Estimating Stand Volume from Nonparametric Distribution of Airborne LiDAR Data

This study was performed to estimate stand-level volume using the characteristics of vertical and horizontal distribution of airborne Light Detection And Ranging (LiDAR) data. It is found that the height distributional parameters, such as percentile, of LiDAR data reflected onand in-canopy in a stand have the relationship with stand volume in previous research. However, we assumed that the nonparametric height distribution form of canopy LiDAR returns would be obviously related with the stand volume directly. Nonparametric height distribution was presented to be a continuous line according to the frequency of LiDAR returns by the height. Thereafter, the sum of each height of all canopy returns, which means the area below the continuous line, was compared to stand volume using National Forest Inventory (NFI) data. In addition, for verifying the volume of test stands, the similarity which is the overlapping ratio between the height distribution curves of sample and test stand was calculated. The relationship between the height sum and stand volume was relatively high to be R=0.83. Based on such relationship, the maximum similarity of each test stand was computed as compared sample stands. As a result, mean similarity and root mean square error (RMSE) of estimated stand volumes were 82% and 34.96m/ha respectively. However, supplementary indices, for non-overlapping part in similar distribution of canopy returns of sample and test stand, are needed to reduce such errors.