Test of different classification methodologies for land cover mapping over France using SPOT/VEGETATION data: applications to the years 2002 and 2003

Author

Han, Kyung-Soo ; Tanguy, Yannick ; Champeaux, Jean-Louis ; Hagolle, Oliver

Author_Institution

CNRM/GMME/MATIS, METEO-FRANCE, Toulouse, France

Volume

4

fYear

2004

fDate

20-24 Sept. 2004

Firstpage

2713

Abstract

The present study aims at testing several methodologies of land cover mapping over France at 1 km resolution based on the remotely sensed observations provided by the operational SPOT 4-5/VEGETATION (VGT) Earth observing system. Neural networks classifications are performed to test alternatives for the classification of multi-temporal remote sensing data, such as normalized reflectance data and 10-day maximum value composite NDVI (normalized difference vegetation index). The new products shows an improvement of the accuracy compared to Global Land Cover 2000 project (GLC 2000) map over France.

Keywords

geophysical signal processing; image classification; image resolution; neural nets; terrain mapping; vegetation mapping; AD 2002; AD 2003; Earth observing system; France; GLC 2000 map; Global Land Cover project; SPOT-VEGETATION data; image classification; image resolution; land cover mapping; maximum value composite NDVI; multitemporal remote sensing data; neural networks classification; normalized difference vegetation index; normalized reflectance data; Clouds; Discrete cosine transforms; Multi-layer neural network; Neural networks; Polynomials; Reflectivity; Remote sensing; Spatial resolution; System testing; Vegetation mapping;

fLanguage

English

Publisher

ieee

Conference_Titel

Geoscience and Remote Sensing Symposium, 2004. IGARSS '04. Proceedings. 2004 IEEE International

Print_ISBN

0-7803-8742-2

Type

conf

DOI

10.1109/IGARSS.2004.1369861

Filename

1369861