Enhancing the SRTM Da ta for Australia

The SRTM dataset is a near-global DEM derived from interferometric radar data c ollected by NASA's Space Shuttle in February 2000. It is unique in providing reasonably good quality topographic data over most land masses but suffers from a number of artefacts including: voids in high relief areas and some other locations; striping artefacts; and offsets due to woody vegetation. It was produced at 1 second (about 30 m resolution) but degraded to 3 second resolution for public release except for the USA. In Australia, the SRTM data provides better quality topographic information than other available sources over much of the continent. Its utility is hampered by the artefacts, particularly the offsets induced by vegetation. Many rivers in inland Australia are surrounded by remnant native vegetation while the surrounding land is cleared for agriculture; the rivers thus appear as raised features in the SRTM data preventing any useful hydrological analysis. This paper describes methods developed to treat the artefacts in the SRTM data to produce a usable DLSM. Negotiations with Australia's Defence Imagery and Geospatial Organisation (DIGO) have resulted in access to the 1 second version of the SRTM data and permission to release processed versions of that data at 1 second resolution to government agencies within Australia. Reduced resolution 3 second versions will be publicly available. The methods described could be applied to SRTM data in other areas that lack better quality DEM data, such as Africa, so long as vegetation mapping is also available to support the removal of vegetation offsets. The methods could also be adapted to treat similar offsets in other radar or photogrammetric products that do not directly map land surface heights in vegetated areas. 2. Removal of St riping Artefacts The first step was to remove striping artefacts, which are clearly visible in some areas of low relief Australia. The stripes have a typical wavelength of about 800 m with an amplitude of about 0.2 - 4 m and are aligned diagonally in a pattern that suggests a close relationship to the orbital paths. A software tool was developed to display tiles of the raw SRTM data (º × º degree) and a 2-D Fourier transform of the data and support identification and removal of frequency components corresponding to the stripes (Figure 1). The tool was used by a team of analysts to remove striping wherever stripes could be discerned across the continent.