Estimating understory dwarf bamboo distribution in forests using digital canopy height models derived from drone LiDAR

Abstract

Understory vegetation plays a critical role in forest regeneration and biodiversity conservation. In particular, the proliferation of dwarf bamboo hinders forest regeneration and complicates management efforts. Therefore, it is essential to develop methods for efficiently and extensively identifying the presence of dwarf bamboo. This study estimated their occurrence using Digital Canopy Height Models (DCHM) created from drone-based LiDAR point cloud. Principal Component Analysis was applied to DCHM images, and a model was built using random forest, achieving high accuracy. The first principal component was related to average canopy height, while other components reflected spatial patterns of overstory trees. These results indicate that canopy structure patterns from DCHM can effectively estimate the distribution of understory dwarf bamboo in forests.