The Effect of Sampling Design on Data Precision and Practicality in Mobile-LiDAR-based Forest Inventories: Case Study from Rize Senyuva

Abstract

This study aims to determine an optimal sampling design for forest inventories performed by mobile laser scanners (LiDAR). To this end, the study area, located in Şenyuva Forest Planning Unit, was first surveyed using the conventional ground measurement method by sample plots. Then, it was scanned by mobile LiDAR with different walking routes. Produced point clouds were clipped with different shapes and sizes for feature extraction. Finally, the two datasets were compared for the same stand parameters. Regarding diameter at breast height (DBH), no significant difference was found between the LiDAR data and ground truth (p>0.05) for the entire area (1,834.4 m2). The difference between the datasets was less than 1 cm (~2%) based on the mean of the two data. The number of trees parameter was completely the same and the deviation in dominant tree height was less than 1 m (~1.5%). In conclusion, specific prescriptions were written out for practitioners, surveying different forest conditions. Accordingly, clipping the LiDAR data by 400-m2 circles by scanning the plot from outside is the best option for practitioners who can accept an estimation error of about 10%. The practitioners, who need more precise estimates, should analyze the same data on the entire plot without clipping. If the plot size is greater than 1 ha, scanning within the stand would be necessary.