Investigation of topographic influences on compact airborne spectrographic imager data
The 'topographic effect' is the distortion in measured radiance caused by differential illumination and viewing angles in undulating terrain. This effect is quantified for CASI imagery of the Reunion Island (French overseas territory, Indian Ocean), in an area dominated by sugar cane cover. High-resolution digital elevation models provide a means to correct the data in a remotely sensed image for the topographic effect. By calculating the slope and aspect of a surface, the sun/surface/sensor orientation can be determined, and its effects included in a standard data processing procedure. Several algorithms of topographic correction using DEMs have been proposed, but none have gained widespread use in the remote sensing community. The goals of our research are to test two methods of topographic correction on a sugar cane stand in Reunion Island, and demonstrate if these methods should be considered before creating any vegetation map using remotely sensed data. A digital elevation model is generated by stereoscopic analysis of aerial photography. The slope and aspect images of the study area are produced. There are the main input for the normalisation models. The Lambertian and non-Lambertian normalisation model were found to apply over correction. Using the regression coefficients obtained for the non-Lambertian model, the Lambertian assumption can be rejected. The results shown that it is not necessary to include a topographic correction step in the vegetation mapping process of the CASI images. This result could not be generalised to whole Reunion Island but only on area with the same topographic characteristics.
Main Author: | |
---|---|
Format: | thesis biblioteca |
Language: | eng |
Published: |
Cranfield University
|
Subjects: | U30 - Méthodes de recherche, B10 - Géographie, |
Online Access: | http://agritrop.cirad.fr/519299/ http://agritrop.cirad.fr/519299/1/BAPPEL-2001-%20investigation%20topografic.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|