Overcoming seasonal hurdles: estimating leaf-on Digital Surface Model from late fall Lidar data for enhanced forest tree height information

Bilguunmaa Myagmardulam*, Kazuyoshi Takahashi

Abstract

A canopy height model (CHM) is crucial for effective forest resource assessment, which requires accurate digital surface models (DSMs) during leaf-on periods. This study addresses the challenge of constructing comprehensive CHMs in regions such as Japan, where LiDAR data is predominantly collected in leaf-off seasons. We introduce a novel approach to estimate leaf-on DSMs using late fall LiDAR data, overcoming the limitations of the leaf-off conditions. Our method applies a spatial filtering operation to identify leaf-off grids based on statistical thresholds, followed by a Savitzky-Golay smoothing filter and a local maximum operation for DSM estimation. This innovative technique significantly improved the mean absolute difference (MAD) in DSM estimates, reducing it from 3.2 meters in leaf-off conditions to 1.19 meters for estimated leaf-on conditions. These results demonstrate our approach's potential for year-round forest monitoring and accurate resource assessment, despite the seasonal constraints in LiDAR data collection.

Keywords: Digital Surface Model; Canopy Height Model; DJI Zenmuse l1; LiDAR

Full Text: