.112 when the overlap points were used. While, the R² improved to 0.762 and RMSE was 0.091 when the overlap points were removed from LiDAR data.[Conclusion] Both for forest stand mean height and LAI estimation, the result acquired from LiDAR data without overlap points were better than the result obtained from LiDAR data with overlap points. Additionally, the result of scotch pine were better than the result of larch pine. For forest stand mean height estimation, the best result from scotch pine and larch pine achieved with different quantile height parameters. The removal of overlap points between LiDAR adjacent swaths could effectively improve the estimation accuracy of forest structural parameters. Therefore, the overlap cut off should be included in the pre-processing of airborne LiDAR point cloud data to improve the forest structural parameters estimation accuracy in the future research.

Key words: airborne LiDAR, side-overlap between adjacent swaths, coniferous forest, mean tree height, leaf area index(LAI)