Comparison of digital elevation models over Australia and external validation using ERS-1 satellite radar altimetry

Abstract

Digital elevation models (DEMs) are widely relied upon as representations of the Earth's topographic morphology. The most widely used global DEMs available are ETOPO5, TerrainBase and JGP95E at a 5-arc-minute spatial resolution, and the GTOPO30 and GLOBE (version 1) global DEMs at a 30-arcsecond spatial resolution. This paper presents the results of intercomparisons of these global DEMs over Australia, and with the GEODATA 9-arc-second DEM (version 1) of Australia. These DEMs were also compared to an independently produced, altimeter-derived orthometric height database. This allows not only a totally independent assessment of the quality of these different DEMs over Australia, but also an insight into the ERS-1 radar altimeter's ability to measure orthometric heights on land. The results of all these comparisons reveal large differences among the DEMs, with the greatest difference between JGP95E and ETOPO5 (mean 49 m, standard deviation 274 m). The comparison with the altimeter derived database shows good agreement with the version 1 GEODATA DEM (mean 2 m, standard deviation 27 m), thus demonstrating that the altimeter is a viable method for quality assessment of DEMs in lowland regions. A further conclusion is that the representation of the Australian land surface in both the JGP95E and TerrainBase global DEMs is more accurate than the higher resolution GLOBE (version 1) global DEM, even though JGP95E displays a disparity along the 140-degree E meridian because of the different data sources used in its construction.