A linear regression method for tree canopy height estimation using airborne lidar data

Abstract

Tree canopy height is one of the most fundamental measurements in forest inventory and is a critical variable in the quantitative assessment of tree (or stand) volume, forest biomass, carbon stocks, growth, and site productivity. In this study, we analyzed two traditional methods for tree canopy height estimation and designed a new linear regression method for improved tree canopy height estimation using airborne light detection and ranging (lidar) data. Examples of two typical crown shapes were used, and theoretical analysis was performed on simulated datasets with varying crown shape, unit penetrability, and laser-missed canopy layer(s). The final result derived from the simulated lidar data illustrates that the linear regression method can improve canopy height estimation. This method was also applied to lidar data covering a tall pine forest in Idaho, USA. An average error of 0.51 m was obtained from a comparison of the lidar-derived tree canopy heights and 79 field measurements. This error was also compared with the estimation error resulting from the use of two traditional methods. Results indicate our method produced more accurate tree canopy height estimates, with a mean error and root mean square error (RMSE) ranging between 25% and 50% lower than those from the two traditional methods.