Species identification of individual trees by combining high resolution LiDAR data with multi-spectral images

  • Authors:

  • J. Holmgren;Å. Persson;U. Söderman

  • Affiliations:

  • Swedish University of Agricultural Sciences, Department of Forest Resource Management, SE-90183 Umeå, Sweden;Global IP Solutions, Magnus Ladulåsgatan 63B, SE-11827 Stockholm, Sweden;FORAN Remote Sensing AB, SE-58330 Linköping, Sweden

  • Venue:
  • International Journal of Remote Sensing - 3D Remote Sensing in Forestry
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The objectives of this study were to identify useful predictive factors for tree species identification of individual trees and to compare classifications based on a combination of LiDAR data and multi-spectral images with classification by the use of each individual data source. Crown segments derived from LiDAR data were mapped to multi-spectral images for extraction of spectral data within individual tree crowns. Several features, related to height distribution of laser returns in the canopy, canopy shape, proportion of different types of laser returns, and intensity of laser returns, were derived from LiDAR data. Data from a test site in southern Sweden were used (lat. 58°30' N, long. 13°40' E). The forest consisted of Norway spruce (Picea abies), Scots pine (Pinus sylvestris), and deciduous trees. Classification into these three tree species groups was validated for 1711 trees that had been detected in LiDAR data within 14 field plots (sizes of 20×50 m2 or 80×80 m2). The LiDAR data were acquired by the TopEye MkII system (50 LiDAR measurements per m2) and the multi-spectral images were taken by the Zeiss/Intergraph Digital Mapping Camera. The overall classification accuracy was 96% when both data sources were combined.