Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Klassning och statistisk separabilitetsanalys av marktäckningsklasser i Halland : analys av multivariata data Landsat TM och ERS-1 SAR

Liljeberg, Marcus (1997) In Lunds universitets Naturgeografiska institution - Seminarieuppsatser
Dept of Physical Geography and Ecosystem Science
Abstract
The objective of this study is to classify Swedish land cover classes using multisource satellite data. Multispectral Landsat TM data from April 1993, and multitemporal ERS-1 SAR data acquired seven times during the growing season of 1993 have been used. Nine agricultural and seven forest land cover classes have been analysed. The statistical separabilities between land cover classes have also been investigated to determine the best combinations of satellite data for discrimination between land cover classes. The Jeffries-Matusita distance is used as a separability measurement. When each data source is analysed separately , separabilities are greater with the SAR data. However, the highest separabilities are generally achieved when the two... (More)
The objective of this study is to classify Swedish land cover classes using multisource satellite data. Multispectral Landsat TM data from April 1993, and multitemporal ERS-1 SAR data acquired seven times during the growing season of 1993 have been used. Nine agricultural and seven forest land cover classes have been analysed. The statistical separabilities between land cover classes have also been investigated to determine the best combinations of satellite data for discrimination between land cover classes. The Jeffries-Matusita distance is used as a separability measurement. When each data source is analysed separately , separabilities are greater with the SAR data. However, the highest separabilities are generally achieved when the two data types are combined. A synergy can therefore be confirmed in this multisource approach, based on statistical separability results. A maximum likelihood algorithm is used for the c1assification. When classifying various combinations of TM and SAR data the highest overall c1assification accuracy using all sixteen land cover classes is achieved when TM and SAR data are combined (58%), compared to c1assification accuracy achieved when using TM-data alone (43%) or SAR-data alone (45%) respectively. (Less)
Please use this url to cite or link to this publication:
author
Liljeberg, Marcus
supervisor
organization
year
type
H1 - Master's Degree (One Year)
subject
keywords
geografi, naturgeografi, separabilitetsanalys, marktäckningsklasser, Halland
publication/series
Lunds universitets Naturgeografiska institution - Seminarieuppsatser
report number
43
language
Swedish
id
2061235
date added to LUP
2011-11-22 12:44:58
date last changed
2011-11-22 12:44:58
@misc{2061235,
  abstract     = {{The objective of this study is to classify Swedish land cover classes using multisource satellite data. Multispectral Landsat TM data from April 1993, and multitemporal ERS-1 SAR data acquired seven times during the growing season of 1993 have been used. Nine agricultural and seven forest land cover classes have been analysed. The statistical separabilities between land cover classes have also been investigated to determine the best combinations of satellite data for discrimination between land cover classes. The Jeffries-Matusita distance is used as a separability measurement. When each data source is analysed separately , separabilities are greater with the SAR data. However, the highest separabilities are generally achieved when the two data types are combined. A synergy can therefore be confirmed in this multisource approach, based on statistical separability results. A maximum likelihood algorithm is used for the c1assification. When classifying various combinations of TM and SAR data the highest overall c1assification accuracy using all sixteen land cover classes is achieved when TM and SAR data are combined (58%), compared to c1assification accuracy achieved when using TM-data alone (43%) or SAR-data alone (45%) respectively.}},
  author       = {{Liljeberg, Marcus}},
  language     = {{swe}},
  note         = {{Student Paper}},
  series       = {{Lunds universitets Naturgeografiska institution - Seminarieuppsatser}},
  title        = {{Klassning och statistisk separabilitetsanalys av marktäckningsklasser i Halland : analys av multivariata data Landsat TM och ERS-1 SAR}},
  year         = {{1997}},
}