Simulating Sampling Efficiency in Airborne Laser Scanning Based Forest Inventory

A simple simulator was developed to test whether airborne laser scanning can be used as a strip sampling tool for forest inventory purposes. The simulator is based on the existing two stages, grid based laser inventory procedure. A population of trees was created using an existing forest stand structure generator. Each tree was represented by means of its 3D-crown model derived from airborne laser scanning measurements and field measured parameters, i.e. total tree height, height of crown base and average crown diameter. Monte Carlo simulations were run to assess the efficiency of volume estimates obtained from airborne laser scanning and ground based inventory. The lowest RMSE for the laser based estimates was 5.1 mha (2.0%) and the highest was 8.4 mha (3.3%), while RMSE for the ground based estimates varied between 13.7 mha (5.4%) and 18.4 mha (7.2%). The LIDAR based estimation was on average 6.3 times more efficient in terms of MSE than ground based sampling. The RMSE of the volume estimates increased with increasing plot size, for a given sampling intensity. The results indicated that forest surveys over large areas carried out using airborne laser scanning as a strip sampling tool can provide accurate estimates, and can be more effective than traditional systematic ground plot based inventories.