Correction of Terrain effects on satellite image radiance

Abstract

With the ever-increasing resolution of satellite sensors, the use of satellite images for precise mapping will become more common. IKONOS is one of the new generation of high resolution satellite images, providing spatial resolution of 1m in Panchromatic, and 4m in Multispectral modes. However, reliable image interpretation results, including positional and classification accuracy, depend on the effectiveness of pre-classification techniques such as geometric and topographic corrections. The problem of differential terrain illumination on satellite imagery has been investigated for over 20 years but there is still no satisfactory solution. Most past research has been conducted on Landsat imagery where the look angle is nadir and the spatial resolution is 30m. The high spatial resolution, and off-nadir look angle of IKONOS imagery requires a more accurate digital elevation model (DEM) for topographic correction. Moreover, there is no research on the impact of the quality of DEM on topographic correction. These issues are particularly relevant for the use of high resolution satellite image in a mountainous region such as Hong Kong. The objectives of this project are to 1) investigate the impact of DEM quality on topographic correction of IKONOS imagery, and 2) compare some existing methods of topographic correction and see how successfully they can be applied to IKONOS imagery of Hong Kong. A variety of DEMs were created using different interpolation methods. An IKONOS-2 image was then ortho-rectified achieving an accuracy of approximately 4m over the whole image. Slope and aspect parameters derived from the different terrain models were applied in different topographic correction algorithms. To evaluate the impact of the DEM on topographic correction, training area statistics and image classification were used to evaluate the accuracy and consistency of the classes for each DEM. This research contributes to existing knowledge of terrain effects on satellite image radiance. It demonstrates that existing algorithms for topographic correction of satellite imagery can be successfully adapted to IKONOS images and illustrates the importance of the quality of the DEM to topographic correction of IKONOS imagery.