An algorithm for generalizing topography to grids while preserving subscale morphologic characteristics—creating a glacier bed DEM for Jakobshavn trough as low-resolution input for dynamic ice-sheet models

The objective of this paper is to derive an algorithm for preserving important subscale morphologic characteristics at grids of lower-resolution, in particular for linear features such as canyons and ridge lines. The development of such an algorithm is necessitated by applications that require reduced spatial resolution, as is common in cartographic generalization, GIS applications, and geophysical modeling. Since any algorithm that results in weighted averages, including optimum interpolation and ordinary kriging, cannot reproduce correct depths, a new algorithm is designed based on principles of mathematical morphology. The algorithm described here is applied to derive a subglacial bed of the Greenland Ice Sheet that includes the trough of Jakobshavn Isbræ as a continuous canyon at correct depth in a low-resolution (5-km) digital elevation model (DEM). Data from recent airborne radar measurements of the elevation of the subglacial bed as part of the CReSIS project are utilized. The morphologic algorithm is designed with geophysical ice-sheet modeling in mind, in the following context. Currently occurring changes in the Earth's climate and the cryosphere cause changes in sea level, and the societal relevance of these natural processes motivates estimation of maximal sea-level rise in the medium-term future. The fast-moving outlet glaciers are more sensitive to climatic change than other parts of the Greenland ice sheet. Jakobshavn Isbrae, the fastest-moving ice stream in Greenland, follows a subglacial geologic trough. Since the existence of the trough causes the acceleration of the slow-moving inland ice in the Jakobshavn region and the formation of the ice stream, correct representation of the trough in a DEM is essential to model changes in the dynamics of the ice sheet and resultant sea-level predictions, even if current ice-sheet models can typically be run only at 5-km resolution. The DEM resultant from this study helps to bridge the conceptual gap between data analysis and geophysical modeling approaches. It is available as SeaRISE Greenland bed data set dev1.2 at http://websrv.cs.umt.edu/isis/index.php/SeaRISE_Assessment.