Exploring the Utility of Hyperspectral Imagery and LiDAR Data for Predicting Quercus garryana Ecosystem Distribution and Aiding in Habitat Restoration

Authors

  • Trevor G. Jones,

    Corresponding author
    1. Department of Forest Resources Management, 2424 Main Mall, University of British Columbia, Vancouver V6T 1Z4, Canada
      T. G. Jones, email tgjones@interchange.ubc.ca
    Search for more papers by this author
  • Nicholas C. Coops,

    1. Department of Forest Resources Management, 2424 Main Mall, University of British Columbia, Vancouver V6T 1Z4, Canada
    Search for more papers by this author
  • Tara Sharma

    1. Parks Canada, Gulf Islands National Park Reserve of Canada, 2220 Harbour Road, Sidney V8L 2P6, Canada
    Search for more papers by this author

T. G. Jones, email tgjones@interchange.ubc.ca

Abstract

In west-coastal Canada Garry oak habitat has been significantly degraded and reduced to 1–5% of its pre-European settlement range. To reverse the at-risk status of species associated with Garry oak habitat, restoration efforts are mandatory. Effective restoration requires understanding habitat distribution in a detailed, accurate, and spatially explicit manner. This research investigates whether classified airborne hyperspectral imagery can provide distribution predictions that are more detailed and accurate than those stemming from conventional aerial photograph interpretation. Furthermore, this research assesses whether including structural information represented by light detection and ranging (LiDAR) data increases classification accuracies. Hyperspectral classification resulted in an overall accuracy of 86.4% with a 0.8 Kappa Index of Agreement (KIA) and Garry oak producer's and user's accuracies of 81.7 and 92.1%, respectively. Including structural information as classification input resulted in an overall accuracy of 87.2% with a 0.8 KIA and Garry oak producer's and user's accuracies of 86.9 and 81.5%, respectively. Both rounds of classification identified the precise location and amount of Garry oak trees/tree clusters at a 2-m spatial resolution providing significant improvement as compared with 1:5,000 scale polygons mapped with a minimum unit of 0.04 ha (i.e. 20 × 20-m) which comprise conventional data. Despite lower user's accuracy for Garry oak, overall, Garry oak producer's and most per-class accuracies (producer's and user's) increased with the inclusion of structural information and therefore its use is recommended. Classification results provide contemporary reference information which can inform required restoration activities and be used to judge their effectiveness.

Ancillary