||Modeling Topographic Effects in Satellite Imagery
||A. Mazimwe, A. Gidudu
||Azimuth Angle, Zenith Angle, Slope, Aspect, Insolation Intensity, Classification Accuracy
||The ability to extract accurate land cover information from satellite imagery depends on the quality of satellite data. Thus a study was made to assess the impact of topographic correction on satellite imagery where illumination effects were modelled with a DEM to quantitatively assess the classification accuracy. During the study, false colour composites were created using raw bands, and radiance band images which were a result of applying correct scene date, and varied sun’s Zenith and Azimuth angles in the intensity formula. Supervised and unsupervised classification algorithms were used to classify the composites followed by quantitative accuracy assessment to compare classification accuracy. Following the application of topographic correction, results at a 95% confidence level showed that 1) supervised classification was significantly better than unsupervised classification, 2) the true radiance composite classification had a significantly better overall accuracy than radiance classifications where azimuth was varied or the Zenith was increased, and 3) accuracy seems to increase significantly when the Zenith angle is lowered from true Zenith while maintaining the true Azimuth. This paper demonstrates that correcting slope-aspect effects significantly improves classification accuracy of satellite imagery necessitating corrections for geometric errors in remotely sensed data.
||The first Conference on Advances in Geomatics Research (2011)