• DocumentCode
    484333
  • Title

    A Multi-Scale Approch for Retreiving Proportional Cover of Life Forms

  • Author

    Gessner, Ursula ; Conrad, Christopher ; Huttich, Christian ; Keil, Manfred ; Schmidt, Michael ; Schramm, Matthias ; Dech, Stefan

  • Author_Institution
    Remote Sensing Unit of DLR, Univ. of Wuerzburg, Wurzburg
  • Volume
    3
  • fYear
    2008
  • fDate
    7-11 July 2008
  • Abstract
    This study presents a multi-scale procedure to derive continuous proportional cover of woody vegetation in savanna ecosystems. QuickBird data was classified to define a continuous training and validation data set of woody cover proportions. Using a regression tree algorithm based on Landsat TM data, this woody cover information was extrapolated to an area of approximately 185 km times 185 km. The resulting 30 m map of the Namibian North-eastern Kalahari Woodland was aggregated to 250 m and 500 m resolutions. Comparisons of the global MODIS VCF product with the regionally adjusted multi-scale fractional cover map indicate that VCF tree cover is generally underestimated in the study area and confusions between tree and dense shrub cover occur.
  • Keywords
    image classification; vegetation mapping; Kalahari Woodland; Landsat TM data; QuickBird data classification; Vegetation Continuous Fields; global MODIS comparison; life forms distribution; northeastern Namibia; regression tree algorithm; savanna ecosystems; shrub cover; woody vegetation proportional cover; Ecosystems; Fires; Humans; Land surface; Regression tree analysis; Remote sensing; Sampling methods; Satellites; Soil; Vegetation mapping; VCF; fractional cover; multi-scale analysis; regression tree; savanna;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Geoscience and Remote Sensing Symposium, 2008. IGARSS 2008. IEEE International
  • Conference_Location
    Boston, MA
  • Print_ISBN
    978-1-4244-2807-6
  • Electronic_ISBN
    978-1-4244-2808-3
  • Type

    conf

  • DOI
    10.1109/IGARSS.2008.4779444
  • Filename
    4779444