|Titre||Landsat 7 SLC-Off gap-filling for interim data continuity in northern regions using a robust radiometric normalisation technique|
|Licence||Veuillez noter que la Licence du gouvernement
ouvert - Canada remplace toutes les licences antérieures.|
|Auteur||Orazietti, J; Olthof, I; Fraser, R|
|Source||Proceedings of the 26th Canadian Symposium on Remote Sensing; 2005, 8p., https://doi.org/10.4095/220696 Accès ouvert|
|Séries alt.||Secteur des sciences de la Terre, Contribution externe 2005116|
|Réunion||26th Canadian Symposium on Remote Sensing; Wolfville, Nova Scotia; CA; juin 14-16, 2005|
|Media||papier; en ligne; numérique|
|Sujets||Nature et environnement|
|Diffusé||2005 01 01|
|Résumé||(disponible en anglais seulement)|
Much of the current climate change research in Canada involves monitoring northern environments, which require a continuous source of data for change detection.
Unfortunately, data continuity of Landsat 7 imagery has been interrupted due to the recent failure of the Scan Line Corrector (SLC), which compensates for the forward motion of the satellite. Although Landsat 7 data are still being acquired, each
scene has gaps of no-data values due to the SLC failure. We have created a methodology to produce a consistent source of data for medium-resolution northern biomass study and change detection until a suitable replacement for Landsat 7 is launched.
The relatively short growing season and low sun angle in the north provide few opportunities for acquisition of usable earth observation data, and coupled with the failure of the SLC, further reduces the amount of usable data per scene. Fortunately,
the large overlap between adjacent orbits in northern Landsat scenes provides a potentially large sample to use for gap infilling. We propose normalization of scenes using overlap regions, semi-invariant targets and a robust regression technique
called Thiel-Sen prior to infilling gaps in SLC-Off imagery using the normalised adjacent scene. The new technique enhances the Level 1G (L1G) products provided by the USGS and is modification to gap-filling procedures in the data flow of the
Gap-Fill Algorithm currently in use by the USGS. The resulting imagery is much more suitable for biomass modelling using vegetation indices such as the Reduced Simple Ratio (RSR) and Normalised Difference Vegetation Index (NDVI). The current
methodology shows great potential for maintaining data continuity of Landsat 7 imagery for northern environmental monitoring procedures requiring the use of vegetation indices.