Application of Inverted Slope Images for Geological Mapping – Reduction of Artifacts in Digital Elevation Models by Filtering in the Frequency Domain

Abstract

Inverted slope images created from digital elevation models (DEMs) can be used to visualize differences in lithological units and provide an indication of areas of geological outcropping. Such images are useful for geological mapping, but when created from DEMs generated from remote sensing images (“image-derived DEMs”), artifacts in the DEMs are exaggerated and give rise to unclear noisy final images. Such artifacts also affect the quality of normal slope images and shaded relief images. In contrast, DEMs created from elevation data in the form of contour maps (“contour-derived DEMs”) contain far fewer artifacts, but their availability is limited. Therefore, in order to obtain clear processed DEM images that can be used for global geological mapping, it is necessary to develop a method for reducing the number of artifacts in image-derived DEMs. In the present study, the effectiveness of a fast Fourier transform (FFT) for artifact reduction in image-derived DEMs was investigated. The FFT converts spatial data into the frequency domain, and artifacts with specific frequencies are removed by low-pass filtering. An inverse FFT is then used to return to the spatial domain, thus producing a filtered DEM with a reduced number of artifacts. The effect of the filter size on the final image quality was also investigated. The image-derived DEMs used in this study were the Shuttle Radar Topography Mission (SRTM) DEM and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global DEM (GDEM) for Cuprite, Nevada. A comparison was made between shaded relief and inverted slope images created following artifact removal from these DEMs and equivalent images produced using the contour-derived National Elevation Dataset (NED) DEMs. It was clarified that for the image-derived DEMs, that a low-pass filter with a size of 20 to 40 % significantly reduced the number of artifacts without significant loss of terrain information. For the ASTER GDEM data, artifact reduction was found to be particularly important because the artifacts obscure the original topographic information in the inverted slope images.