Abstract for presentation at The 13th Australasian Remote Sensing and Photogrammetry Conference

Modelling of Foliage Projected Cover Using Landsat-7 Spectral Imagery: Spectral Indices for Greeness and Brightness

  • Mr Trevor Moffiet, Newcastle University, Faculty of Science & IT, Australia
  • K Mengersen, Queensland University of Technology, Faculty of Science, Australia
  • Robert King, University of Newcastle, Faculty of Science & IT, Australia
  • Mr John Armston, Queensland Department of Natural Resources, Mines and Water, Australia
  • Christian Witte, Queensland Department of Natural Resources, Mines and Water, Australia

    • Field and LiDAR data from the Injune Collaborative Landscape Project are being used to develop a new statistical modelling technique for relating Landsat spectral imagery to Foliage Projected Cover (FPC). The aim is to build a single Bayesian model for the prediction of FPC from Landsat spectral imagery which will incorporate a number of different surrogate estimates of FPC together with their uncertainties. Surrogate estimates of FPC include field transect estimates, tree basal area estimates and LiDAR estimates. Uncertainty includes sampling error as well as intrinsic variation in the relationships between the different types of estimates. The new approach requires the development of a new spectral vegetation (greenness) index to assist with the modelling of FPC. In this paper we present a method for creating indices for greenness and brightness by a polar transformation of the first two principal components of the spectral imagery. The new indices are compared to the normalised difference vegetation index and the tasselled cap indices. Elements of the larger study are presented here only as a background to the development of these new indices.
    The study supports the Queensland government's Statewide Landcover and Trees Study (SLATS) and Australian Greenhouse Office initiatives of landcover change analysis. An anticipated outcome of the larger study is a method for estimation of FPC by Landsat, which will provide a measure of uncertainty of predictions. Outcomes of the study should initially assist with validation of current models for regional mapping of FPC and cover change. The large-scale systematic LiDAR sampling of the Injune study area coincident with Landsat-7 ETM+ spectral imagery from the same season in year 2000, has been fundamental to the development of these new methods.

    Conference Organiser - ICMS Pty Ltd