UAV Imagery for Measuring Large Wood Volume and Distribution: Ground Truthing Results and Sources of Error

Abstract

abstract: UAV imagery offers great potential in determining precise environmental measurements. This study identifies and quantifies errors and their causes/sources between UAV imagery- and field-derived measurements of large wood (LW) and cross-sectional valley topography in several forested headwater stream valleys in the Ozark Highlands, Missouri. UAV imagery was collected at six sites following best practices for producing high resolution orthophotos and elevation models through Structure-from-Motion photogrammetry. Field-based measurements of LW were completed along 5 m wide valley cross-sections in which all LW > 1 m in length and 0.1 m in diameter were recorded. Topographic surveys were performed along the centerlines of the LW transects. The same assessments were performed using UAV imagery and ArcGIS. Our sites showed high (>= 0.85) R2 values between field and UAV-GIS topographic surveys. Differences between measurements of LW dimensions tended to be larger with R2 ranging from 0.42 to 0.88. Overall 78 percent of standing trees and 60 percent of LW pieces were identified in the UAV imagery. The main source of detection and measurement error was due to vegetation/canopy cover. This study furthers our understanding of the applications and limitations of UAV derived data for environmental assessments.