Improving the Accuracy of ASTER EO data

In order to improve our understanding of processes such as sea level rise and climate change, there is a requirement for enhanced assessment of historical and present-day glacial change in sensitive polar regions such as the Antarctic Peninsula. Vulnerable coastal societies such as the UK require better, more reliable estimates of sea level rise, and such information is essential in shaping government policy and mitigating impacts upon existing infrastructure. However, accurate assessment of glacial change in the remote and hazardous Antarctic Peninsula is not straightforward. A range of existing approaches for measuring elevation change exist. These include aerial photography, satellite laser altimetry (e.g. ICESat), and radar-based approaches (e.g. ERS-1/2). However, for a variety of reasons, nearly all suffer from drawbacks.Satellite-based stereo imaging offers perhaps the strongest solution, allowing the production of a digital elevation model (DEM) â a continuous surface which models elevation changes in a three-dimensional manner. Currently, DEMs derived from NASAâs Advanced Spaceborne Thermal Emission and Reflection radiometer (ASTER) sensor* offer greatest potential, with good coverage of polar regions and a spatial resolution of 30 metres. However, the potential of ASTER is limited by its relatively weak accuracy, with elevation measurements correct to only around 10-20 metres, and deteriorating further over steep terrain. Although this can be improved through the use of ground control points, which are acquired through field survey, such an approach is not possible in inaccessible and hostile polar environments.This research investigates an automated mathematical approach which, instead, performs statistical minimisation of a poorly-controlled DEM surface with respect to aDEM of higher accuracy. This approach, known as âsurface matchingâ offers significant potential for improving the accuracy of ASTER data and facilitating the wider application of the technique in order to quantify glacial change in a systematic manner across the Antarctic Peninsula.Through collaboration with the British Antarctic Survey (BAS) the RICS research was applied to a test site located on the western Antarctic Peninsula at Pourquoi Pas Island.This encompassed a glacier system, extending over an area of 86 km2, and providing a variety of terrain types over which to investigate the approach. In addition to an ASTER DEM, a high-accuracy DEM, derived from aerial photography captured at the same time as the ASTER data, was available for the site.The goal of the research was to unlock the potential of ASTER data for delivering enhanced quantification of glacial change in the Antarctic Peninsula, a region which is thought to be particularly sensitive to climate change. In order to achieve this, the research investigated the presence of elevation biases in the ASTER DEM, and introduced a capacity within the matching algorithm for evaluating and correcting for any such errors. In the dataset under consideration here, no such bias was found to exist. Following this, the prime focus of the project was in establishing an optimum matching strategy for registering an ASTER DEM to a well-controlled photogrammetric DEM. The surfaces were acquired only three weeks apart in time, therefore minimising any terrain-related changes, and facilitating rigorous validation of matching performance. Local, patch-based and global matching strategies were explored over the 86 km2 extent of the glacier system.The patch-based approach delivered significantly better performance for the selected areas of interest than the two alternative approaches, providing strong evidence to support the broader application of this strategy at other glacier sites across the Antarctic Peninsula. Following validation of these results through comparison to GPS check points, ASTER DEM quality was explored with respect to terrain type. This revealed strongest agreement between the ASTER and reference DEMs over relatively flat glacier surfaces, with significant errors occurring over steep terrain and in regions of shadow. These findings will be particularly useful in quantifying uncertainties associated with glacial change, through the extension of the research to other similar sites. The outcomes of this research are expected to lead to enhanced estimates of glacier mass balance change across the Antarctic Peninsula, which in turn, will reduce uncertainties associated with the contribution of this component to sea level. Ultimately, this will inform UK government policy on climate change impacts and the mitigation of the effects of rising sea levels, shaping RICS best practice guidelines in relation to sustainable engineering. * The ASTER imaging instrument is a payload on the Terra satellite, launched in December 1999 as part of NASA\'s Earth Observing System (EOS). ASTER is a cooperative effort between NASA, Japan\'s Ministry of Economy, Trade and Industry (METI) and Japan\'s Earth Remote Sensing Data Analysis Center (ERSDAC). The full report of this research is available for download from www.rics.org/site/download_feed.aspx?fileID=11596&fileExtension=PDF

For more information visit:

www.rics.org