One of the most fundamental geophysical measurements of the Earth is that which describes the shape of its land surface. Topographic data are required by virtually all scientific disciplines engaged in studies at or near the land surface. Topography is also civilization's most heavily used non-atmospheric geophysical measurement. NASA's ASTER and SRTM projects have each recently completed independent near-global digital elevation measurements (or measurement-amenable data acquisitions) at comparable resolutions that approach 30 meters spatially and 10 meters vertically. But problematic data gaps are widespread in SRTM data, and ASTER DEMs cover the world�s landmass in 45,000 independent panels. Fusion of these data would be a major step toward completion of a high-resolution and comprehensive global elevation model, a valuable asset for Earth sciences as well as civil applications.
We propose to study the technical problems, develop the methods, and demonstrate the potential of generating an integrated and improved global elevation model. We will undertake this effort on several fronts and from a global perspective. Firstly, we will characterize in detail the systematic and physiographic causes of errors and voids in the DEMs in order to define requirements for accurately correcting and completing the DEM in those locations. Problems of these data sets can be difficult to identify and evaluate when used individually, but are usually obvious when the data sets are contrasted. Secondly, we will develop optimal methods for merging the DEMs, particularly regarding blending for a seamless combined product. Thirdly, we will develop new and innovative methods of filling gaps in geographic locations that are particularly problematic for both data sets, such as very steep terrain. We note that major errors in ASTER DEMs can occur where accurate topography is clearly perceived in the ASTER imagery. The data are adequate to define the topography, but new techniques are needed to more accurately extract it. Topographic information is available in all ASTER bands (visible, reflectance infrared, and thermal infrared), not just in the stereo pair, and integration of these inputs can potentially fill the gaps and improve the entire DEM.
Understanding the characteristics and differences of the DEMs is important not only for their fusion into a combined product, but also for their differentiation where surface change measurements can be used to assess natural hazards. Consequently, our secondary objective will be to demonstrate hazard assessments that may be quantified by temporal differences in the DEMs. Differentiation of DEMs is becoming an important measure in assessing glacial mass changes, and more accurate DEMs would have direct application to quantifying that climate change indicator. Major geologic hazards, such as landslides and some volcanic events, can also be quantified by such methods. From differing temporal perspectives, topographic measurements present both a single picture of what has happened to Earth�s surface in the past and a picture sequence of what is happening now. Fusion and differentiation of the ASTER and SRTM DEMs will contribute to both of those perspectives.
Our primary products will be the algorithms required for completion of a high-resolution global elevation model plus spatially comprehensive demonstration DEMs of scientifically significant and globally representative sites. At present there is no mechanism for creating a continuous, high-resolution global DEM, yet the data exist. Our success will enable follow-on production of a near-global high-accuracy DEM derived by integrating the ASTER and SRTM data sets.
