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

Deriving Terrain Roughness Information through Texture and Spectral Analysis of Merged SAR and Optical Datasets

  • Dr Alan Forghani, Australian Centre for Remote Sensing (ACRES), Geoscience Australia, Australia
  • Mr Medhavy Thankappan, Australian Centre for Remote Sensing (ACRES), Geoscience Australia, Australia
  • Mr Craig Smith, Australian Centre for Remote Sensing (ACRES), Geoscience Australia, Australia
  • Dr Bob Cechet, Risk Research Group, Geospatial & Earth Monitoring Division, Geoscience Australia, Australia
  • Dr Krishna Nadimpalli, Risk Research Group, Geospatial & Earth Monitoring Division, Geoscience Australia, Australia

    • The Wind Risk Assessment Project at Geoscience Australia used Landsat data to derive terrain surface roughness information for four major metropolitan areas across Australia in 2005. It was necessary to investigate the applicability of multi-sensor approaches in order to produce a regional/national terrain surface roughness map based on the Australian/New Zealand wind loading standard (AS/NZS 1170.2, 2002). Object-based classification of various optical datasets yielded accuracies of 79% and 93% over metropolitan (Sydney) and rural/urban areas respectively. To further improve the quality of the terrain surface roughness product, the potential of SAR-based texture information was explored.
    This paper reports on the development of a methodology based on an integrated texture and spectral analysis of merged SAR and Landsat data. A RADARSAT-1 Fine Beam image acquired on 12 December 2001 over Wollongong, NSW, was processed to derive a roughness map using texture-based occurrence and co-occurrence measures. Haralick texture measures derived from SAR data was then merged with multispectral information from Landsat imagery, acquired on 5 October 2002. Ten categories of terrain surface roughness were derived by classifying the merged SAR and optical datasets. Visual assessment of the texture and spectral-based product revealed additional topographic information (eg slopes, ridges and valleys) critical for the Wind Risk Assessment, compared to results based solely on the spectral-based product from Landsat imagery.

    Conference Organiser - ICMS Pty Ltd