Header

UZH-Logo

Maintenance Infos

Multitemporal Unmixing of MERIS FR Data


Zurita-Milla, Raúl; Gómez-Chova, Luis; Clevers, J G P W; Schaepman, Michael E; Camps-Valls, Gustavo (2007). Multitemporal Unmixing of MERIS FR Data. In: 10th Intl. Symposium on Physical Measurements and Spectral Signatures in Remote Sensing (ISPMSRS), Davos, Switzerland, 12 March 2007 - 14 March 2007, 238-243.

Abstract

The possibilities of using MERIS full resolution (FR) data to extract sub-pixel land cover composition over The Netherlands are explored in this paper. More precisely, the use of MERIS FR time series is explored in this paper since it should facilitate the discrimination of spectrally similar land cover types because of their seasonal variations. The main steps of the methodology used to extract sub-pixel information can be summarized as follows. First, a set of seven MERIS FR Level 1b images that covered the period February to December 2003 were selected. Second, the images were projected into the Dutch national coordinate system. Special attention was paid to this process in order to account for the orbital differences of each MERIS acquisition. Third, a cloud screening algorithm was applied to all MERIS images. Next, the MERIS level 1b TOA radiances were converted into surface reflectance. After that, the latest version of the Dutch land use database (LGN5) was used to support the selection of the endmembers from the MERIS images. Finally, a constrained linear unmixing algorithm was applied to each of the MERIS scenes and to the multi-temporal dataset. The results were validated both at sub-pixel and per-pixel scales using the LGN5 as a reference. The paper concludes by describing the potential and limitations of the selected approach to extract sub-pixel land cover information over heterogeneous and frequently clouded areas.

Abstract

The possibilities of using MERIS full resolution (FR) data to extract sub-pixel land cover composition over The Netherlands are explored in this paper. More precisely, the use of MERIS FR time series is explored in this paper since it should facilitate the discrimination of spectrally similar land cover types because of their seasonal variations. The main steps of the methodology used to extract sub-pixel information can be summarized as follows. First, a set of seven MERIS FR Level 1b images that covered the period February to December 2003 were selected. Second, the images were projected into the Dutch national coordinate system. Special attention was paid to this process in order to account for the orbital differences of each MERIS acquisition. Third, a cloud screening algorithm was applied to all MERIS images. Next, the MERIS level 1b TOA radiances were converted into surface reflectance. After that, the latest version of the Dutch land use database (LGN5) was used to support the selection of the endmembers from the MERIS images. Finally, a constrained linear unmixing algorithm was applied to each of the MERIS scenes and to the multi-temporal dataset. The results were validated both at sub-pixel and per-pixel scales using the LGN5 as a reference. The paper concludes by describing the potential and limitations of the selected approach to extract sub-pixel land cover information over heterogeneous and frequently clouded areas.

Statistics

Citations

Downloads

45 downloads since deposited on 18 Apr 2013
18 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Conference or Workshop Item (Paper), not refereed, original work
Communities & Collections:07 Faculty of Science > Institute of Geography
Dewey Decimal Classification:910 Geography & travel
Language:English
Event End Date:14 March 2007
Deposited On:18 Apr 2013 11:32
Last Modified:07 Dec 2017 21:02
Publisher:ISPRS
Number:XXXVI-7C50
Official URL:http://www.isprs.org/proceedings/XXXVI/7-C50/papers/P6.pdf
Related URLs:http://www.isprs.org/proceedings/XXXVI/7-C50/ (Organisation)

Download

Download PDF  'Multitemporal Unmixing of MERIS FR Data'.
Preview
Content: Published Version
Language: English
Filetype: PDF
Size: 1MB