Fast Stream Extraction from Large, Radar-Based Elevation Models with Variable Level of Detail

M. Metz1, H. Mitasova2, R.S. Harmon3
1 Research Group Cognitive Ethology , German Primate Center
Kellnerweg 4, 37077 Goettingen, Germany
Telephone: +49-(0)921-2614299
2 Department of Marine, Earth and Atmospheric Sciences, North Carolina State University
Campus Box 8208, Raleigh, North Carolina 27695, USA
Telephone:001- 919-513-1327
3 Environmental Sciences Division, Army Research Office, U.S. Army Research Laboratory,
4300 South Miami Boulevard, Durham, North Carolina, 27703-9142, USA

Interferometric Synthetic Aperture Radar for Elevation (IFSARE) and Shuttle Radar Topographic Mission (SRTM) surveys provide a new generation of digital surface models (DSM) in regions that have had only limited elevation data coverage. These new topographic data sets are increasingly used to improve mapping of geomorphic and hydrologic features in remote, hard to reach areas and at regional to global scales (e.g. Kinner et al. 2005, Lehner and Döll 2004, World Wildlife Fund 2009). Extraction of hydrologic features from radar-based elevation data poses several challenges: (a) elevation surfaces include tree canopy that often requires depression filling of large areas (Figures 1, 2); (b) depending on the size of the study region and resolution, data sets can be massive and require extensive processing time. Significant effort has been devoted to development of new flow tracing and watershed analysis algorithms that support efficient processing of large DSMs and address the issue of depression filling (e.g., Arge et al. 2003; Danner et al. 2007).

We present a new implementation of method for flow routing, flow accumulation, and watershed analysis based on a least-cost path search algorithm (A* Search, Hart et al. 1968; module r.watershed in GRASS GIS, Ehlschlaeger 1989). This implementation dramatically improves computational efficiency while preserving its high accuracy routing capabilities through nested depressions, even for a challenging triple-canopy tropical rainforest environment with tree heights of more than 30m above the land surface. The new implementation that includes both single (SFD) and multiple flow direction (MFD) routing is compared with previously developed methods in terms of performance and accuracy. The impact of mapping technology (IFSARE, SRTM) and resolution on the extracted stream networks is also analyzed.

AttachmentSizeHitsLast download
metz2009geomorphometry.pdf389.82 KB20702 days 5 hours ago
Your rating: None Average: 4.5 (2 votes)