The near-global elevation product from NASA’s Shuttle Radar Topographic Mission (SRTM) has been widely used since its release in 2005 at 3 arcsecond resolution and the release of the 1 arcsecond version in late 2014 means that the full potential of the SRTM DEM can now be realised. However the routine use of SRTM for analytical purposes such as catchment hydrology, flood inundation, habitat mapping and soil mapping is still seriously impeded by the presence of artefacts in the data, primarily the offsets due to tree cover and the random noise. This paper describes the algorithms being developed to remove those offsets, based on the methods developed to produce the Australian national elevation model from SRTM data. <br><br> The offsets due to trees are estimated using the GlobeLand30 (National Geomatics Center of China) and Global Forest Change (University of Maryland) products derived from Landsat, along with the ALOS PALSAR radar image data (JAXA) and the global forest canopy height map (NASA). The offsets are estimated using several processes and combined to produce a single continuous tree offset layer that is subtracted from the SRTM data. <br><br> The DEM products will be made freely available on completion of the first draft product, and the assessment of that product is expected to drive further improvements to the methods.