Comparison of ERS SAR and Landsat TM forest classification

Terhikki Manninen, Tuomas Häme, Anssi Lohi

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

Abstract

The seasonal variation of intensities in SAR images is typically higher on agricultural fields than in forests. Therefore multitemporal use of SAR data enables the distinction of fields and forests [1, 2]. Single SAR images are not as easy to use for forest mapping, although at certain times of the season the intensities of the fields and forests are clearly different. Due to the speckle direct classification is usually not succesful. Moreover, the modest resolution of ERS SAR makes reliable forest detection even more difficult [2]. Recently a method for distinguishing boreal forests and open land areas using single ERS images has been developed [3]. For boreal forests an unsupervised classification method utilizing Landsat TM data has been used in several contexts lately [4, 5]. The method is based on clusters labeled using their reflectance values. The ERS SAR single image forest/nonforest classification results are compared in this paper with the Landsat TM classification results in two test sites in Southern Finland.
Original languageEnglish
Title of host publicationSecond International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications
Place of PublicationThe Netherlands
PublisherEuropean Space Agency ESA
Pages347-354
VolumeSP-441
ISBN (Print)92-9092-764-X
Publication statusPublished - 1998
MoE publication typeNot Eligible
Event2nd International Workshop on Retrieval of Bio and Geo-physical Parameters from SAR Data for Land Applications - Noordwijk, Netherlands
Duration: 21 Oct 199823 Oct 1998
Conference number: 2

Conference

Conference2nd International Workshop on Retrieval of Bio and Geo-physical Parameters from SAR Data for Land Applications
CountryNetherlands
CityNoordwijk
Period21/10/9823/10/98

Fingerprint

Landsat thematic mapper
synthetic aperture radar
boreal forest
unsupervised classification
speckle
ERS
comparison
reflectance
seasonal variation
method

Cite this

Manninen, T., Häme, T., & Lohi, A. (1998). Comparison of ERS SAR and Landsat TM forest classification. In Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications (Vol. SP-441, pp. 347-354). The Netherlands: European Space Agency ESA.
Manninen, Terhikki ; Häme, Tuomas ; Lohi, Anssi. / Comparison of ERS SAR and Landsat TM forest classification. Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications. Vol. SP-441 The Netherlands : European Space Agency ESA, 1998. pp. 347-354
@inproceedings{4138f9b0236a4e729f13286fb8ba4279,
title = "Comparison of ERS SAR and Landsat TM forest classification",
abstract = "The seasonal variation of intensities in SAR images is typically higher on agricultural fields than in forests. Therefore multitemporal use of SAR data enables the distinction of fields and forests [1, 2]. Single SAR images are not as easy to use for forest mapping, although at certain times of the season the intensities of the fields and forests are clearly different. Due to the speckle direct classification is usually not succesful. Moreover, the modest resolution of ERS SAR makes reliable forest detection even more difficult [2]. Recently a method for distinguishing boreal forests and open land areas using single ERS images has been developed [3]. For boreal forests an unsupervised classification method utilizing Landsat TM data has been used in several contexts lately [4, 5]. The method is based on clusters labeled using their reflectance values. The ERS SAR single image forest/nonforest classification results are compared in this paper with the Landsat TM classification results in two test sites in Southern Finland.",
author = "Terhikki Manninen and Tuomas H{\"a}me and Anssi Lohi",
year = "1998",
language = "English",
isbn = "92-9092-764-X",
volume = "SP-441",
pages = "347--354",
booktitle = "Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications",
publisher = "European Space Agency ESA",
address = "France",

}

Manninen, T, Häme, T & Lohi, A 1998, Comparison of ERS SAR and Landsat TM forest classification. in Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications. vol. SP-441, European Space Agency ESA, The Netherlands, pp. 347-354, 2nd International Workshop on Retrieval of Bio and Geo-physical Parameters from SAR Data for Land Applications, Noordwijk, Netherlands, 21/10/98.

Comparison of ERS SAR and Landsat TM forest classification. / Manninen, Terhikki; Häme, Tuomas; Lohi, Anssi.

Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications. Vol. SP-441 The Netherlands : European Space Agency ESA, 1998. p. 347-354.

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

TY - GEN

T1 - Comparison of ERS SAR and Landsat TM forest classification

AU - Manninen, Terhikki

AU - Häme, Tuomas

AU - Lohi, Anssi

PY - 1998

Y1 - 1998

N2 - The seasonal variation of intensities in SAR images is typically higher on agricultural fields than in forests. Therefore multitemporal use of SAR data enables the distinction of fields and forests [1, 2]. Single SAR images are not as easy to use for forest mapping, although at certain times of the season the intensities of the fields and forests are clearly different. Due to the speckle direct classification is usually not succesful. Moreover, the modest resolution of ERS SAR makes reliable forest detection even more difficult [2]. Recently a method for distinguishing boreal forests and open land areas using single ERS images has been developed [3]. For boreal forests an unsupervised classification method utilizing Landsat TM data has been used in several contexts lately [4, 5]. The method is based on clusters labeled using their reflectance values. The ERS SAR single image forest/nonforest classification results are compared in this paper with the Landsat TM classification results in two test sites in Southern Finland.

AB - The seasonal variation of intensities in SAR images is typically higher on agricultural fields than in forests. Therefore multitemporal use of SAR data enables the distinction of fields and forests [1, 2]. Single SAR images are not as easy to use for forest mapping, although at certain times of the season the intensities of the fields and forests are clearly different. Due to the speckle direct classification is usually not succesful. Moreover, the modest resolution of ERS SAR makes reliable forest detection even more difficult [2]. Recently a method for distinguishing boreal forests and open land areas using single ERS images has been developed [3]. For boreal forests an unsupervised classification method utilizing Landsat TM data has been used in several contexts lately [4, 5]. The method is based on clusters labeled using their reflectance values. The ERS SAR single image forest/nonforest classification results are compared in this paper with the Landsat TM classification results in two test sites in Southern Finland.

M3 - Conference article in proceedings

SN - 92-9092-764-X

VL - SP-441

SP - 347

EP - 354

BT - Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications

PB - European Space Agency ESA

CY - The Netherlands

ER -

Manninen T, Häme T, Lohi A. Comparison of ERS SAR and Landsat TM forest classification. In Second International Workshop on Retrieval of Bio- & Geo-physical Parameters from SAR Data for Land Applications. Vol. SP-441. The Netherlands: European Space Agency ESA. 1998. p. 347-354