Spatial Patterns of Trees from Airborne LiDAR Using a Simple Tree Segmentation Algorithm

Abstract:

Objectives for management of forest ecosystems on public land incorporate a focus on maintenance and restoration of ecological functions through silvicultural manipulation of forest structure. The spatial pattern of residual trees – the horizontal element of structure – is a key component of ecological restoration prescriptions. We tested the ability of a simple LiDAR individual tree segmentation method – the watershed transform – to generate spatial pattern metrics similar to those obtained by the traditional method – ground-based stem mapping – on forested plots representing the structural diversity of a large wilderness area (Yosemite NP) and a large managed area (Sierra NF) in the Sierra Nevada, Calif. Most understory and intermediate-canopy trees were not detected by the LiDAR segmentation; however, LiDAR- and field-based assessments of spatial pattern in terms of tree clump size distributions largely agreed. This suggests that (1) even when individual tree segmentation is not effective for tree density estimates, it can provide a good measurement of tree spatial pattern, and (2) a simple segmentation method is adequate to measure spatial pattern of large areas with a diversity of structural characteristics. These results lay the groundwork for a LiDAR tool to assess clumping patterns across forest landscapes in support of restoration silviculture. This tool could describe spatial patterns of functionally intact reference ecosystems, measure departure from reference targets in treatment areas, and, with successive acquisitions, monitor treatment efficacy.