Generation and refinement of a continental-scale digital elevation model by integrating cartographic and remotely sensed data: A GIS-based approach.

摘要

Digital Elevation Models (DEMs) are of fundamental importance to a wide range of geo-scientific and environmental studies. This research addresses technical and theoretical issues involved in the construction of a large-scale DEM with multiple data sources. Enormous data volume, diverse data sources and different terrain types complicate the large-scale DEM generation process. To tackle technical difficulties and challenges, several new methods have been developed and some existing algorithms have been improved. Having these newly developed and extended methods and algorithms along with available GIS functions, we constructed a complete, consistent, and high-resolution DEM over the Antarctic by integrating a variety of cartographic and remotely sensed data.;This research demonstrates that the GIS technology represents an effective means of integrating not only various types of topographical data, but also a pool of alternative computational methods and techniques for the creation of large-scale DEMs. Nevertheless, currently available GIS software did not provide an entire solution to the development and refinement of large-scale DEMs.;In this research, a number of error detection techniques have been developed for efficiently checking and eliminating gross errors in source data. New interpolation procedures were designed for gridding traverse data and contour-based topographical data. The determination of an optimal interpolation spacing was pursued. A conceptual framework was proposed for integrating and fusing complementary and competitive topographical data. In the case of the Antarctic, a set of criteria were developed for the selection of competitive data to exploit the best elements of available sources. It is also demonstrated that spatially and functionally complementary topographical data sources can be synergistically combined during the interpolation stage to create a more reliable and accurate elevation model than what a single source could create alone. A new algorithm was developed for merging individual DEM data sets into a seamless DEM mosaic along irregular lines. For DEM refinement, the potential of the state-of-the-art remote sensing techniques, including shape-from-shading, SAR stereo and interferometric SAR (InSAR), are discussed and demonstrated. A technical strategy is formulated for future improvement of the Antarctic DEM.;The accuracy and error pattern of the resultant Antarctic DEM were evaluated in a number of ways. Comparisons with high-resolution satellite images at different scales demonstrate that the DEM not only describes the broad topographical configuration and true ice drainage pattern, but also captures unprecedented detail of the terrain surface. We believe that this DEM product represents the most detailed and accurate digital description of the Antarctic topography published to date.